Document [original]

FAKULTÄT FÜR

ELEKTROTECHNIK,

INFORMATIK UND

MATHEMATIK

Low Pressure Chemical Vapor Deposition of Silicon

Nitride and Silicon Oxynitride Layers and their

Application in Optical Waveguide based

Chemical Sensors

Zur Erlangung des akademischen Grades

DOKTORINGENIEUR (Dr.-Ing.)

der Fakultät für Elektrotechnik, Informatik und Mathematik

der Universität Paderborn

genehmigte Dissertation

von

M.-Eng. Ahmed Tamim

Paderborn

Referent: Prof. Dr.-Ing. U. Hilleringmann

Korreferent: apl. Prof. Dr. rer. nat. Reinhart Job

Tag der mündlichen Prüfung: 24.08.2007

Paderborn, den 05.09.2007

Diss. EIM-E/SEN

FAKULTÄT FÜR

ELEKTROTECHNIK,

INFORMATIK UND

MATHEMATIK

Low Pressure Chemical Vapor Deposition of Silicon

Nitride and Silicon Oxynitride Layers and their

Application in Optical Waveguide based

Chemical Sensors

A Thesis Submitted to the Faculty of Electrical Engineering,

Computer Science and Mathematics – University of Paderborn in

Partially Fulfillment of the Requirements for the Degree of

Doctor-Engineer (Dr.-Eng)

In

Electrical Engineering

By

M.-Eng. Ahmed Tamim

Reviewers:

Prof. Dr.-Eng. Ulrich Hilleringmann

apl. Prof. Dr. rer. nat. Reinhart Job

Date of Thesis Submission: 30.05.2007

Date of Defence Examination: 24.08.2007

Paderborn, Germany

Acknowledgement

I would like to express my deepest gratitude towards my supervisor Prof. Dr.-Eng.

Ulrich Hilleringmann, for his guidance, moral support, assistance and encouragement

during my study here in Paderborn.

I would also like to thank apl. Prof. Dr. rer. nat. Reinhart Job for acting as second

reviewer, and Prof. Dr. techn. F. Gausch, Prof. Dr.-Eng. R. Noe, Prof. Dr.-Eng. U.

Rückert and Prof. Dr.-Eng. A. Thiede for being in my examination committee.

I would also like to thank also Dr.-Eng. Christoph Pannemann and all my

colleagues in sensor technology department for their help and friendship.

Finally I would like to thank my parents, my wife, my children and my brothers.

Acknowledgement

Abstract

This work involves optimization and characterization of low pressure chemical

vapour deposition (LPCVD) of silicon nitride and silicon oxynitride layers. The

optimized parameters of this deposition were used in fabricating the guiding layer of an

optical Mach-Zehnder interferometer which was used as a transducer for a waveguide

based sensor used to detect chemical gases such as ammonia. In this sensor a titanium

heater with aluminium contacts was integrated near to the reference arm in order to

increase the sensitivity of the sensor by using the thermo-optical effect. A chemo-

optical sensitive material to ammonia (its refractive index changes with changing the

amount of ammonia that diffuses into it from the ambient air) was spin coated on a

sensing window in the sensing arm.

The work in this thesis is split into three main areas of study. The first is designing

a monomode silicon oxynitride waveguide using the imaginary distance beam

propagation method (ID-BPM), also a design of waveguide parameters which increase

the sensitivity, designing the Mach-Zehnder interferometer and the heater. The masks

for the whole structure were designed using the Cadence program.

In the second area, LPCVD of silicon nitride and silicon oxynitride films were

optimized by the adjustment of the deposition temperature and the gases flow rate. The

homogeneity of the deposited layers, the deposition rate and the thickness variations

along the wafers and the boats were discussed.

In the third area, the detailed fabrication procedures of the sensor are discussed

including several important standard processes such as the thermal oxidation of the

silicon substrate, the low pressure chemical vapour deposition (LPCVD), the

photolithography for mask transfer, the reactive ion etching for a ridge waveguide

formation, the metal evaporation and wet etching. Also the characterizations of the

sensor were carried out. The waveguide was analyzed to make sure that the sensor is

working as expected. The heater also was tested to see how much dissipated power the

resistor can withstand during its heating, the heating effect on its resistance and the

heating effect on the transmission intensity were checked too. Finally, the refractive

index change in the sensitive layer during exposure to ammonia was measured. Also the

sensor response, the response time and sensitivity of the sensor were discussed.

Abstract

Zusammenfassung

In dieser Arbeit wurden die Abscheidung von Siliziumnitrid Und

Siliziumoxynitridfilmen nach dem Low Pressure Chemical Vapor Deposition

(LPCVD)-Verfahren zur Integration eines optischen Gassensors optimiert und die

resultierenden Schichten optisch charakterisiert. Mit den gewonnenen

Abscheideparametern wurden die Wellenleiter eines Mach-Zehnder-Interferometers

hergestellt, das als Signaltransformator in einem Ammoniaksensor dient. Dieser Sensor

nutzt ein Heizelement aus Titan mit Aluminiumkontakten, integriert neben dem

Referenzarm, um die Empfindlichkeit des Interferometers durch Ausnutzung des

thermo-optischen Effekts auf das Maximum abzugleichen. Ein chemo-optisch mit

Ammoniak reagierendes Material wurde im Spincoating-Verfahren auf die Sensorfläche

im Messzweig aufgetragen.

Im ersten Teil dieser Arbeit wurde ein Monomode-Siliziumoxynitrid-Wellenleiter

mit der Imaginary-Distance-Beam-Propagation-Methode (ID-BPM) entworfen. Die

Wellenleiterparameter wurden entsprechend gewählt, um das Mach-Zehnder-

Interferometer mit dem Heizelement zu optimieren. Die Masken für die vollständige

Struktur wurden mit dem Programm Cadence entworfen. Im zweiten Teil wurde die

Abscheidung von Siliziumnitrid und Siliziumoxynitrid im LPCVD-Verfahren anhand

von 45 Testwafern bezüglich der Abscheidetemperatur und der

Gasströmungsgeschwindigkeit optimiert. Die Homogenität der abgeschiedenen

Schichten, die Abscheiderate und die Schichtdickenvariationen wurden untersucht. Im

dritten Teil wurde die Herstellung des Sensors einschließlich der wichtigsten

Prozessschritte behandelt. Ebenso wurden die Wellenleiter sowie der gesamte Sensor

charakterisiert. Der Heizer wurde hinsichtlich der maximal zulässigen Leistung

analysiert, insbesondere die Auswirkung der Erwärmung auf den elektrischen

Widerstand getestet. Schließlich wurde die Änderung des Brechungsindexes in der

empfindlichen Schicht in Abhängigkeit von der Ammoniakkonzentration bestimmt.

Ebenso wurden das Sensorübertragungsverhalten, die Ansprechzeit und die

Empfindlichkeit des Sensors untersucht.

Zusammenfassung

Contents i

Contents

INTRODUCTION ....................................................................................... 1

1 PROCESSING OF WAVEGUIDE SENSORS ..................................... 3

1.1 Silicon technology .................................................................................................. 3

1.2 Low pressure chemical vapor deposition (LPCVD) ............................................... 3

1.3 Silicon nitride (Si3N4) ............................................................................................. 5

1.4 Silicon oxynitride (SiON) ....................................................................................... 6

1.5 Optical waveguides ................................................................................................. 7

1.5.1 Light coupling into waveguides ..................................................................... 13

1.5.2 Optical losses ................................................................................................. 15

1.6 Chemical sensors .................................................................................................. 16

1.6.1 Background .................................................................................................... 16

1.6.2 Historical Perspective of chemical sensors .................................................... 19

1.6.3 Applications of chemical sensors .................................................................. 25

1.7 Optical waveguide based chemical sensors .......................................................... 27

1.7.1 Refractive chemical sensors ........................................................................... 28

1.8 Ammonia sensors .................................................................................................. 33

2 THEORETICAL CONSIDERATIONS .............................................. 35

2.1 Introduction ........................................................................................................... 35

2.2 Waveguide design ................................................................................................. 36

2.3 Design of Mach–Zehnder interferometer (MZI) .................................................. 50

2.4 The heater design .................................................................................................. 55

3 DEPOSITION OF SILICON NITRIDE AND SILICON

OXYNITRIDE BY LPCVD ..................................................................... 59

3.1 Introduction ........................................................................................................... 59

3.2 Description of LPCVD process ............................................................................ 59

3.3 Measuring of layer thickness and refractive index by ellipsometer ..................... 63

3.4 Preparation of the silicon wafers .......................................................................... 64

3.5 Deposition process ................................................................................................ 64

3.5.1 Silicon nitride (Si3N4) deposition .................................................................. 64

Contents

ii

3.5.1.1 Deposition of Si3N4 at 740 °C ................................................................. 65

3.5.1.2 Deposition of Si3N4 at 760 °C ................................................................. 66

3.5.1.3 Deposition of Si3N4 at 780 °C ................................................................. 67

3.5.1.4 Temperature optimization ....................................................................... 68

3.5.2 Silicon oxynitride (SiON) deposition ............................................................. 76

4 SILICON OXYNITRIDE WAVEGUIDE BASED AMMONIA

SENSOR ..................................................................................................... 81

4.1 Introduction ........................................................................................................... 81

4.2 Sensor fabrication .................................................................................................. 81

4.2.1 MZI waveguide fabrication ............................................................................ 81

4.2.2 Heater fabrication ........................................................................................... 91

4.2.3 Sensitive layer fabrication .............................................................................. 93

4.3 Experimental results .............................................................................................. 95

4.3.1 Waveguide characterization ........................................................................... 95

4.3.2 Light modulation ............................................................................................ 97

4.3.3 Ammonia sensing ......................................................................................... 100

5 CONCLUSIONS ................................................................................... 107

ABBREVIATIONS ................................................................................. 111

REFERENCES ........................................................................................ 113

LIST OF FIGURES ................................................................................ 117

LIST OF TABLES .................................................................................. 121

Introduction 1

Introduction

The main goal of this thesis work is the optimization of LPCVD to find the

deposition parameters such as deposition temperature, deposition pressure and gases

flow rate which will be used in fabricating the guiding layer of the MZI waveguide.

