Analytical and Numerical Analysis of Low Optical Overlap Mode Evanescent Wave
Chemical Sensors
A thesis presented to
the faculty of
the Russ College of Engineering and Technology of Ohio University
In partial fulfillment
of the requirements for the degree
Master of Science
Anupama Solam
August 2009
© 2009 Anupama Solam. All Rights Reserved. 2
This thesis titled
Analytical and Numerical Analysis of Low Optical Overlap Mode Evanescent Wave
Chemical Sensors
by
ANUPAMA SOLAM
has been approved for
the School of Electrical Engineering and Computer Science
and the Russ College of Engineering and Technology by
Ralph D. Whaley, Jr.
Assistant Professor of Electrical Engineering and Computer Science
Dennis Irwin
Dean, Russ College of Engineering and Technology 3
ABSTRACT
SOLAM, ANUPAMA, M.S., August 2009, Electrical Engineering
Analytical and Numerical Analysis of Low Optical Overlap Mode Evanescent wave
Chemical Sensors (106 pp.)
Director ofThesis: Ralph D. Whaley, Jr
Demands for integrated optical (IO) sensors have tremendously increased over the years due to issues concerning environmental pollution and other biohazards. Thus, an integrated optical sensor with good detection scheme, sensitivity and low cost is needed.
This thesis proposes a novel evanescent wave chemical sensing (EWCS) technique for ammonia (gaseous) and nitrite (aqueous) detection utilizing a low optical overlap mode
(LOOM) waveguide structure. This design has the advantage of low modal fill factor and more field interaction with the sensing region compared to fiber sensors; hence eliminating the difficulty faced by traditional EWCS designs. Effective refractive index is a crucial parameter for analyzing the sensitivity, which is compared using the analytical and numerical methods such as finite element method (FEM) and semi-vectorial beam propagation method (BPM). The sensitivity of LOOM structures are calculated and compared analytically and numerically. Finally, these waveguide structures are analysed when integrated with a Mach-Zehnder Interferometer.
Approved: ______
Ralph D. Whaley, Jr.
Assistant Professor of Electrical Engineering and Computer Science
4
ACKNOWLEDGMENTS
It’s a wonderful opportunity to thank my advisor, Dr. Ralph Whaley, Jr. PhD. for his consistent support and guidance in every step for the completion of my thesis. I highly respect and appreciate his ideas and inputs, and learned a lot of things besides academics such as being systematic, polite, humble and enthusiastic. Secondly, I would like to thank my committee members Dr.Savas Kaya, Dr.Avinash Kodi and Dr.Karen Coschigano for their extended support. I am grateful to Dr.Gines Lifante, who had provided me literature help and replies to various questions through email and also helping me with the results for Semi-vectorial BPM simulations. I also thank the Rsoft research group, mainly Mr.
Matthew Frank, for troubleshooting various problems while working with the FEM software. My father Mr. Sharath Chander, my mother Mrs. Jhansi Rani and my sister
Anusha have always been encouraging and showed tremendous interest towards my study and research, which contributed to my thesis to a great extent. I thank my friend
Bhargav Kota, who has been motivating and always there for any kind of help and support. I would also like to thank my friends from India, Priyanka Kamath, Deepthi
Chandu, Lakshmi Iyer and Kriti Reddy who stood by me always. I thank my seniors at
Ohio University: Chandana Venkatayogi, Jyothsna Jakka, Vijaya Hari, Tanu Sharma,
Sulalita Chaki and Mohor Chatterjee for their encouragement and understanding. Krishna
Manoharan, one of our research members, helped a lot with his inputs and suggestions throughout the process. I thank him and Aarthi Srinivasan, for her warmth and help. I also thank my relatives Mr. P.Sri Kumar and Mrs. Shailaja for following me up all the while. Last, but not the least, I would like to thank our janitor, Mrs. Rose Mary.
5
TABLE OF CONTENTS
Page
ABSTRACT ...... 3
ACKNOWLEDGMENTS ...... 4
LIST OF TABLES ...... 9
LIST OF FIGURES ...... 10
CHAPTER 1: INTRODUCTION ...... 12
1.1 Overview of Optical Sensors ...... 14
1.2 Bio-Chemical Sensor ...... 16
1.3 Advantages of Planar Structures over Fibers for Optical Bio-sensing ...... 17
1.4 Importance and Phenomenon of Evanescent Wave Sensing ...... 18
1.5 Types of Evanescent Wave Sensors ...... 21
1.6 Working Principle of Integrated Optic (IO) Sensors ...... 23
1.7 Single Mode Operation ...... 25
1.8 LOOM Structure ...... 26
1.9 Analytes ...... 28
1.9.1 Gaseous Detection of Ammonia ...... 28
1.9.2 Aqueous Detection of Nitrites...... 28
1.10 Material Used ...... 29
1.10.1 Indium Phosphide (InP) ...... 29
1.10.2 Amorphous Zinc Oxide (a-ZnO) ...... 30 6
CHAPTER 2: ANALYTICAL AND NUMERICAL METHODS TO OBTAIN
EFFECTIVE INDEX ...... 31
2.1 Characteristics of Optical Waveguides ...... 31
2.2 Channel Waveguides ...... 33
2.3 Analytical Solution: Effective Index Method (EIM) ...... 36
2.3.1 Theoretical Calculation of Effective Index for a Symmetric Planar
Waveguide Structure ...... 40
2.3.2 Wave Equations of a planar waveguide ...... 42
2.3.3 Other Important Waveguide Parameters ...... 43
2.3.4 Mode Parameters ...... 44
2.3.5 Eigen Value Equation ...... 45
2.4 Effective Index Calculations using Finite Element Method (FEM) ...... 49
2.5 Effective Index Calculations Using Semi-Vectorial Beam Propagation
Method (BPM) ...... 53
2.6 Vertical Cut Profile: Electric Field ...... 53
2.7 Confinement Factor ...... 55
2.8 Modal Fill Factor ...... 56
2.9 Gaseous and Aqueous Detection ...... 57
2.10 Modal Analysis of the Buried Structure ...... 58
CHAPTER 3: SENSITIVITY ANALYSIS ...... 60
3.1 Analytical Method ...... 60
3.2 Numerical Method: Sensitivity Analysis using FEM Approach ...... 62 7
3.3 Sensitivity Analysis of Buried Waveguide Structure ...... 64
3.4 Surface Sensitivity of an Air LOOM Structure ...... 66
CHAPTER 4: MACH ZEHNDER INTERFEROMETER ANALYSIS ...... 67
4.1 Principle of Operation of an MZI ...... 68
4.2 Analysis of MZI ...... 69
CHAPTER 5: RESULTS AND DISCUSSION ...... 70
5.1 Analytical and Numerical Results for Effective Index Calculations ...... 70
5.2 Comparison of the Modal Fill Factor and Confinement Factor ...... 72
5.3 Sensitivity Analysis ...... 73
5.4 Mach-Zehnder Interferometer Output ...... 75
CHAPTER 6: CONCLUSION ...... 77
REFERENCES ...... 79
APPENDIX A.0 EIM CALCULATIONS FOR STRUCTURE 1(TE) ...... 86
APPENDIX A.1 EIM CALCULATIONS FOR FINAL STRCUTURE 1(TE) ...... 87
APPENDIX A.2 EIM CALCULATIONS FOR STRUCTURE 1 (TM) ...... 88
APPENDIX B.0 FEM SETTINGS ...... 89
APPENDIX C.0 CALCULATION OF ELECTRIC FIELD PLOTS ...... 90
APPENDIX D.0 SENSITIVITY CALCULATIONS FOR STRUCTURE 1(TE) ...... 91
APPENDIX D.1 SENSITIVITY CALCULATIONS FOR STRUCTURE 1(TM) ...... 92
APPENDIX D.2 MOST SETTINGS ...... 93
APPENDIX E.0 MZI OUTPUT FOR STRUCTURE 1 ...... 94
APPENDIX E.1 MZI OUTPUT FOR STRUCTURE 2 ...... 98 8
APPENDIX F.0 MFF AND CF CALCULATIONS FOR STRUCTURE 1 ...... 102
APPENDIX F.1 MFF AND CF CALCULATIONS FOR STRUCTURE 2 ...... 104
APPENDIX G.0 ABSORPTION SPECTRA OF GASEOUS AMMONIA ...... 106
9
LIST OF TABLES
Page
Table 1: Optical and geometrical properties of the LOOM based waveguide structures.47
Table 2: List of effective index values obtained...... 48
Table 3: Rf values for different cladding thicknesses...... 65
Table 4: Comparison of Neff between analytical and numerical methods...... 70
Table 5: Percentage of error calculated for the three methods...... 71
Table 6: Comparison of the MFF and CF...... 72
10
LIST OF FIGURES
Page
Figure 1: Layout of the Evanescent wave chemical sensors used...... 14
Figure 2: Schematic Figure of IO EWS...... 16
Figure 3: Operating principle of a biochemical sensor ...... 17
Figure 4: Classification of biochemical sensors ...... 17
Figure 5: Classification of optical sensors ...... 18
Figure 6: Single mode waveguide with evanescent wave penetration ...... 19
Figure 7: Penetration depth of an evanescent wave ...... 20
Figure 8: Three cases when the incidence angle of the light ...... 21
Figure 9: Shows types of EWS ...... 23
Figure 10: An integrated optic sensor with an adlayer ...... 24
Figure 11: Basic integrated sensor effects ...... 25
Figure 12: Dispersion curve ...... 26
Figure 13: The structure is a nanoscale membrane waveguide...... 27
Figure 14: The SEM shows an InP-based LOOM guide ...... 27
Figure 15: A CCD image of a guided LOOM mode...... 28
Figure 16: Fundamental structure of a dielectric waveguide ...... 32
Figure 17: Index profile distributions of step index and graded index waveguides ...... 33
Figure 18: Shows the 3 kinds of channel waveguides ...... 35
Figure 19: Rib waveguide and diffused waveguides structures ...... 36
Figure 20: Effective index approximation method...... 37 11
Figure 21: Flowchart of EIM ...... 38
Figure 22: Two LOOM based structures...... 40
Figure 23: Matlab output ...... 46
Figure 24: Figure showing mesh plot of structure 1...... 50
Figure 25: Figure showing the mode profile of structure 1 ...... 51
Figure 26: Vertical Electric field mode profile of structure 1...... 51
Figure 27: Horizontal electric filed mode profile of structure 1 ...... 52
Figure 28: Figure showing the index profiles for structure 1...... 52
Figure 29: showing the Hx and Ey field amplitude of an TM mode...... 55
Figure 30: shows the active region of a rib waveguide structure...... 56
Figure 31: Waveguide structures for gaseous and aqueous detection ...... 58
Figure 32: Mode plot and the vertical cut profile of the buried waveguide structure ...... 58
Figure 33: Figure showing the steps to obtain effective index ...... 62
Figure 34: MOST outputs for (a) TE and (b) TM, for structure 1 ...... 63
Figure 35: Absorbance versus Concentration plot for buried waveguide structure ...... 65
Figure 36: Absorbance versus Concentration for Air LOOM waveguide structure ...... 66
Figure 37: Schematic diagram of an IO MZI...... 68
Figure 38: Comparisons of the analytical and numerical methods for structure 1 ...... 74
Figure 39: Comparison of analytical and numerical methods for structure 2 ...... 75
Figure 40: Plot of power outputs P1 and P2 for structure1 at 1550nm...... 76
Figure 41: Plot of power outputs P1 and P2 for structure2 at 550nm ...... 76
12
CHAPTER 1: INTRODUCTION
In the recent years, optical sensors have attracted considerable attention, especially in the application of biochemical species detection. They have excellent advantages such as good compactness and robustness, immunity to electromagnetic interference, high sensitivity, shorter response time, low cost, and high compatibility with fiber optic networks [0]. However, optical fiber-based systems do not seem promising with respect to fabrication, efficiency and miniaturization. However, planar waveguide-based platforms employing evanescent wave sensing techniques have shown tremendous improvement [2] and evanescent wave sensors have proven to be highly sensitive [3-5].
Over a decade, biochemical sensors have shown profitable improvement, in various sensing applications and in industry, with a revenue report from high-tech market research firms indicating an increase in compound annual growth rate of about 11.5%, with the next generation biochemical sensors revenue rising from $2.3 billion in 2002 to nearly 4 billion in 2007 [6-7]. The applications are broad-reaching, including the areas such as environmental, chemical, biomedical, industrial and other process control fields
[8]. Added to this, it has been suggested that “modern optical sensors provide high- quality applications for monitoring environmental processes, process control, food analysis, medical diagnostics, investigation of the biosphere, drug detection and even extra-terrestrial research” [6, 9]. Recent rise in terrorist activities has also demanded for the need to detect biochemical explosives [6]. Irrespective of the application, realizing a highly sensitive, miniaturized, reduced cost biochemical sensor is desirable. With the
growing interest in the integration of waveguide sensors with existing electronic and 13
photonic platforms, comes the demand for high-performance sensing architectures that
can be directly incorporated into the standard CMOS process sequence.