This layer will be used to verify the design requirements such as refractive index value,

thickness and optical properties which will make the designed waveguide monomode

and will result in high surface sensitivity. This waveguide will be used as a transducer

for a waveguide based sensor used to detect chemical gases such as ammonia. In this

sensor a titanium heater with aluminium contacts will be integrated near to the reference

arm of the MZI in order to increase the sensitivity of the sensor by using the thermo-

optical effect. For ammonia sensing a chemo-optical material sensitive to ammonia will

be placed at a sensing window in the sensing arm. The processing facilities in our

research laboratory were employed throughout the study. This thesis is consisting of

five chapters;

Chapter one includes background on silicon technology, low pressure chemical

vapor deposition, the optical guiding materials such as silicon nitride and silicon

oxynitride, the optical waveguides and chemical sensors.

Chapter two concerns the sensor design including the design of the waveguide in

order to verify two conditions: it has to be monomode and it should show a high surface

sensitivity. It explains the design of the Mach-Zehnder interferometer and the design of

the heater.

In chapter three the optimization of low pressure chemical vapor deposition for

silicon nitride and silicon oxynitride is presented. The gases flow rate, the deposition

temperature, the thickness variation along the wafers, the deposition rate and the

homogeneity of the deposited layers were discussed.

Chapter four describes the process steps applied in the fabrication of the silicon

oxynitride waveguide based ammonia sensor and the characteristics of the sensor such

as total losses, light modulation and ammonia sensing.

Finally, in chapter five the conclusions that are drawn from this study are

presented.

2

1 Processing of waveguide sensors 3

1 Processing of waveguide sensors

1.1 Silicon technology

Silicon technology still remains the most dominant force in electronic integrated

circuits and it seems that it will continue in the near future. The work towards

implementation of silicon technology in the integrated optics area started ever since the

notion of integrated optics was introduced. There are several key factors for this

application to be successful. First of all, any candidate technology must provide

waveguide structures with low propagation losses. Moreover, it should be compatible

with microelectronics device processing and provide a high reproducibility at

reasonably low cost. The coupling between optical integrated circuits (OIC) and other

elements such as optical waveguides, light sources, and photodetectors is another

important issue of consideration. Si-based technologies offer crucial advantages on all

these points, which make it more attractive than other competing technologies.

In order to look for silicon compatible materials, one of the requirements is the

availability of thin dielectric films with a large refractive index difference between the

core material and the cladding material. In fact, the device size scales down with

increasing the refractive index difference. Unfortunately, it is quite difficult to achieve

low propagation losses for large refractive index differences and the scattering losses

increase as the square of the refractive index difference, and therefore high index core

waveguides are very sensitive to interface roughness. A second requirement on the

materials is the compatibility with the conventional silicon technology. The materials

have to withstand the process conditions such as purity, temperature, compatibility,

stability, functionality and should be processable with the tools already common in the

microelectronics industry. Low pressure chemical vapor deposition (LPCVD) is a very

important process for silicon technology.

1.2 Low pressure chemical vapor deposition (LPCVD)

Chemical vapor deposition (CVD) is defined as the formation of a non-volatile

solid film on a substrate by the reaction of vapor phase chemicals (reactants) that

contain the required constituents.

A CVD process can be summarized as consisting of the following sequence of

steps:

• A given composition (and flow rate) of reactant gases is introduced into a

reactant chamber

• The gas species move to the substrate

• The reactants are absorbed on the substrate

1 Processing of waveguide sensors

4

• The atoms undergo migration and film-forming chemical reactions

• The gaseous by products of the reaction are desorbed and removed from the

reaction chamber

Gas

Control

Pressure

Sensor

Mechanical

Pump

Sensor

Three-zone resistance heated furnace

Exhaust

Pressure

Temperature

Control

Figure (1-1): Block diagram of a low pressure chemical vapor deposition system

LPCVD can be used for deposition of silicon nitride (Si3N4) and silicon oxynitride

(SiON). The low pressure chemical vapor deposition (LPCVD) process shown in figure

(1-1) deposits the films at relatively high temperatures and low pressures and depends

upon four critical parameters which are the total flow rate of the gasses, gas ratio of

ammonia (NH3), oxygen (O2) and triethylsilane (C6H16Si), chamber pressure and

chamber temperature. The most significant factor in the (LPCVD) silicon nitride

furnace is the gas ratio. By modifying the ratio at which gases enter the gas chamber

one can drastically change the film properties obtained at a specified point within the

chamber. By increasing the amount of ammonia that is in the chamber, the film stress

becomes increasingly tensile, whereas, by a decrease of the amount of ammonia in the

chamber the stress of the film becomes increasingly compressive. By balancing these

two extremes, one can create a point in the chamber that closely approximates zero

stress.

The second significant factor in a LPCVD silicon nitride furnace is the chamber

pressure. By increasing the pressure of the chamber, the extent of the reaction is not

allowed to move too far down the chamber (towards the vacuum). Hence by increasing

the pressure of the chamber for a fixed gas ratio, total flow and temperature, the

resulting film will become more compressive. On the other hand, by decreasing the

1 Processing of waveguide sensors 5

pressure within the chamber, the extent of the reaction will migrate further down the

chamber and the resulting thin films will be more tensile.

A similar relationship is involved with the total flow rate of the furnace. By

increasing the total flow, the reactions can be pushed further down the chamber (with

high flow rates) or keep them confined to the point of entry (with low flow rates). By

increasing the flow rates, the film will be more tensile, and by decreasing the flow rates,

the film will be more compressive [1]. Reduced pressures tend to reduce unwanted gas-

phase reactions and improve film uniformity across the wafer. Increasing the chamber

temperatures tend to increase the deposition rate.

1.3 Silicon nitride (Si3N4)

Bulk silicon nitride (Si3N4) is a hard, dense, refractory material. Its structure is

quite different from that of silicon dioxide: instead of flexible, adjustable Si-O-Si bridge

bonds, the Si-N-Si structure is rendered rigid by the necessity of nitrogen forming three

rather than two bonds. CVD silicon nitride is generally amorphous, but the material is

much more constrained in structure than the oxide. As a consequence, the nitride is

harder, has higher stress levels, and cracks more readily.

The dense structure of silicon nitride does not provide the open channels found in

oxide structures; thus, the nitride is widely employed in electronics as a barrier material.

Even hydrogen diffuses slowly in the nitride film, and other small positive ions (Na+ or

K+) are effectively blocked by thin nitride layers. Since oxygen diffuses very slowly

through the nitride, deposited nitride can prevent an oxidation of the underlying silicon:

this property is exploited in local-oxidation-of-silicon (LOCOS) transistor isolation.

Nitride layers are also employed as etch stop layers both for wet etching and plasma

etching.

Deposited nitrides almost always contain hydrogen, typically much more than in

the comparable oxide films. The source of the hydrogen is the silane precursor and

possibly also the ammonia oxidant employed in most deposition schemes, but the

presence of hydrogen in the film is a consequence of the nitride structure. It is very

difficult for the atoms in an amorphous but constrained film like silicon nitride to

occupy all positions allowing the valence of each silicon and nitrogen atom to be filled:

that is, a lot of broken bonds are present. These bonds are readily occupied by hydrogen

atoms. The amount of hydrogen and the bonding (Si-H or N-H) can be measured by

infrared spectroscopy, and are important in characterizing the properties of plasma

nitrides. The stoichiometry of nitride films also varies widely so that the refractive

index can vary from about 1.8 to 2.2. This is another useful control parameter for nitride

deposition.