Goals of this Thesis
The main aim of this thesis is to introduce the concept of Low Optical Overlap Mode
(LOOM) structure into integrated optical sensors incorporating an evanescent wave
chemical sensing technique. LOOM structures incorporating amorphous zinc oxide (a-
ZnO) material are analyzed using Finite Element Method (FEM). The structure utilizing
InP as the core material is a non-resonant structure as the wavelength of operation does not match the absorbance peak of the analyte, while the resonant structure with a-ZnO as
the core material works at the wavelength where the analytes show peak absorbance. The
critical parameter in assessing the sensitivity of the sensor is effective refractive index,
which is analyzed and compared using both analytical and numerical methods, FEM and
semi-vectorial BPM as previously mentioned. Modal analysis of LOOM structures using
FEM software are used to solve for the field components using a non-uniform mesh for
both guided and leaky modes. This thesis has two areas of sensing as shown in Figure 1:
gaseous and aqueous. An air LOOM and buried waveguide structures are designed for
detecting ammonia (gaseous) in air and nitrites (aqueous) in water. Sensitivity analysis is
performed and compared using the above mentioned methods. The other parameters such
as confinement factors and modal fill factors are also compared. Choice of the materials
used is important, and this is discussed in detail in the future. These LOOM structures
when integrated with Mach-Zehnder Interferometer architectures are also realized. 14
Evanescent wave sensor using LOOM mechanism
Gaseous ammonia Nitrite detection in detection water
Rib waveguide Rib waveguide structure structure Buried waveguide structure (InP) (a‐ZnO) (a‐SiC:H) non‐resonant resonant
Figi ure 1: Layout of the Evanescent wave chemical sensors used.
1.1 Overview of Optical Sensors
The current trends used for the development of optical sensors include miniaturization, low cost, compatibility with mass production processes, disposability besides multiple functionalities on a single platform and multiple analyte detection [2]. Interest and progress in the integrated optics sensors over the last two decades have tremendously increased due to the shortcomings of the conventional Electromagnetic Field (EMF) sensors, utilizing active metallic probes, thus producing noise. Some of the broad 15 scientific applications of EMF sensors are as process control, microwave-integrated circuit testing, electromagnetic compatibility measurements and electric field monitoring in medical apparatuses. Comparatively, photonic EMF sensors have very good galvanic insulation, high sensitivity and a very wide bandwidth. [10]. Optical sensors are widely divided based on the two segments, sensing mechanism and sensing architecture. The various sensing mechanisms are fluorescence, surface plasmon resonance, raman scattering, absorption change, photon migration spectroscopy and effective index change in the guiding structures which are discussed in detail in [1]. Based on the second category, they are further classified in terms of architecture as interferometer, anti resonant reflecting optical waveguides (ARROW), hollow waveguides, surface plasmon resonance, bragg gratings, silicon slot waveguides, integrated optical micro cavities and low optical overlap mode (LOOM) structures, which are also explained in detail in [1,
11]. Figure 2 shows a schematic overview of an integrated optical evanescent wave sensor (IO EWS).
16
Figure 2: Schematic Figure of IO EWS.
1.2 Bio-Chemical Sensor
Optical Bio-chemical sensing is an extremely wide field incorporating various devices
such as contact-less monitors, fiber-optic based devices, planar waveguides, evanescent
wave interrogation and interferometers. Some of the diverse applications in current
analytical chemistry are environmental monitoring, chemical process analysis, food
storage and manufacturing, clinical monitoring, space and aeronautics [12]. The bio-
sensing applications in the life sciences, such as binding and dissociation kinetics of antibodies and receptor-ligand pairs, protein-DNA and DNA-DNA interactions, epitome
mapping, phage display libraries and whole cell and virus protein interactions [13]. As
quoted in [1] “A biosensor is a device which consists of a biologically or biophysically-
derived sensing element integrated with a physical transducer that transforms a
measurand into an output signal.” Figure 3 shows the operating principle of a
biochemical sensor. In optical biosensors, there are two kinds of sensing mechanisms,
homogeneous sensing and surface sensing. The former relates to a change in the
refractive index of the cover medium due to the presence of analytes, and the latter is due
to an increase in the thickness of the adlayer deposited [1]. The adlayer refers to the
sensing layer which is deposited on top of the core layer to specifically bind the analytes.
It is used mainly to improve the specificity of the sensor. The two sensing mechanisms
are discussed in detail in later chapters.
17
Figure 3: Operating principle of a biochemical sensor [1].
The structure of a biosensor is shown in the Figure 4, which is based on [14]. They are classified based on different parameters such as nature of the sensor, monitored parameters, type of a transducer, operational mode, number of analytes, etc.
Figure 4: Classification of biochemical sensors.
1.3 Advantages of Planar Structures over Fibers for Optical Bio-sensing
In spite of the attractive advantages of optical fiber (OF) bio-chemical sensors, such as sensing in the inaccessible areas, transmitting optical signals over great distances with 18 low power loss, high immunity to electromagnetic disturbances, flexibility in their geometry, resistance to corrosion , electrically less passive, and better compatibility with telemetry; [15-17] they have a few disadvantages when compared to planar devices in terms of fabrication and mass production. In planar optical devices, deposition and patterning of reagents is easier, also different materials can be used. They have high precision as multiple devices can be fabricated on a single sheet. The devices are robust and hence have better flexibility. Planar devices are easily incorporated into optically associated instrumentation. Lastly, manufacturing equipment for optical planar devices is easily available [15]. Figure 5 shows the classification of optical sensors.
Figure 5: Classification of optical sensors.
1.4 Importance and Phenomenon of Evanescent Wave Sensing
In comparison to various other sensing methods, evanescent wave sensors (EWS) have various advantages, one of them being miniaturization, as the mode of operation is light, 19
and the sensor does not incorporate any bulky components. The interaction of the
evanescent field with the analyte forms the main concept of the sensing mechanism
which is explained in more detail in the forthcoming sections; and because the power in
the sensing region is high, these sensors have high sensitivity. They have immunity from
scattering effects and have better control in design besides reduction in cost [18].
Evanescent waves are those waves which penetrate into the cladding or the region where
light is reflected off the surface. In optical terminology, when total internal reflection
occurs at an angle greater than the critical angle, the sinusoidal waves reflect off from an
interface, and the waves which penetrate into, are called the evanescent waves [19].
Figure 6 shows the evanescent field decaying exponentially into the cover region.
Figure 6: Single mode waveguide with Evanescent wave Penetration.
The above mentioned mechanism is used in sensing, when the analytes or species to be detected are in contact or are made to interact with the evanescent field (which decay exponentially from the surface of the waveguide), forms the basis for evanescent wave 20 sensing. The operating wavelength is selected such that the analytes show peak absorption, so that there is a change at the output either in the intensity or power or any other parameter, when measured using a detector. In this thesis, the operating wavelength is chosen to be 1550nm and 550nm; while 1550nm is the non resonant wavelength which is purely simulated to compare and check with the previous literature; while for the
550nm; where many analytes show peak absorption as shown in Appendix G.0 is the resonant structure as mentioned earlier [20].
Figure 7 shows the penetration depth, when reflected off from a surface of higher index incident onto a lower index. The penetration depth (δ) is the exponential decay of the electric field, where N is the effective refractive index of the medium, and nc is the index of the cladding region, containing the analyte and is calculated using [21-22]
2 2 −1/ 2 (Equation 1.1) δ = ()λ / 2π [N − nC ]
Figure 7: Penetration depth of an Evanescent wave. 21
Figure 8 which is based on [22] shows the three cases when the incident light travelling in medium 1 with higher refractive index, strikes medium 2 having lower index. The three cases show the incident angle less than the critical angle, equal to critical angle and greater than critical angle; when the concept of evanescent wave arises, due to total internal reflection.
Figure 8: Three cases when the incidence angle of the light (a) less than critical angle (b) equal to critical angle and (c) greater than critical angle.
1.5 Types of Evanescent Wave Sensors
Evanescent wave sensors are those, in which the light of the evanescent wave interacts with the analyte, and this is further classified, based on the kind of interaction between the two [23]; and is as follows:
22
1. Open-Clad-Type Evanescent Wave Sensor
This type of a sensor typically has a three layer structure with the substrate,
guiding film; as well as the cladding, as shown in the Figure 9(a). It is called
Open clad type as the cladding acts as the absorptive medium.
2. Buffered-Clad-Type Evanescent Wave Sensor
This is a four layer type of a sensor as shown in the Figure 9(b). The cladding still
acts as the absorptive medium with the buffer layer as an addition to the previous
type of a sensor. Buffer layer is used to increase or decrease the sensitivity or
sometimes protects the guiding film from harmful substances [23].
3. Sensing-Layer-Type Evanescent Wave Sensor
This is also a four layer type of a sensor, where the cladding is no longer acting as
the absorptive medium, while the sensing layer does. Here, the analytes do not
absorb the working wavelength light directly and these sensors are the most
common type of Evanescent wave sensors [23].
23
Figure 9: (a) shows open clad type EWS, (b) shows buffered clad typpe EWS and (c) shows sensing layer type EWS [22]
1.6 Working Principle of Integrated Optic (IO) Sensors
Generally, efff ective refractive index (Neff) of a guided mode is considered to be the most crucial physical parameter under measurement. It is because for an IO sensor, a chemically selective layer or adlayer (as shown in Figure 10) based on [1], is deposited on the waveguide surface to bind the analyte (liquid or gaseous). Subsequently the evanescent field senses the changes in the index distribution caused by the layer, thus innducing an effective refractive index change [21, 24]. Figure 11, based on [21], depict a basic integrated sensor effect. Neff depends on the polarization (Transverse Electric, TE; 24 or Transverse Magnetic TM), mode number (m), wavelength (λ), other properties of the film, such as its refractive index, thickness and index of the substrate and cover.
The following effects cause effective refractive-index changes
1. Changes in the cover index containing the sample or anlayte.
2. Changes formed due to the formation (thickness) of an adlayer due to
adsorption or bound molecules formed from the bulk gaseous or liquid
samples.
3. Only in case of a micro porous waveguide, an effect called adsorption
or desorption of molecules causes changes in its refractive index, thus
causing changes to the effective refractive index. In this case, the
evanescent field is not responsible but only the field in the waveguide
[21].
When the above three effects together result in the effective index change, then
⎛ ∂N ⎞ ⎛ ∂N ⎞ ⎛ ∂N ⎞ (Equation 1.2) ΔN = ⎜ ⎟d + ⎜ ⎟Δn + ⎜ ⎟Δn ⎜ ∂d ⎟ F′ ⎜ ∂n ⎟ C ⎜ ∂n ⎟ F ⎝ F′ ⎠ ⎝ C ⎠ ⎝ F ⎠
Figure 10: An integrated optic sensor with an adlayer 25
Figure 11: Basic integrated sensor effects.
1.7 Single Mode Operation
The main reason for choosing the fundamental mode or single mode of operation is to avoid the modal dispersion caused in multimode waveguide structures [25]. The mode number m determines the number of modes, with m=0, corresponding to the fundamental mode and m=1, 2, 3…. corresponding to first, second, third and higher order modes. The figure below shows the modal dispersion curves taken from [26] for both transverse electric (TE) and transverse magnetic (TM) modes with core index being 3.38, cladding and substrate index being 3.17. It is a plot of the normalized parameters. As it can be seen for the figure that the first curve represents the fundamental mode (TE and TM), while the others correspond to the higher order modes. 26
Figure 12: Dispersion curve [24].
1.8 LOOM Structure
The difficulty faced by most evanescent wave sensor architectures lies in the lack of strong interaction between the analytes (target species) and the evanescent field. This is primarily due to the use of strongly confined waveguide designs [11]. The concept of
Low Optical Overlap Mode (LOOM) structures arises, where a nanoscale core region is used, which results in a large and distended mode [27].These structures have a very small percentage (usually less than 1%) of the optical mode confined in the guiding material.
This is discussed in detail in the future chapters. These large modes are sufficient to provide a strong interaction with the analytes, as the field is pushed out of the core region, enhancing the field interaction with the analytes. All the upper and lower cladding 27 layers are air, while the core is a dielectric or semiconductor depending on the application. Figures 13, 14 and 15 show the LOOM structure, SEM of the structure and
CCD image of the LOOM taken from [11].