Silicon nitride thin films have been extensively used in various technological

areas, especially in microelectronic devices such as passivation layers, interlevel

insulators, and dielectrics. All these applications are due to silicon nitride’s remarkable

1 Processing of waveguide sensors

6

physical properties such as a high dielectric constant, high insulation strength, low

creepage, and good resistance against sodium as well as water vapor. In addition, they

also have excellent mechanical properties and stability. In addition to the above physical

and mechanical properties, silicon nitride films have another special characteristic,

namely the spectral selectivity, which can be used in optical coatings, used as a

diffusion mask for local oxidation of silicon in MOS processing and used as a guiding

layer in optical waveguides [2].

1.4 Silicon oxynitride (SiON)

There are some strong requirements for the devices for optical communication

such as low propagation loss, efficient fiber-to-chip coupling, low insertion loss, small

bend radii, and low fabrication cost that should be met for successful realization of the

technology.

Most of the materials used in integrated optics so far are classified as either low

contrast (e.g. silica, LiNbO3, polymers) or high contrast (e.g. InP, silicon on insulator).

The satisfaction of all of the above mentioned requirements simultaneously for both of

these systems is rather difficult if not impossible. In the low contrast systems the fiber-

to-chip coupling efficiency is excellent due to the large size of the single mode

waveguides, but they have a low integration density since large bend radii (10-30 mm)

are required for low loss operation. On the other hand, high contrast systems allow very

small bending radii (down to 0.15 mm) with low losses, but efficient fiber-to-chip

coupling is difficult to obtain due to small waveguide dimensions.

The commercially planar waveguide technology is typically based on low-index

contrast (0.7 %) silica material. This technology platform ensures planar waveguide

components with low propagation loss (less than 0.05 dB/cm) and low fiber-to-chip

coupling losses. However, to increase the level of functional integration and reduce the

fabrication cost, it is necessary to decrease the component size. Therefore core materials

with higher refractive index have become the center of attention in the field of

integrated optics devices for telecommunication. The propagation loss of high refractive

index contrast waveguides is influenced by the material absorption loss and scattering

by surface roughness of the waveguide core.

In recent years, growing attention has been paid to silicon oxynitride (SiOxNy or

SiON for short) as a potential material for integrated optics. This attention has been

motivated mainly by its excellent optical properties such as low absorption losses in the

visible and near infrared, a large refractive index range and compliance with standard

CMOS technology. SiON combines the dielectric properties of SiO2 together with good

chemical inertness and low permeability of Si3N4. In addition, the index of refraction of

SiON layers can be easily adjusted continuously over a wide range between 1.45 (SiO2)

- 2.0 (Si3N4) which comes to be a very attractive property that allows fabrication of

1 Processing of waveguide sensors 7

waveguides with desired characteristics of fiber match and compactness. Moreover, the

growth of SiON layers on silicon substrate is done by well established standard silicon

integrated circuit processing technology which is also a key point towards a low cost

mass production.

In addition, after the optimization of the processes it has become a well-controlled

technology with small technological tolerances. The relatively high index contrasts

which can be obtained enable a high functional density on the chip. It is quite natural to

apply in this technology the layer stack on the top of a silicon wafer which has several

advantages. The SiON technology is also used for integrated optical telecom devices.

The high contrast pure silicon technology using SOI wafers has only recently reached a

state in which it can be applied in practice; because of the high refractive index contrast

which can be obtained it might be most useful for circuits in which high compactness is

required such as in sensor arrays.

Amorphous silicon oxynitride layers are deposited by various techniques. The

most used growth technologies are plasma enhanced chemical vapor deposition

(PECVD) and low pressure chemical vapor deposition (LPCVD). The PECVD process

is found to be more efficient in controlled deposition of films with refractive indices

below 1.7. The deposited layers have good uniformity of the refractive index and in

layer thickness with run-to-run reproducibility [41].

1.5 Optical waveguides

Integrated optical systems generally consist of an optical chip which is provided

with a network of optical waveguides. The light is propagating through these

waveguides according to certain field patterns (the guided modes).

Figure (1-2): Total internal reflection of light [3]

1 Processing of waveguide sensors

8

Just as in optical fibers light is confined to the waveguide as a result of total

internal reflection (TIR) against material boundaries. This happens when light arrives

through an optically dense medium of refractive index n1 at its boundary with an

optically rarer medium of refractive index n2 (i.e., n1 > n2) and the angle of incidence θi

is greater than the critical angle θc where, sin θc = n2 / n1, as shown in figure (1-2) [3]

and figure (1-3).

Figure (1-3): Different light rays with different incidence angles are arriving to the

boundary of two mediums from a higher refractive index medium to a lower refractive

index medium where the total internal reflection occurs when the angle of incidence is

greater than the critical angle

Light in optical fibers is exposed to multiple total internal reflections which may

occur as a consequence of the core layer having a higher value of refractive index than

the cladding layer as shown in figure (1-4).

Cladding n2

Core n1, (n1 > n2)

c

θ

Li

g

ht source

n1

n2

Figure (1-4): Multiple total internal reflections in an optical fiber

1 Processing of waveguide sensors 9

Most IO systems are structured as a planar multilayer stack deposited on top of a

substrate (e.g. a glass slide or a silicon wafer). Sometimes the higher index core is

defined by locally implementing (mostly by diffusion) ions. Layer thicknesses are on

the order of magnitude of 0.1–10 µm just as the cross sectional dimensions of the core.

Lateral dimensions of a functional optical chip are in between some millimeters and

some centimeters.

In some IO systems there is no lateral patterning; in these systems, called

slabguides, the light is confined in the transverse direction only and not in the lateral

direction. These slabguides are useful for straightforward propagation of relatively wide

beams (hence with a low lateral divergence) only. They are seldom applied in integrated

optics, but nevertheless they will be met in several sensors. Mostly used are the

channel-type waveguides in which the light is also confined in the lateral direction [4].

Waveguides typically consist of a core with a higher refractive index n1 than the

surrounding cladding layer n2. This structure is transferred to the planar technology for

silicon wafer processing using different types of anorganic films, which are standard

materials in MOS technology. Silicon technology offers different transparent materials

with a wide range of refractive indices which can be used for light guiding films and

waveguide structures. The most common material for the cladding layers is silicon

dioxide (SiO2). It is normally used as a dielectric layer in MOS transistors. The

refractive index is about 1.46, and the absorbance rate is extremely low in the visible

spectral range. SiO2 can be deposited by thermal oxidation of silicon or by chemical

vapor deposition (CVD) using low pressure (LPCVD) or plasma enhanced CVD

(PECVD).

n1

n2

Silicon

Figure (1-5): The rib waveguide

Optical waveguides on silicon can use different types of setup. Figures (1-5), (1-6)

and (1-7) depicts the most common structures of waveguides suitable for monolithic

integration.

1 Processing of waveguide sensors

10

The rib waveguide [figure (1-5)] consists of a low refracting layer which is

covered by a layer with a higher refractive index as the light guiding film. The top layer

is partly etched to form a rib. The electromagnetic wave can only propagate in the area

of the rib, outside the rib no mode can exist due to the reduced film thickness. Because

of the roughness of the etched surface the propagation loss of this kind of waveguide is

large.

n1

Silicon

n2

Figure (1-6): The channel waveguide

n1

n2

Silicon

Figure (1-7): The strip loaded waveguide

Types of figure (1-6) waveguides (channel waveguides) depict less propagation

loss. This kind of waveguide can easily be integrated on silicon, but propagation loss is

not optimized due to the roughness of the vertical rib side walls caused by dry etching.

1 Processing of waveguide sensors 11

The strip loaded film waveguide shown in figure (1-7) exhibits very low

propagation loss. The surface cladding layer is deposited on top of the light guiding

film, and the etching process to form the strip does not affect the quality of the light

guiding layer, so propagation loss is very low. The integration process is as simple as in

case of the rib loaded waveguide.

In channel and rib waveguides, monomode behavior depends on the thickness, the

width and depth of the core and on the difference between the core and cladding

refractive indices. If the difference between the refractive index of the core and cladding

is large (higher than 10 %), monomode behavior is achieved with a core thickness of

hundreds of nanometers. The cladding thickness is a few micrometers due to the small

penetration of the evanescent field into the cladding. However, the rib depth must be

around several nanometers for monomode waveguides.

As shown in figure (1-8), the modes are in different locations for the rib and strip

loaded waveguides. For rib waveguides, the mode is confined just below the rib, which

gives a symmetric mode profile. In practice, it is difficult to produce such devices with

smooth rib sidewalls and since the mode is confined just below the rib, there will be

significant scattering loss. With strip loaded geometry, the mode is confined to the

higher index core below a lower index loading strip layer. Since the mode is not

contained in the strip, this configuration is insensitive to sidewall roughness.

(a) (b)

Figure (1-8): Schematic cross-section of waveguide structures with the location of the

mode in each case (a) Rib waveguide (b) Strip loaded waveguide

The difficulty in designing strip loaded waveguides is that their fundamental

modes are inherently elliptical, making them difficult to couple to optical fibers. It

1 Processing of waveguide sensors

12

requires much effort to obtain circular mode profiles suitable for coupling to optical

fibers.