Figure 13: The structure is a nanoscale membrane waveguide where the guided optical mode is represented by the dashed circle lines [11].
Figure 14: The SEM shows an InP-based LOOM guide indicating (a) membrane, (b) rib, (c) effluent vias, and (d) MEMS release V-groove [11].
28
Figure 15: A CCD imaage of a guided LOOM mode with dashed lines indicating V- groove position [11].
1.9 Analytes
1.9.1 Gaseous Detection of Ammonia
Ammonia is one of the most harmful gaseous pollutants present in the atmosphere. It is corrosive in nature, colorless and has a strong odor. Ammonia is present in air, wateer and soil naturally. The main source of ammonia entering the human body is via breathing.
The Environmental Protection Agency (EPA) reports show that exposure to excess amounts of ammonia cause irritation and burns to eyes, mouth, skin and lungs and sometimes may cause permanent blindness, lung disease or even death. The maximum permissible level of ammonia in air is 50ppm, above which it causes irritation and bad odor. The absorption spectra of gaseous ammonia are shown in Appendix G.0. Thus, detection of ammonia in air is important for human safety [28].
1.9.2 Aqueous Detection of Nitrites
One of the harmful agents present in water is nitrogen compounds, which are present in the form of nitrites, nitrates and ammonia [29]. Nitrites and nitrates are compounds of 29 nitrogen atom that contain two oxygen atoms attached to the nitrogen, while for nitrates; they are joined to three oxygen atoms. Nitrites and nitrates are readily converted into each other. Runoff from the golf courses, home lawns, gardens, farms, and septic systems, excess usage of fertilizers and bad management of animal feedlots are some of the sources for nitrate content in water. Division of Environmental and Occupational
Health Services Consumer and Environmental Health Services report that the maximum contaminant levels of nitrates can be 10 ppm, nitrites can be 1 ppm, while both combined can be 10 ppm in water. Excess nitrogen compounds in human body react with hemoglobin in the blood and this reduces its ability to carry oxygen to various parts of the body. It even causes a disease called blue baby syndrome in infants. Also, it is found that they cause cancer in animals [30]. Thus determining the nitrite/ nitrate content in water is important to monitor the pollution levels.
1.10 Material Used
1.10.1 Indium Phosphide (InP)
Indium Phosphide is applicable only at a wavelength of 1.55µm and not below, as this is very lossy at lower wavelengths. Since, the waveguide absorption loss is the product of the modal fill factor and the material absorption, a lower waveguide loss is possible, for a
LOOM structure with high quality core material [11]. Compared to most frequently used semiconductors such as silicon and gallium arsenide, for high power and high frequency electronic applications, InP has an edge due to its high electron velocity. It is used widely in the integration of active devices (lasers), as it has a direct band gap. 30
1.10.2 Amorphous Zinc Oxide (a-ZnO)
Amorphous Zinc Oxide (a-ZnO) has received popularity in the recent past due to its physical and chemical properties and wide range of applications. It is a wide band gap material of 3.37 eV at room temperature; which makes it a promising candidate for short wavelength emitters [31-32] besides having higher breakdown voltages, ability to sustain high electric fields, high-power operation ability and reduced electronic noise [32]. It has excellent chemical and thermal stability, has electrical and optoelectronic properties for solar cells [33], and has high exciton binding energy which enables semiconductor lasers with reduced thresholds [34]. It is observed that this material is lossless at a wavelength around 550 nm [35], because of the wide band gap, while most other materials are lossy around that wavelength and thus it is used as a waveguide at lower wavelengths. Since most of the analytes as discussed in the above section have high absorption at lower wavelengths, using this material proves to improve the sensitivity of the sensor device.
Since the absorption of light by the anlaytes determines the sensor sensitivity, i.e by reducing the output at the detector, which detects the amount of anlaytes present, thus, it is demanded that the material used is lossless at that wavelength. Thus, choosing a-ZnO for EWS waveguide technique proves to be advantageous.
31
CHAPTER 2: ANALYTICAL AND NUMERICAL METHODS TO OBTAIN
EFFECTIVE INDEX
Effective indices of optical waveguides are often calculated using a wide variety of methods, both analytical and numerical. Analytical methods include the Effective Index method and Marcitili method which are simple and often yield an accurate result; while numerical methods (such as Beam Propagation methods (BPM) and Finite Element
Method (FEM)) solve for much more complex waveguide structures, which incorporates a numerical technique such as Fast Fourier Transform, Finite Difference and Finite
Element Methods [37].
The main aim of this chapter is to calculate the effective index using the analytical method: “Effective Index Method” and the numerical methods: “Finite Element Method” and “Semi-vectorial Beam propagation Method”. Using the values obtained from the effective index method, the vertical cut profile for the electric field is plotted analytically and compared with the other methods. Modal analyses of the structures are performed analytically and critical parameters such as modal fill factor and confinement factor are determined in this chapter.
2.1 Characteristics of Optical Waveguides
In general the fundamental structure of a dielectric waveguide consists of a high index core region, also the film, surrounded by a lower index cladding region, the bottom cladding, (sometimes called the substrate) and the upper cladding (cover) as shown in
Figure 16. The propagation of a guided optical wave is in the longitudinal direction, parallel to the planar structure. There are two kinds of waveguides, planar optical 32 waveguides and non planar optical waveguides. They are divided based on the feature of optical confinement, which is the interaction between the cladding region and the core. In the planar waveguides, the optical confinement is restricted to one direction, while for the non planar waveguides; it is surrounded by all directions as shown in the Figure 16(a) and 16(b) [38].
(a) 1D (b) 2D
Figure 16: Fundamental structure of a dielectric waveeguide and (a) 1D planar Optical waveguide and (b) 2D non planar optical waveguuide.
Further, the planar waveguides are divided based on the refractive index distribution, as step index and graded index waveguides. Step Index waveguides have constant refractive index in the core, sandwiched between the substrate and the cover. They have an abrupt index difference between the core and the cladding regions. Frequently, the cover is air, 33 with an index of 1. When the cover index and the substrate index are the same, the waveguide is said to be Symmetric, if different, the term Asymmetric is used [38].
Graded index waveguides have the refractive index dependent upon the deposition profile
[39]. Figure 17 shows the index profiles of a step and graded index waveguide.
Figure 17: Index profile distributions of step index and graded index waveguides.