The propagation of light through the waveguides is formally described by

Maxwell theory. It can be derived that for each wavelength, λ, the guided light

propagates as a running wave in certain patterns of the electro-magnetic field, each

pattern with its own specific velocity; one speaks about guided modes. The number of

guided modes which can propagate through the waveguide depends on the values of the

refractive indices, the geometry of the channel structure and the wavelength used.

Structures through which one mode can propagate at only a given wavelength are easily

designed and realized. These monomodal waveguides are required by most IO sensing

principles. When lowering the cross sectional dimensions or the refractive index of the

core a state can be reached such that at lower values of these parameters no guided

mode can be supported anymore. This state is called the cut-off.

Generally, the electro-magnetic field of the modes consists of three electrical and

three magnetic field components. If one (for slab modes) or two (for channel modes) of

them are known Maxwell theory delivers some simple relations for calculating from the

known components all other ones. From Maxwell theory it can also be derived that in a

good approximation for most waveguide types the guided modes can be classified into

two groups: TE-modes and TM-modes, where the electrical, respectively the magnetic,

field has a dominant component along the lateral direction only as shown in figure (1-9)

[5]. Three of the field components are zero leaving for TE-modes Hy and Hz and for

TM-modes Ex and Ez only. Generally, the component of the field along the propagation

direction (defined as the z-axis) is relatively small. The approximation is exact for slab

modes and is very good for the commonly used rib-type channel modes.

From Maxwell theory it can be shown that for a given vacuum wavelength of the

light, λ0, the ith TE mode can be mathematically described by:

()()

(

)

[

]

tzkNjyxEtzyxE eff

i

y

i

y

i

ωλλλ

−= 0000 exp,,,,,, (1-1)

Figure (1-9): The propagation of light as a running wave of the electro-magnetic field

[5]

1 Processing of waveguide sensors 13

The formula shows a harmonic wave running in the Z-direction, with an angular

frequency ω. Here k0 = 2Л / λ0 = ω / c, where c is the vacuum velocity of light. Eiy (x, y,

λ0) is called the ith field profile of TE-mode. Note that not all optical power propagates

through the core layer, but that parts of it propagate through the outer layers (the

cladding layers). In these layers the field strength decays exponentially with the distance

to the boundary with a decay length (1/e value) in medium i with refractive index ni

being:

22 22

0

1

eff i eff i

c

kN n N n

ω

=

−−

(1-2)

For most systems this decay length is on the order of magnitude of 0.1–1 µm.

These fields outside the core layer are called the evanescent fields of the mode. These

evanescent fields play an essential role in most IO sensors.

The propagation velocity of the mode is equal to c / Neff, where Neff, the so-called

effective refractive index, being limited to the range ncore > Neff > (nc, nb), has taken the

place of the refractive index of a homogeneous medium, through which a planar

electromagnetic wave propagates. Modes are numbered according to their Neff-values,

the highest Neff corresponding to the lowest mode number. In the case that the structure

is lossy (e.g. due to some absorption or scattering) the losses are expressed by adding an

imaginary component to Neff.

The number of guided modes that can propagate through the waveguide, their

field profiles and their Neff-values depend on the refractive index distribution (meaning

the values of the refractive indices of all relevant materials and the geometry of the

cross section of the straight waveguide) and the wavelength used. Structures are

possible through which for a given wavelength one TE mode (and often one TM mode)

can propagate only: a monomodal waveguide.

1.5.1 Light coupling into waveguides

There are various ways of coupling light from the outside world into a guided

mode of the chip (excitation of a guided mode): selective excitation using prisms [6] as

shown in figure (1-10) or gratings [7] as shown in figure (1-11) for coupling the power

of collimated free space beams into a specific waveguide mode or butt-end coupling

using fibers [8] or (especially in the laboratory) free space beams focused on the

entrance plane of the waveguide.

The first two methods are also utilized for the sensing itself. Because of

reciprocity of light all these methods can also be used for coupling light out of the chip.

1 Processing of waveguide sensors

14

Figure (1-10): Prism coupler

Figure (1-11): Grating coupler

Substrate

Buffer

Film

Air gap

Substrate

Waveguide

g

r

d

1 Processing of waveguide sensors 15

1.5.2 Optical losses

The most important characteristic with the modal properties of a waveguide is its

attenuation or loss, that a guided mode experiences as it travels through the waveguide.

The loss arises mainly due to intrinsic material properties or imperfections that may

come during the fabrication processes. A common measure of the optical loss is dB/cm

and it is defined as the logarithm of the power (intensity) ratio as:

()

10log

/

in

out

P

Loss dB cm l

= (1-3)

Where l, Pout and Pin are the waveguide length, output and input power,

respectively. The measured total loss is the sum of the individual losses due to different

mechanisms.

A mode traveling in the waveguide experiences a scattering loss, in case of facing

imperfections with index fluctuations of the order of the wavelength of light. According

to the location of occurrence, the scattering losses in a waveguide can be divided into

two as volume and surface scattering. In general, volume scattering results from

imperfections such as voids or particle contaminations. The loss due to them is

proportional to the total number of scattering centers per unit length and is negligible

compared to surface scattering loss.

The surface scattering loss might be of much more importance since the

propagating wave interacts strongly with the surfaces that are more imperfect in terms

of roughness. Namely, as the ridge waveguide is defined by etching, the surface may

experience damage, in addition to which the side walls of the rib itself can be rather

rough. In order to avoid these complications, a careful optimization of the reactive ion

etching (RIE) process applied for the waveguides, has to be worked out.

Although both types of scattering occur at different locations, they result from the

same physical process of scattering of light. The power radiated as a result of this

mechanism is given by the well-known expression of Rayleigh scattering

() ()

4

0

2

3

0

2

4

3

4

12

λε

επ

πε

εω

Ec

c

E

Prad

Δ

=

Δ

= (1-4)

where

ε

Δ is the excess polarizability and E is the electric field component of light. The

important feature of the equation is that the radiated power is inversely proportional to

the fourth power of the wavelength.

1 Processing of waveguide sensors

16

Another type of energy loss in waveguides is in the form of the loss of the optical

power from the mode into the media surrounding the waveguide. It can be observed in

both slab and ridge waveguides in terms of leakage of the mode into the substrate or air.

In fact, scattering losses may be included in this category, too. Radiation losses can also

take place in the waveguides having sharp bend or curvature, but this is not an issue for

our case. The SiO2 lower cladding thickness of 2

μ

m ensures no radiation leakage in

the substrate [41].

1.6 Chemical sensors

1.6.1 Background

Sensors technology has been identified as one of the most important technologies

for the 21st century. Sensors are important components in quality control as well as for

online control of different processes in industry.

Sensors are usually designed to monitor one thing at a time and are generally

devices which transduce a physical or chemical parameter into an electrical or optical

signal. Physical parameters which are commonly monitored with sensors are

temperature, pressure, force, magnetic field, etc, while chemical parameters of interest

most often are the concentrations of chemical substances.

A sensor can be described by the:

- detection mechanism

- parameters it is sensitive for, selectivity

- concentration range, sensitivity

- speed of response

- operation temperature

- life time

Physical sensors are different than chemical sensors. Physical sensors are

normally influenced by one or very few parameters only but chemical sensors might be

influenced by hundreds of parameters.

Sensor technology in general is multidisciplinary and this is particularly true for

chemical sensors, simply because knowledge in physics and electronics has to be

combined with knowledge in disciplines such as electrochemistry, biochemistry, etc.

This is an inhibiting factor in the development of chemical sensors but more and more

1 Processing of waveguide sensors 17

scientists are now crossing discipline boundaries and new multidisciplinary research

groups are being set up.

Chemical sensors usually consist of a sensitive layer or coating and a transducer

[9], [10]. Upon interaction with a chemical species (absorption, chemical reaction,

charge transfer, etc.), the physicochemical properties of the coating, such as its mass,

volume, optical properties, or resistance, etc. reversibly change. These changes in the

sensitive layer are detected by the respective transducer and translated into an electrical

signal such as frequency, current, or voltage, which is then read out and subjected to

further data treatment and processing.

Various inorganic materials serve as chemically sensitive layers that

can be coated onto the different transducers [11]. The different sensitive materials and

their operation conditions, such as elevated temperature, impose certain requirements on

the transducer design [12].