2.2 Channel Waveguides
Channel waveguides are an important group of a non planar waveguide. The three main types are the Buried waveguides, Stripe waveguides and the Rib waveguides [38].
2.2.1 Buried Channel Waveguides
This type of waveguide is formed by inducing a high index core region, which is buried in the lower index substrate region. The typical cross section geometry of the core is rectangular, though it can have any shape. The Figure 18(a) shows a buried channel waveguide with a rectangular core, having a width ‘w’ and a thickness ‘d’ [38]. It is 34 difficult to grow this type of a waveguide structure and requires re-growth, the advantage of using the buried channel waveguides is its high confinement factor.
2.2.2 Stripe-Loaded Waveguides
This type of a waveguide includes a strip on top of a core with an index of n1 greater than n3, that of the strip. The waveguide core beneath the loading strip has a thickness ‘d’, and the width ‘w’, described by the loading strip as shown in Figure 18(b) [39]
2.2.3 Ridge Waveguides
A ridge waveguide is similar to the stripe waveguide, except that the ridge on top is the core, with its index higher than the surrounding regions, thus has better optical confinement. Here, the thickness of the core is‘d’, that of the ridge and the width is ‘w’ as shown in the Figure 18(c). [38]
35
18(a) 18(b)
18(c)
Figure 18: showing the 3 kinds of channel waveguides, 18(a) buried waveguide, 18(b) Stripe waveguide and 18(c) ridge waveguide
The other types of the Channel waveguides are described as below
2.2.4 Rib Waveguides
This has a structure similar to strip or ridge waveguide, but the only difference is that the strip/ridge together with its lower surface constitutes the core region. The thickness and width being ‘d’ and ‘w’, defined by the rib, as shown in the Figure 19(a) [38]. These kinds of waveguides are modeled in this thesis.
36
2.2.5 Diffused Waveguides
This kind of waveguide is fabricated when a high index core region is grown in the substrate with a process of diffusion. Due to the diffusion process, the boundaries are not sharp, but curved as shown in the Figure 19(b). It has a thickness‘d’ and a width ‘w’ [38].
(a) (b)
Figure 19: Rib waveguide and diffused waveguides structures as in (a) and (b).
2.3 Analytical Solution: Effective Index Method (EIM)
EIM is one of the simplest approximate methods to find the effective indices in channel waveguides with arbitrary index profiles and waveguide geometries [38]. Since it is difficult to analyze the channel waveguides (rib waveguides) as the optical confinement is dependent on two directions, it is hence necessary to understand the physics associated with the structures to analyze the index distribution and hence, the analytical approach becomes critical. The basic concept is to convert a channel waveguide into two planar waveguides, treated independently to find the effective index of the propagation mode for each, as shown in the Figure 20.
37
Figure 20: Effective index approximation method, showing two planar waveguides in two different directions.
The EIM solves for two planar waveguides, described in two directions: x and y, while planar waveguide 1 has the light confined in x direction, while the planar waveguide 2 has it in the y direction. EIM is a good approximation, if it satisfies the following conditions [38].
1. The width of the waveguide is larger than its thickness, w>d
2. The wave guiding along its width is not stronger than its perpendicular or in its
thickness direction.
EIM can be applied to Step index and Graded index waveguides, and also to all of the non planar waveguides described before (if the two conditions mentioned above are satisfied) [38].
The technique of EIM is to solve for the propagation constants of the planar waveguides, and then obtain the effective indices; Figure 21 shows the flowchart to obtain the effective index, using EIM. Since effective index is mode dependent, we must specify the 38 mode (number of the mode) we are looking for. In this technique we assume that when the electric field is in the y direction or parallel to the planar surface, it is Transverse
Electric (TE)-like mode, while when it is in x direction or perpendicular to the planar surface, then it is Transverse Magnetic (TM)-like mode [38].
Figure 21: Flowchart of EIM.
A LOOM based rib waveguide structure is considered, with its core being InP, (n1=3.17) at a wavelength of 1550nm, substrate and cover being air, (n2,3=1) thickness of the membrane is 50nm, and the rib thickness is 50nm, and the width ‘a’ of the rib is 3µm, as shown in Figure 15. This structure is divided into three regions, Region 1, Region 2 and
Region 3 as in Figure 20. Each of these regions are treated as a symmetric slab planar waveguide structure with a propagation constant β and hence the effective index Neff, 39
(Neff = λβ/2π) are calculated. Since the effective index is mode dependent as mentioned before, the x-dependence of the y component for the TE like mode, or the electric field,
Êm, y (x), and the magnetic field Ĥm, y (x) for TM like mode is calculated [38]. Here, ‘m’ corresponds to a specific mode number. The above mentioned procedure is used to calculate β1, β2 and β3, for the three regions and hence the corresponding Neff1, Neff2 and
Neff3 are calculated as shown in Table 2 for the resonant and non-resonant structures.
Region 1 consists of a core index n1 = 3.17 and n2 = n3 = 1, with a width of 100nm, while regions 2 and 3, have the similar index distribution, but the width is 50nm each. For the structure proposed here, β2 = β3. After obtaining the three effective indices, we treat the structure with n1=Neff1 and n2 = n3 = Neff2 or Neff3. Now, we apply the above procedure again to calculate the propagation constant for the final structure, having a y-dependence on the y-component for the TE like mode or the electric field Ên, y(y) and the magnetic field Ĥn, y(y) for TM like mode. Firstly, the three regions are analyzed in the x-direction
(vertical), the corresponding effective indices are obtained and those effective indices are considered to get a final structure in the y-direction (horizontal), with a width of 3µm.
This final structure is again a three slab symmetric waveguide as mentioned before, but for this final structure, to obtain a TE- like mode, we consider the TM field of the vertical structure; and for TM-like mode, it’s the TE field of the initial vertical structure [38]. The two LOOM waveguide structures, with structure 1 having a core of InP and structure 2 having a-ZnO are as shown in Figure 22.
40
Figure 22: Two LOOM based structures, with structure 1 having InP as the core and structure 2 having a-ZnO as core material and the other dimensions are as shown above.
2.3.1 Theoretical Calculation of Effective Index for a Symmetric Planar Waveguide
Structure
From [38], we understand that “a waveguide mode is a transverse field pattern whose amplitude and polarization profile remains constant along the longitudinal direction”.