The large amount of parameters that influence the sensor signal of a chemical

sensor gives a lot of effects that has to be taken into account when using chemical

sensors:

- Chemical sensors are normally non-specific sensors, there are very few 100

% selective chemical sensors

- Cross-sensitivity: some chemicals may interact to give a different signal

from the component in a mixture compared to the single component

- Temperature sensitivity, chemical reactions, adsorption, desorption, and

diffusion processes have a large temperature dependence, and temperature

control is crucial

- Memory effects: the history of the sensor influences the response to a certain

chemical component, so frequent recalibration is necessary or advanced

signal processing

- Drift problems related to the sensor technology in use, e.g. MOSFET

sensors, SAW sensors or conducting polymers may have electronic drift not

related to chemical interaction

- Other drift problems, that is, chemical sensors in general suffer from long

term stability problems

Hence, there are some special measurement techniques used for chemical sensors:

- Twin sensors are used to get rid of drift problems not caused by the chemical

species to be measured

- Recalibration normally has to be used frequently. Special calibration

algorithms for chemical sensors are being developed

- Regeneration of the sensor by heating might reduce memory effects

- Short test gas pulses might reduce memory effects

- Scanning of the temperature enhances selectivity

1 Processing of waveguide sensors

18

- Large areas of catalytic metal influences selectivity

- Advanced signal processing enhances the sensitivity

- Sample handling and sample treatment, e.g. to keep a constant temperature

and humidity of the samples

The response of a sensor to a certain chemical component can be measured in

different ways [13]:

- The difference in sensor signal between a reference and the test species

- The ratio between the sensor signal at a reference related to the test species

- The derivative of the initial change of the signal due to a chemical species

- The integral of the change of the signal due to a chemical species

Semiconductors play an important role in sensor technology because of their

variable conductivity. The concentration as well as the mobility of charge carriers in a

semiconductor is sensitive not only to physical parameters such as temperature, light

and mechanical tension but also to chemical parameters.

Integrated optical (IO) devices are increasingly used as transducers for

optochemical sensing applications [14], [15]. The utilization of microtechnology for the

integration of these devices offer some advantages as a better control of the light path

by the use of optical waveguides, high sensitivity, mechanical stability, miniaturization,

and the possibility of mass production [16]. The use of standard silicon microelectronics

technology for the fabrication of these devices allows a high homogeneity of the

waveguide material and the possibility of final integration of optical and electrical

functions on the same chip [17].

In general, integrated optical sensors make use of the evanescent field detection

principle. In an optical waveguide, confined light travels within the core layer.

However, part of the guided mode (evanescent field) travels through a region that

extends outward into the media surrounding the waveguide. When there is a change in

the optical characteristics of the outer medium (i.e., refractive index change), a

modification of the optical properties of the guided mode is induced via the evanescent

field. To detect this variation, Mach-Zehnder Interferometer (MZI) can be used. The

refractive index change is evaluated by the intensity modulation produced by the

interference of light traveling through the two branches of the interferometer, one of

which is exposed to the refractive index variations of the outer medium.

To obtain an efficient transducer for detecting molecular interactions, the optical

waveguides of the sensor should have a high surface sensitivity. The strength and

distribution of the evanescent field in the outer medium need to be maximized in order

to assure a high response for changes in the optical properties of the surrounding

medium.

1 Processing of waveguide sensors 19

For sensing applications, the optical waveguides that form the MZI sensor must

have two features: monomode behavior and a high surface sensitivity. Optical

waveguides based on Total Internal Reflection (TIR) have been experimentally proved

to have high surface sensitivity [18]. For implementation in a MZI, these waveguides

should have monomode behavior that, assuming a fixed wavelength, can be controlled

with a proper design of the thicknesses and refractive indices of the waveguide layers.

Finally, a rib structure is designed to achieve lateral light confinement. Width and depth

of this rib are also essential for assuring a single-mode optical waveguide.

There has been extensive research reported using optical fibers and planar

waveguides for chemical sensors and optical guided-wave chemical sensing is an

extremely promising and fast growing technology. However, planar waveguide devices

utilizing modern microfabrication technologies facilitate monolithic integration of

several components in a single chip for simultaneous multiagent detection. These

compact integrated optical devices are robust, give much flexibility in material selection

to optimize the sensitivity and have the potential for low-cost manufacturing [19].

Since so many different sensors are found one can ask: which one is the best?

There are so many different sensing situations and criteria to be considered that the

general answer is: the sensor which will do the job. There is, however, one

characteristic of a chemical sensor which sets it apart from the others and that is

ruggedness. This term describes the ability of the device to maintain its performance

specifications even under adverse operating conditions. In a pragmatic sense it is

understood as reliability.

Ruggedness may have different meanings; a mechanically rugged device is able to

withstand mechanical shock, vibrations, mechanical stresses, and so on. Chemical

ruggedness has a slightly more subtitle meaning. It is related to selectivity and means

that the output of the sensor is unaffected by unforeseen chemical changes in the

operating environment. Generally speaking, rugged sensors are those commercially

most successful because they are reliable or because they are the only ones which can

provide information under conditions in which other sensors would be unable to

operate.

It is generally possible to distinguish two types of interaction of the chemical

species with the sensors: a surface interaction in which the species of interest is

adsorbed at the surface, and a bulk interaction in which the species of interest is

absorbed and partitions between the sample phase and the sensor.

1.6.2 Historical Perspective of chemical sensors

Classical solutions to chemical sensing tasks have been dominated by complex,

expensive laboratory methods such as gas chromatography and ion-mobility

1 Processing of waveguide sensors

20

spectroscopy. Although these methods are accurate for detecting chemical

concentrations and in discriminating among chemicals, their cost is often prohibitive for

many low-end, chemical sensing applications ranging from residential sensing of toxic

chemicals to the detection of seafood freshness and breath alcohol analysis. To lower

the cost of chemical sensing systems sufficiently to compete in these low-cost markets,

a new approach to chemical sensing needs to be adopted.

In order to address the needs for chemical sensing systems in consumer and other

low-end markets, it has been useful to use miniaturization techniques that perhaps

sacrifice some of the accuracy of laboratory methods for lower cost, faster response

times and greater accessibility. Since the early 1970’s, the microelectronic chemical

sensor has been investigated as this low-cost, miniaturized alternative to laboratory

chemical sensing methods. However, the miniature chemical sensor has been plagued

by problems with:

- Reproducibility: inconsistent responses to the same chemical over time

- Selectivity: difficulty in discriminating among chemicals

- Sensitivity: difficulty in detecting low concentrations of particular

chemicals

- Stability: difficulty in detecting chemicals of interest across changes in

ambient conditions

- Response time: typically on the order of tens of seconds to minutes

Chemical sensing systems that overcome some or all of these problems have had

difficulty in keeping system cost down at a manageable level for the corresponding

market for particular sensing applications. However, some progress has been made in

the research community since the 1970’s in addressing this delicate balance between

cost and robustness of viable chemical sensing systems. There are three general tasks of

interest in these systems: concentration detection, chemical discrimination, and response

time optimization. Most microelectronic chemical sensors are able to detect

concentrations reasonably well at medium to high concentrations.

Since the 1970’s, the microelectronic chemical sensor has been explored as a low-

cost alternative to laboratory chemical sensing methods. Many of the microelectronic

sensor technologies are based simply on conductivity changes in a material in response

to chemicals in the environment. The simplest of these conductivity-based sensors, the

thin-film sensor, was first introduced into the research community in the early 1970’s.

As shown in figure (1-12) the thin-film sensor is simply a film of chemically sensitive

material, such as tin oxide [20] or polypyrrole [21] whose conductivity changes in

response to reducing chemicals in the sensing environment. The thinness of the film is

required, because these conductivity changes are primarily based on surface interactions

and the surface must be a significant part of the entire sensor in order to detect these

changes. The output of the sensor may be read either as a current or as a voltage.

1 Processing of waveguide sensors 21

Vsensor

+ _

isensor

Figure (1-12): Basic structure of the thin-film sensor

The metal-oxide, thin-film sensors are the only miniaturized chemical sensors that

have had significant impact in commercial markets. For example, tin oxide (SnO2) and

iron oxide (Fe2O3) [22] have frequently been used to detect hydrocarbons and

combustible gases in a variety of applications. By far, the most popular of these sensors

has been the Taguchi-type sensor, manufactured by Figaro Engineering in Japan; these

sensors are made up of primarily tin oxide modified with various catalysts and additives

to detect particular hazardous gases such as carbon monoxide and methane [23].

Perhaps the most important promise of these thin-film sensors for the development

of viable chemical microsystems is their compatibility with standard integrated circuit

fabrication processes. Metal oxides and conducting polymers can be deposited onto

standard integrated circuit substrates, often after circuits have already been fabricated on

the same substrate.

Also, in the early 1970’s, the ion sensitive field effect transistor (ISFET) was

developed in the research community. As shown in figure (1-13) an ISFET is simply a

MOSFET without a gate. The oxide layer of the FET is replaced with an insulating,

chemically sensitive membrane. Charges from sensitive chemicals accumulate on top of

this insulating membrane and are amplified through the operation of the FET [24].

Although the amplification properties of the transistor in these devices seem very

attractive for sensing chemicals, the vulnerability of the insulating membrane to

environmental poisoning and subsequent transistor breakdown has prevented the ISFET

from gaining popularity in commercial markets.

1 Processing of waveguide sensors

22

Chemically sensitive insulator

Drain

Source

Figure (1-13): ISFET (ion selective FET)

Since the insulator layer provides no optical shielding from the surrounding FET

device, light sensitivity has also proven to be a problem with these devices. As a result,

the ChemFET has demonstrated more potential for integration into practical chemical

sensing applications although it is less selective and chemically sensitive than the

ISFET as shown in figure (1-14). Unlike the ISFET, the ChemFET uses a standard

oxide layer as the insulator and a chemically sensitive metal, such as palladium, as the

gate [25]. The addition of the gate minimizes light sensitivity problems that are a

problem in ISFETs. Likewise, potential poisoning of the oxide layer is minimized not

only by the inherent physical barrier provided by the metal gate but also by the fact that

the silicon dioxide is fairly resilient to environmental poisoning.

Modifications and hybrids of the ChemFET and the ISFET such as the surface

accessible FET or SAFET [26] [see figure (1-15)] and the Suspended Gate FET or

SGFET [27] [see figure (1-16)] were also introduced into the research community in the

1970’s; despite the enhanced selectivity and sensitivity of these devices over the

ChemFET, however, they share the common flaw of a short lifetime due to the

accelerated degradation of the partially or completely exposed oxide layer. Because of

its relative low sensitivity to environmental degradation, the ChemFET is thought to be

the most promising of the MOSFET-based chemical sensors in spite of its relatively low

selectivity compared to other chemically sensitive FETS.

The ChemFET sensors are also well suited to monolithic integration onto standard

integrated circuit substrates. Although fabrication of these devices on standard

substrates is more complex and slightly less advanced than that of the thin film devices,

the technology for integration is nevertheless currently available in the research

community.

1 Processing of waveguide sensors 23

Chemically sensitive gate

Source

Oxide

Drain

Figure (1-14): ChemFET Structure

Chemically sensitive insulator

Figure (1-15): SAFET structure

Drain

Source

Standard MOS gate

1 Processing of waveguide sensors

24

Chemically sensitive mesh

Drain

Source

Standard

MOS gate

Chemically

sensitive

insulator

Figure (1-16): SGFET structure

The MOS-based and conductivity based devices described above belong to the

largest class of microelectronic chemical sensors: those based on a single stage of

transduction between chemical input and electrical output signal. Significant

improvements in the sensor response times are attained as a result of this single level of

transduction. Furthermore, the cost of implementing and manufacturing systems that

use single transduction stage devices for chemical sensing is minimized by the fact that

standard microfabrication techniques are frequently sufficient to produce many of the

MOS-based and conductivity based sensors.

Although the ChemFET and thin-film sensors are undoubtedly the most popular of

the single transduction-stage chemical sensors, a variety of other chemical sensors have

also been explored in the last three decades for accomplishing chemical analysis tasks.

Chemically sensitive MOS capacitors are similar to ChemFETs; however, since they

lack the source and drain of the Chem-FET, their capacitive output is difficult to capture

and to process for chemical analysis. Chemically sensitive Schottky barrier diodes

contain a chemically sensitive metal as the top layer of the diode [28]; the diode barrier

height alters in the presence of a reducing chemical; however, as in the case of the MOS

capacitor, this change in barrier height is difficult to measure in a reliable and

reproducible manner. The solid electrolyte has also shown potential for improved

performance over ChemFET and thin-film sensors, since many of the available

electrolytes that are chemically sensitive have an ionic sensitivity that is highly selective

to particular chemicals [29], [30]. However, fabrication difficulties involved in bulk

micromachining and establishing a reference electrode for these devices have prevented

most electrolyte-based devices from becoming commercially viable [31].

1 Processing of waveguide sensors 25

Multiple stage transduction devices have demonstrated use for chemical sensing

applications. The thermal sensor uses a thermistor to measure the heat generated by

reactions between particular chemicals and the chemically selective layer coating the

temperature sensors. However, the low efficiency of these devices in capturing reaction

heat limits their suitability for the low concentration limits required for many chemical

sensing applications. The mass sensors, on the other hand, suffer from the opposite

problem. Based on sensing the added mass of a chemical reacting with the sensor

surface, these devices are extremely sensitive, making them also very noisy and

inherently vulnerable to interference. The piezoelectric sensors which sense the added

force of an additional mass on the sensor surface have very broad selectivity. The

surface acoustic wave SAW devices have improved selectivity over their piezoelectric

counterparts and can often be made using standard IC fabrication techniques. In these

sensors, a chemical reacts with a chemically selective layer on the sensor surface,

causing a frequency, phase, or amplitude shift in the acoustic wave traveling across the

device. Finally, the optical sensor has two advantages over all other types of chemical

sensors. These sensors are selective and the sensing environment is not required to

interact physically with the sensor. This characteristic makes optical sensors well suited

to remote sensing and applications where electricity in the sensing environment can be

hazardous.

1.6.3 Applications of chemical sensors

The applications for chemical sensors are innumerable. Chemical analysis is

presently carried out, or should be carried out, in almost all areas of technology.

Obvious is the use of chemical sensors in laboratories, in medical clinics, industrial

processes and bioreactors but there is, most certainly, a number of other places where

these devices can find applications [32]. Conceivable applications might be for food

freshness monitoring in shops, field analysis of plants and soil, portable air pollution

monitors and many more.

In the medical area there is a pronounced need for chemical sensors. Medical

diagnosis makes significant use of chemical analysis and incredible numbers of tests are

carried out in clinical laboratories. The costs for these activities are very high and there

is a significant time delay involved.

New sensors that enable rapid analysis will significantly reduce the costs and will

also enable appropriate treatment to be set in at an earlier stage. Sensors which are

cheap and easy to use will also facilitate decentralization of the chemical analysis to the

hospital wards and to the general practices. Even sensors which can be used at home

e.g. by chronically ill patients can be conceived.

Another important opportunity for chemical sensors is continuous monitoring of

chemical parameters during intensive care. This is an area which is subject to a lot of

research. Equipment for monitoring blood gases such as oxygen and carbon dioxide are

1 Processing of waveguide sensors

26

used today. Sensors for these as well as sensors for monitoring of potassium, calcium

etc. are at present being developed further, especially with respect to miniaturization

and biocompatible encapsulation. Silicon technology has opened up new possibilities

for miniaturization but the delicacy and sensitivity to moisture and ionic contaminations

of silicon devices emphasis the importance of good encapsulation.

In industrial processes there are a number of chemical parameters which should be

measured in order to optimize the processes both with respect to efficiency and quality.

Looking closer into different industrial processes one finds that surprisingly few

parameters are actually measured. With some exceptions the few analyses which are

undertaken are done so by bringing samples to a chemical laboratory. This is not only

expensive but also includes a time delay which quite often makes the information

useless for process control.

To have the ability of using modern computer technology for efficient process

control, continuous on line measurements have to be done by means of chemical

sensors. The reason why this is done to such a small extent today is not only the lack of

sensors for many substances but also the fact that existing chemical sensors are often

not sensitive, selective and stable enough for the use in industrial processes.

Factors that limit the usability of existing sensors are things like temperature and

pressure, interference from other chemical parameters, fouling and mechanical shock.

Many existing sensors would certainly be useful if they were further developed to

withstand such conditions. The main reason why this is not done is that the demands are

very diversified and the sensor needs to be designed specifically for each application.

This is expensive and the market for each version of a sensor is too small to cover the

development costs. A good sensor for process control purposes can be sold at very high

prices but very few sensor manufacturers are prepared to take the economical risks

involved. This means that the potential users of process control sensors have to

undertake, or at least pay for, the development of sensors in their particular applications.

This fact is slowly being accepted since improved process efficiency and quality are

often the most powerful means for competition and improvements in process control

can therefore be highly valuable in economic terms.

Environmental monitoring is another field of applications for chemical sensors in

general and gas sensors in particular. In more and more countries people are becoming

aware of environmental hazards in their lives. Long-term exposures to even trace

amounts of hazardous gases have been shown to have severe medical implications.

Health authorities have a demand for portable instruments for monitoring and the

industry is interested in alarm systems for protection against elevated levels of toxic and

explosive gases.

The most difficult task is to develop gas sensors with sufficiently high sensitivity

and selectivity to measure concentrations at the very low exposure limits. This can be

done, and has been done, for some of very reactive gases but is more difficult for less

1 Processing of waveguide sensors 27

reactive, but still toxic gases. The need for sensors for environmental monitoring is

appreciated amongst health authorities but the markets for sensors for toxic gases are

generally too small to attract commercial interest. The development of these sensors is

therefore strongly dependent on governmental funding. Sensors for explosive gases

have been shown to have much larger markets and have consequently attracted more

commercial interest. Research on sensors for gas alarms is carried out by a large number

of companies.

It is also worth bearing in mind that chemical sensors can be used, and are used, to

measure nonchemical parameters. A well known example of this is leak detection by

means of tracer gases. Other indirect measurements such as flow measurement using

traces can be conceived and new sensors will probably give rise to such new

possibilities.