The electric and magnetic fields of a mode can be written as
ˆ (Equation 2.1) Eυ (r,t) = Eυ (x, y)exp(iβυ z −iωt)
ˆ (Equation 2.2) Hυ (r,t) = Hυ (x, y)exp(iβυ z −iωt)
ˆ ˆ Where, ‘υ ’ is the mode index, Eυ (x, y) and Hυ (x, y) are the mode field patterns and βυ is the propagation constant of the mode [37].
Using the basic Maxwell Equations, we obtain for a linearly, isotropic dielectric waveguides [38-39]. 41
δH (Equation 2.3) ∇× E = −μ 0 δt
δE (Equation 2.4) ∇ × H = ε δt
Since, the optical fields in the waveguides are in the form of equations (2.1) and (2.2), the above Maxwell equations (2.3) and (2.4) are simplified as following
∂Eˆ (Equation 2.5) z − iβEˆ = iωμ Hˆ ∂y y 0 x
∂Eˆ (Equation 2.6) iβEˆ − z = iωμ Hˆ x ∂x 0 y
∂Eˆ ∂Eˆ (Equation 2.7) y − x = iωμ Hˆ ∂x ∂y 0 z
∂Hˆ (Equation 2.8) z − iβHˆ = iωεEˆ ∂y y x
∂Hˆ (Equation 2.9) iβHˆ − z = iωεEˆ x ∂x y
∂Hˆ ∂Hˆ (Equation 2.10) y − x = iωεEˆ ∂x ∂y z
From the above equations, the transverse components of the electric field and magnetic fields are expressed in terms of the longitudinal components [38] as:
∂Eˆ ∂Hˆ (Equation 2.11) ()k 2 − β 2 Eˆ = iβ z + iωμ z x ∂x 0 ∂y
∂Eˆ ∂Hˆ (Equation 2.12) ()k 2 − β 2 Eˆ = iβ z + iωμ z y ∂y 0 ∂x 42
∂Hˆ ∂Eˆ (Equation 2.13) ()k 2 − β 2 Hˆ = iβ z + iωε z x ∂x ∂y
∂Hˆ ∂Eˆ (Equation 2.14) ()k 2 − β 2 Hˆ = iβ z + iωε z y ∂y ∂x
Where, k 2 = ω 2μ ε (x, y) (Equation 2.15) 0
Thus, once the longitudinal components Ez and Hz are known, the other field components
Ex, Ey, Hx, and Hy can be calculated.
2.3.2 Wave Equations of a planar waveguide
The two modes of a planar, 3-slab waveguide are either TE or TM. The homogeneous wave equations of a planar waveguide with an index distribution n(x) and is independent of the y-axis [36] is as follows:
TE Mode:
Using equations (2.11)–(2.14), for any TE mode we have Ez=Ex=Hy=0 and
∂Hz/∂y=0.Thus, the only non zero components left are Hx, Ey and Hz [36] .The wave equation for Ey can be derived using other field components as,
∂ 2 Eˆ (Equation 2.16) y + ()k 2 − β 2 Eˆ = 0 ∂x2 y
ω 2 (Equation 2.17) where, k 2 = ω 2μ ε (x) = n2 (x) 0 c2
And, the other components are obtained in terms of Ey as follows
ˆ β ˆ (Equation 2.18) H x = − E y ωμ 0 43
1 ∂Eˆ (Equation 2.19) Hˆ = y z iωμ ∂x 0
TM Mode:
ˆ In case of a TM mode, Hz=Hx=Ey=0 as ∂ Ez / ∂y =0. Thus, the non zero components are
Ex, Hy, Ez The wave equation for Hy can be derived using other field components [38] as
∂ 2 Hˆ 1 dε ∂Hˆ (Equation 2.20) y + ()k 2 − β 2 Hˆ = y ∂x 2 y ε dx ∂x
Where, k 2 is the same as defined in equation (2.17).
β (Equation 2.21) Eˆ = − Hˆ x ωε y
1 ∂Hˆ (Equation 2.22) Eˆ = y z iωε ∂x
As mentioned previously, the structure is broken down into 3 regions, with each region being a symmetric waveguide structure. In the analysis below, we can see how the effective index of region 1 is solved.
2.3.3 Other Important Waveguide Parameters
Asymmetry Factor: This is the measure of the asymmetry of the waveguide meaning that the cladding regions are different, which depends on the polarization of the mode [39].
2 2 n2 − n3 (Equation 2.23a) aE = 2 2 For TE modes n1 − n2
4 2 2 n1 n2 − n3 (Equation 2.23b) aM = 4 2 2 For TM modes n3 n1 − n2 44
Where, a = 0 for symmetric waveguides as, n2 = n3.
Normalized Waveguide Thickness or the V-number: It is a dimensionless normalized parameter of a waveguide and determines the number of modes supported and is defined as
2π (Equation 2.24) V = d n2 − n2 λ 1 2
Where k is the propagation constant,
2π (Equation 2.25) k = λ
2.3.4 Mode Parameters
A guided mode is one in which the power is retained inside the core as the wave, is perfectly reflected off both the interfaces. This mode can exist only if it satisfies a transverse resonance condition, such that the reflected wave has constructive interference with itself. The transverse component (x-component) of the wave vector inside the core is h1 = k1cosθ, where θ is the angle of incidence and the longitudinal component (z- component) is β = k1sinθ. Similarly, the transverse components for the substrate and cover regions can be defined as h2 = k2cosθ, and h3=k3sinθ, where k1, k2 and k3 are the propagation constants in the respective regions [38], defined as
2πni (Equation 2.26) ki = ,i = 1,2,3 λ
Applying boundary conditions at the n1/n2 and n1/n3 interfaces, the above can be rewritten as, 45
2 2 2 (Equation 2.27) k − β = h1
2 2 2 (Equation 2.28) β − k2 = γ 2
2 2 2 (Equation 2.29) β − k3 = γ 3
But, equations 2.28 and 2.29 are imaginary as β>k2>k3, and the guided mode fields decay exponentially in the transverse directions of the substrate and the cover regions, thus introducing new parameters called decay parameters as γ2=|h2| and γ3=|h3| [38].
The only parameter to be determined is β, which specifies the waveguide mode as the transverse field parameters h1, γ2 or h2 and γ3 or h3 are characterized by k1, k2 and k3, which are well determined. Another interesting feature relating mode number and the evanescent field is that, as the mode number increases, the penetration of the wave corresponding to that mode deepens. This phenomenon occurs because as there is an increase in the mode number, the propagation constant decreases, thus lowering γ3 and hence, enhances the penetration depth [39].