1.7 Optical waveguide based chemical sensors

Sensors that can translate the concentration of various chemical compounds to the

electrical domain are in great demand by industry. These sensors are required in process

control, environmental monitoring, health care, biotechnology and the automotive

industry for example. In some cases it is advantageous to use the optical domain as an

intermediate between the chemical and electrical one. This can be useful for several

reasons, amongst them are:

• The absence of electromagnetic interference

• No danger for explosions due to electrical passivity

• High sensitivity to many measurands

• Specific optical sensing mechanisms, such as chemical analysis by means of

material specific absorption wavelengths

Many characteristics of light are available to carry the sensor information. For

example the phase, the intensity (distribution), the direction of propagation, the

polarization or the wavelength can be used. In many situations it is preferred to

incorporate optical sensors in a waveguide system, either fiber or integrated optics

based. Compared to the use of free space beams this guided wave approach offers the

potential advantages of:

• a better control of the light path, no alignment of the experimental setup is

required

• evanescent field sensing, enabling the use of thin chemo-optical interfaces since

large interaction lengths can be used

• the possibility of remote sensing by the use of a fiber network

• a higher mechanical stability

• a reduced size, weight and price

1 Processing of waveguide sensors

28

Waveguide based sensors compared to integrated optics are very cheap and have

the strong advantage that they can be easily applied for distributed sensing. The sensor

can be incorporated into the waveguide that transports the light from the source to the

detector and the different sensors can be read out. Nevertheless, waveguide sensors can

not compete with integrated optics with respect to:

• Robustness

• Compact optical circuitry, enabling a higher complexity (e.g. multiple sensors

on one chip)

• Design flexibility with respect to the geometry as well as the choice and

combination of materials (for example active and passive materials)

• Ease of access to the optical path in evanescent-field sensing

• Potential of integration with micro-electronics, micro-mechanics and micro total

analysis systems

• Potential of ‘cheap’ batch-wise mass production

• Benefits from the developments in guided wave devices and microsystems in

optical telecommunication

The degree of integration of a sensing scheme depends entirely on the intended

application. Selection criteria will for example arise from the required sensitivity,

selectivity, sensing environment, size or price. On one hand very sensitive sensing

systems are known where only the sensing element is a planar waveguide and where the

light from a bulk laser is coupled into and out off the waveguide by means of prism-,

grating- or end-fire coupling. This type of devices will be very suitable for applications

in e.g. medical laboratories. On the other hand, channel waveguide based sensors where

the source, a phase modulator and a detector are monolithically integrated on the chip

have been reported. Due to their small size and weight these sensors might be especially

suitable for applications in e.g. space or micro total analysis systems. These two

examples illustrate that there is no ultimate choice for using end-fire, butt-end, grating-,

prism-, hybrid- or monolithically-coupled light sources or detectors, for the use of

channel or slab waveguides, or for on- or off-chip modulation, referencing and signal

processing [6]–[8].

1.7.1 Refractive chemical sensors

Refractive sensing is very important because of several reasons:

• Association of a limited number of receptor–target pairs only will accomplish

useful changes of absorptive and luminescent properties while the refractive

index will always change

• Most physically interesting sensor principles are of the refractive type

1 Processing of waveguide sensors 29

• Measuring absorption changes by an IO sensor does not differ in essence from

that of the corresponding bulky methods [33]. Also, they are less sensitive and

generally show a poor selectivity

• In luminescent sensors the luminescent properties of the sensitive material are

sensitive to the analyte. Up to now in luminescent sensors the IO system has

been used only for propagation of a mode of a wavelength suited to excite the

luminescent particles [34]; the number of luminescent particles is derived from

monitoring the free space emission. The low collection efficiency of a guided

mode and the lack of efficient IO filters to effectively split the weak emitted

mode from the intense excitation mode hamper a complete IO luminescence

based sensor. Note that as a consequence of the fact that the exciting mode

propagates parallel to the thin luminescent layer, the excitation power is used

much more effectively than when using a perpendicularly incident free space

excitation beam. Although not treated here in full detail, luminescence sensors

are often applied because of their extremely high sensitivity.

The great majority of the refractive sensors rely on the induced changes of the

effective refractive index Neff. In order to get large effects the wave guiding structure in

the sensing region has to be optimized to maximum partial sensitivity ∂Neff/∂nsensitive layer.

Various papers [35], [36] have been published on how to do so, resulting e.g. in maps in

which all influences on the sensitivity can be represented by a pair of normalized

parameters only. Sensitivities not only depend on the geometry, but also on the choice

of materials. Out of the three technologies mentioned earlier the SiON technology can

result in sensitivities close to 0.25, while the sensitivities of in-diffused glasses are an

order of magnitude lower, mainly as a consequence of the low index contrast between

the core and the substrate layer. However, by applying in the sensing region thin high

index layers on top of the structures much higher sensitivities can be obtained.

Sensitivities of polymer IO waveguides are in between. In special structures and special

free standing waveguides ∂Neff/∂nsensitive layer values somewhat larger than unity can be

obtained [37].

In chemical sensing the cross effect arises from the in-diffusion and the

subsequent capture of the target units (molecules or ions) at receptor sites in a so-called

chemo-optical material. This capture-based sensing is generally called affinity sensing.

Essentially during the capture of target molecules by receptor molecules a small volume

of gas or solution is replaced by the target molecules, while as a consequence of their

chemical bond, also their electron distribution will be changed. Both effects together

always result in a change of the refractive index; for a limited number of specific

receptor–analyte pairs the association also accomplishes a useful change of the

absorption in the visible / near infrared or of the luminescent properties. The larger the

concentration of accessible receptors in the sensitive layer, the larger are the optical

changes which can be obtained.

Generally, this chemically sensitive material is applied as a cladding layer within a

window obtained by removing locally the originally applied cladding layer. This

1 Processing of waveguide sensors

30

sensitive material is probed by the evanescent field of the mode: evanescent field

sensing. This configuration has several advantages:

Firstly, it affords in a natural way for the realization of general IO sensing

platforms which later on can be provided with a specific cladding layer which enables

the capture of the target units aimed at in the specific application. Also, if required, the

sensitive layer can be easily removed making room for applying another one.

Secondly in-diffusion especially of large target molecules can be a slow process;

hence to obtain acceptable response times the diffusion path has to be short as can be

realized by using thin chemo-optical cladding layers.

Thirdly these thin layers can be very effectively probed by the evanescent field of

a guided mode. Although only a part of the modal power experiences the change of the

optical properties of the cladding, the interaction length is large, sometimes even on the

order of magnitude of centimeters. So the effects on the light beam properties are much

larger than in the case that the thin layer were interrogated by a perpendicularly incident

free space beam, for then the interaction length would be equal to the thickness of the

sensitive layer.

Generally the capture of the target molecules by receptors is part of a chemical

equilibrium reaction and as a consequence the fraction of filled receptors will be an S-

shaped function of the logarithm of the concentration of the target molecules in the

environment as shown in figure (1-17). The concentration at which half of the receptors

are filled can be derived to be the reciprocal of the association or affinity constant, Kass,

of the target–receptor pair. Hence, large affinity constants in principle allow for

measurements in low-concentration ranges. This association constant is equal to the

ratio of the association rate and the dissociation rate. In practice (e.g. at

immunosensing) a high association constant mostly appears to imply a very low

dissociation rate and as a consequence the fraction of filled receptors can follow only an

increase of the concentration of analyte with a commonly acceptable time lag, but not a

decrease.

Also materials with physical/chemical adsorption of analyte molecules can be

applied; for example polymer layers with some porosity or gels. Here gelatin is

mentioned, a material sensitive to relative humidity, and over the 100% R.H. range the

corresponding change of the refractive index is about 0.055 [38].

When measuring the concentration of a chemical compound in a solution in the

absence of other dissolved substances: for the refractive index or absorption of the

solution is already a measure for the concentration. There are however several reasons

to use also sensitive layers in these cases. Firstly as a result of the generally high

association constants the concentration of the target molecules in the sensitive layer will

be much larger than that in the solution; hence, such a sensitive layer generally strongly

1 Processing of waveguide sensors 31

enhances the effects. Secondly use of a sensitive layer enables the use of structures from

which also some reference signals can be obtained by which e.g. the influence of

temperature changes of the solution can be strongly reduced.

If several types of target molecules are solved all with unknown concentrations, an

additional requirement of the receptors arises: they have to be selective, implying that

they can associate with the target molecules only and hence that any association with

other types of molecules is excluded. And in fact here one arrives at one of the weak

points of all chemical sensors based on association: complete selectivity can never be

reached. Nevertheless, if the nature of all types of target molecules in the solution is

known it is sometimes possible to obtain data on all concentrations. If the structures of

all other compounds differ strongly from that of the measurand, high specificity with

respect to these other compounds is feasible. Also all concentrations may be determined

by using sensor arrays, each sensor supporting a layer that shows different sensitivities

to all present compounds, while these sensitivities are different for different layers. By

application of neural networks or chemometric methods all concentrations can be

derived from the primary sensor data. In this review these aspects will not be treated

any further. For example it is left up to the chemists to develop useful receptors and

adequate methods for immobilizing them in a matrix material, for attaching this material

to the core layer of the sensor inside the sensing window and for regeneration of filled

receptors.