2.3.5 Eigen Value Equation
Equation for TE [38]
⎛ h d mπ ⎞ V 2 − h2d 2 (Equation 2.30) tan⎜ 1 − ⎟ = 1 , m= 0, 1, 2.. ⎝ 2 2 ⎠ h1d
Equation for TM [38]
2 2 2 2 ⎛ h1d mπ ⎞ n1 V − h1 d (Equation 2.31) tan⎜ − ⎟ = 2 , m= 0, 1, 2,.. ⎝ 2 2 ⎠ n2 h1d
Where, ‘m’ determines the mode number. 46
The equations above mentioned are solved using Matlab, graphically by plotting left hand side and right hand side, as a function of (h1d). The solutions give the value of β, and hence the effective index, Neff, could be calculated from the relation Neff = λβ/2π,
as discussed before.
Figure 23 shows the graphical solution obtained by plotting the left and right hand side of the eigen value equation, for a TM mode, which determines h1d for structure1.
Figure 23: Matlab output showing a plot of left and right hand side of the Eigen value equation for structure1(TM).
After obtaining the neff for structure1 as shown in Appendix A.0, which is for region 1 in
Figure 20, a similar analysis for region 2 and region 3 is performed as in Appendix A.0, but the value of d is changed to 50nm from 100nm. Also, for this structure region 2 and region 3 are identical to each other. After obtaining the effective indices for the 3 regions, 47 as previously mentioned, the structure is analyzed as a 3 slab waveguide, with a width of
3µm, as shown in Figure 20.
Again, a similar analysis is done to obtain the final effective index of a quasi-TE mode, for the entire structure as in Appendix A.1. Similarly, calculations are performed for a
TM mode as in Appendix A.2, in the similar way as done for the TE modal analysis.
Table 1 shows the optical and geometrical properties of the LOOM based waveguide structures
Table 1: Optical and geometrical properties of the LOOM based waveguide structures.
Structure 1 Structure 2 Structure 3
Operating wavelen nm) 1550 1550 550
Substrate index ns 1 1 1
Core index nf 3.17 1.79 1.79
Cover index nc 1 1 1 thickness h1(nm) 100 100 100
Thickness h2 (nm) 50 50 50
Width (nm) 3000 2000 2000
Table 2 shows the effective indices obtained for two structures, using InP and a-ZnO, both at 1550nm and a-ZnO at 550nm when solved in the x-direction. 48
Table 2: List of effective index values obtained.
Structure Effective index Region II Region I Region III Region I (100 nm Regions II, III (50 nm
TE mode TM mode TE mode TM mode
1.
1.8114 1.005 1.3204 1
2.
1.071 1.009 1.0746 1.005
3.
1.357 1.103 1.152 1.0234
The results obtained for the final structures, in the y-direction are discussed in the chapter
5 “Results and Discussion”. The numerical methods are discussed as below 49
2.4 Effective Index Calculations using Finite Element Method (FEM)
This is the second method based on Computer Aided Design (CAD) employed for the effective index calculations. FemSIM is a product or simulation tool of Rsoft Design
Group based on the Finite Element Method. It is a generalized mode solver that calculates a transverse or a cavity mode on a non-uniform mesh, for any arbitrary structure. Some of the other waveguide related applications of finite element method are plasmonics, surface plasmon resonance, microwave crystal waveguides, photonic crystal waveguides, optical fibers and waveguides [40]. FemSIM divides the entire waveguide into small rectangular blocks or mesh, where boundary conditions are applied and
Maxwell’s equations are solved for each block [35]. It calculates for the effective index for each mode specified, and filters the TE and TM modes accordingly.
For TE modes, the symmetry options in the mesh options are made to be symmetric, and the modes are sorted with respect to highest effective index in the mode options, while for TM modes, it is set to anti-symmetric in mesh options and the modes are sorted from lowest losses. By incorporating these settings in the software, we can filter both TE and
TM modes, for the structures under examination.
The simulation results obtained from the FEM are shown in the Figures 24-28. Figure 24 represents the mesh plot of the rib waveguide structure, which is divided into rectangular blocks and Maxwell’s equations are solved across each box along the boundaries. Figure
25 shows the mode output plot, in terms of the intensity, where scale on the right side of the Figure shows the color having maximum and minimum intensity. This gives the 50 effective index of the waveguide structure. Figures 26 and 27 show the vertical and horizontal electric field mode profiles, cut at x= -0.0036 and y= -0.013 respectively.
Figure 28 shows the contour map of transverse index profile with the core having an index of 3.17 and the other region being air [41].
Figure 24: Figure showing mesh plot of structure 1. 51
Figure 25: Figure showing the mode profile of structure 1.
Figure 26: Vertical Electric field mode profile of structure 1. 52
Figure 27: Horizontal electric filed mode profile of structure 1.
Figure 28: Figure showing the index profiles for structure 1. 53
Appendix B.0 gives the details of the FEM settings. The results obtained are compared to the effective index method in the chapter 5 Results and Discussion.
2.5 Effective Index Calculations Using Semi-Vectorial Beam Propagation Method
(BPM)
Beam Propagation Method (BPM) is the most common tool used for the study of optical devices, due to its simplicity and high speed of performance [42]. It solves for the transverse electric fields considering the boundary conditions, and also takes the polarization property of the propagating electromagnetic waves into account and is based on finite difference scheme [43]. Conventional BPM uses fast fourier transform (FFT) technique which has a limitation as the boundary conditions are ignored, while the recently developed finite difference (FD) techniques incorporate the vector boundary conditions. Neglecting the vector boundary conditions has an adverse affect, as the discontinuity at the index interfaces having large differences, affects the transverse component of the electric field [42]. The simulations and results using this method are performed and calculated by Dr. Gines Lifante, professor at Universidad Autonoma de
Madrid, Spain using finite difference BPM algorithms. The result obtained using this method is compared with the above methods in the results and discussion chapter.
2.6 Vertical Cut Profile: Electric Field
Modal Analysis of the Air LOOM structures is performed using the analytical method.
Firstly, the electric field is plotted analytically as shown in Appendix C.0. From the effective index method, the effective indices values are obtained, and plugging in these 54 values, the electric field equations are plotted. The propagation constant β can be obtained as Neff is known, and then γ2, γ3 and k are calculated using equations 2.29-2.31, and these are substituted in the field equations as shown below [39]:
-γ x Ae2 x 0