0

0,2

0,4

0,6

0,8

1

1,2

Log c

F

Figure (1-17): Fraction, F, of the sites which are associated with a target molecule as a

function of the concentration, c, of the target molecules (Kass is the affinity constant)

Log (1/Kass)

1 Processing of waveguide sensors

32

The response time of the sensor not only depends on the association rate but also

on the time needed for the target molecules to reach the receptors (diffusion time).

Under special conditions response times can be reduced appreciably by introducing

kinetic measurements: not the equilibrium state is registered, but here it is sufficient to

monitor the first part of the process in which the equilibrium state is reached only.

In so-called direct sensing only, the target molecules are identical to the analyte

molecules. In this operation mode in principle continuous measurements are feasible

and it is very effective for e.g. an alarm sensor. In indirect sensing mostly either analyte

or receptor molecules are applied which are provided with optical labels.

It may be clear that these indirect formats do not allow for continuous

measurements and after each measurement the receptors have to be emptied completely

by adding appropriate regenerating solutions. Another drawback is the consumption of

expensive labeled molecules, while also auxiliary facilities have to be added to the

sensing system for providing and mixing several solutions. However their sensitivity is

generally greater than obtained by direct sensing methods and both indirect formats are

often applied in applications where a concentration has to be measured only once, e.g.

in immunosensing. Most markets prefer application of rapid label-free methods [39]

provided that their resolution is sufficient for the specific application. And in fact

obtaining the high resolutions (and hence the low detection limits) required for direct

assays at acceptable response times is one of the main challenges for refractive type IO

sensors.

With respect to the thickness of the sensitive cladding layer one can define two

extremes: in the first one the sensitive layer is thick enough that the power of the

evanescent field outside the sensitive layer can be neglected; the sensor is insensitive to

the refractive index of the solution and one speaks about homogeneous or bulk sensing.

At the other extreme the sensitive cladding is very thin, in the limit being a

monomolecular layer only; in good approximation the modal field is constant over the

whole of the thickness. Association can result in both an increased average thickness of

the layer and a change of its refractive index. In practice both effects can be taken into

account simultaneously by approaching the effects as arising from a change of the

effective thickness of the layer only. In surface sensing the Neff-value is also sensitive to

the refractive index of the solution and it will also experience influence of non-specific

adsorption of other substances on the sensitive layer. Application of reference windows

can (sometimes partly) compensate for these disturbing effects.

Optical environment sensors have recently become a fast-expanding technology

where the change of the phase shift is converted into a variation of intensity. The main

focus of this research is to demonstrate a chemical sensor, particularly as a sensor of

gaseous species such as ammonia.

1 Processing of waveguide sensors 33

1.8 Ammonia sensors

Ammonia is widely used in the production of explosives, fertilizers and as an

industrial coolant. The toxic qualities of this gas are well documented, and acute

poisoning can result from inhalation of only small doses of ammonia vapor. Exposure

limits of 25 ppm over an 8 h period and 35 ppm over a 10 min period have been

recommended and have recently been legislated for.

Locations employing industrially sized cooling systems, such as food production

and storage plants, report a number of exposure incidents every year. As industry is

becoming increasingly safety conscious, efficient sensor devices to monitor personal

exposure of workers undertaking any risk of contact with ammonia are desirable, as

rapid evacuation of personnel from contaminated areas may be all that is required to

prevent serious illness.

In the field of air quality monitoring, a wide variety of instrumentation can be

employed for ammonia analysis. Devices employing spectroscopic or electrochemical

methods are usually very accurate, sensitive and selective. However, they are also

expensive, static instruments and requiring the presence of an experienced operator.

Simpler detector systems based on the semiconductivity of SnO2 thin films do exist for

monitoring ammonia leakage in the working environment, and have been

commercialized as personal monitoring systems. However, these devices come at a

much higher cost, are not selective towards ammonia and display a restricted active

sensor lifetime.

Figure (1-18): Evanescent field penetrates the cover layer

Recently a miniaturized electrochemical sensor device has become available for

ammonia detection. This device uses interchangeable sensor heads biased towards

detection of a specific gas, with the added requirement of peripheral filters to achieve

1 Processing of waveguide sensors

34

effective selectivity towards ammonia. A different approach can be applied when

considering a dedicated ammonia sensor, which similarly takes advantage of

miniaturized, low power components and selective response to ammonia in the required

range. Optochemical transduction is a way to develop a non-consumptive, compact

ammonia sensor device with low power demand that could be used for in real-time

analysis.

A variety of optical electrode devices have been used that utilize the reaction of

dissolved ammonia vapor with a pH-dependent dye material, which undergoes a

suitable colour change [40]. In general, these are based on monitoring the absorption or

fluorescence characteristics of indicator dyes entrapped within a membrane deposited

onto a waveguide substrate. Ammonia interacts with the immobilized indicator,

resulting in changes in absorbance or emission spectra, which are monitored using a

suitable detector module via an optical fiber or planar waveguide. Evanescent wave

absorption is an effective technique for performing such analysis. Light energy

associated with the guided mode penetrates into a coating of lower refractive index than

the substrate as shown in figure (1-18). This evanescent field is able to interact with dye

materials contained within the coating. When there is a change in the optical

characteristics of the outer medium (refractive index change), a modification of the

optical properties of the guided mode is induced via the evanescent field and this can be

detected by MZI.

In the Mach-Zehnder interferometer (MZI) shown in figure (1-19), the incoming

coherent light wave is equally split into two channels, and after a certain distance of

propagation, the waves from both channels are recombined together. One of the

channels [the phase-modulating arm (sensing arm for sensing applications)] is exposed

to the outer medium for a certain distance L (interaction length), while the other channel

(the reference arm) is isolated from the medium. During this distance, the wave in the

phase-modulating arm will experience a phase shift with respect to the wave in the

reference arm. At the output port, light coming from both channels will interfere and

show a sinusoidal variation corresponding to the accumulated phase difference, which is

related to the change of the refractive index of the surrounding medium.

Figure (1-19): Mach-Zehnder interferometer

Sensin

g

are

a

φ

r

2 Theoretical considerations 35

2 Theoretical considerations

2.1 Introduction

Optical devices based on integrated waveguides are crucial for the future

development of optical communication systems. In recent years, growing attention has

been paid to silicon based dielectrics such as silicon oxides, nitrides, and oxynitrides as

potential materials for integrated optics [41].

There is also a growing need for sensitive chemical sensors in areas such as

process technology, health care, environmental control, biotechnology, etc. For many

applications these sensors should show a high resolution over a wide dynamic range,

they should be very sensitive, selective, fast, small and cheap, and should be suited for

remote sensing. An attractive option fulfilling all these requirements is offered by

integrated optical (IO) sensors. In these optical chips the chemical parameter to be

sensed, the measurand, influences the sensing region directly or by means of a chemo-

optical transduction layer the propagation properties of a guided light propagating

through the chip. Using appropriate optical circuitry, these changes of propagation

properties can be converted into a change of the optical output power. In turn, this is

measured using a photo detector and appropriate electronics.

Generally, selectivity is provided by the chemo-optical transduction layer,

alternatively called sensitive layer. This layer contains receptor units that are selectively

associated with the chemical entities of the measurand. In addition, this layer can

concentrate the measurand molecules and enhance the optical effects of their presence.

Mostly the relevant changes of optical parameters occur in a region just outside

the core layer of the waveguide, the evanescent field region. Hence, such chemical IO

sensors are often called evanescent field sensors. The changing optical parameters are

the refractive index (n), the absorption coefficient and luminescence parameters.

The interferometric sensors are generally supposed to have the highest sensitivity

potential. This is mainly based on the possibility of using large interaction length values

resulting in an enhanced sensitivity. Amongst the interferometric sensors it is only the

MZI sensor that contains an easily accessible reference arm. When used properly, this

reference arm can make the sensor (nearly) insensitive to many perturbing effects. The

large intrinsic stability results in an improved sensor resolution, making the MZI sensor

very attractive. The next task is to design an optical waveguide based chemical sensor

and a heater which will be used to increase the sensitivity of the sensor.

2 Theoretical considerations

36

2.2 Waveguide design

Chemical sensors are the non-communication application field where integrated

optic technology is expected to play an increasing role and where it is already

successful commercially. The sensing in the waveguide is performed by the evanescent

tail of the modal field in the cover medium. This sensing operation consists of

measuring the change of the effective index of a propagating mode when a change of

the refractive index takes place in the waveguide cover. The waveguide characteristic

equation or/and a calibration allows the retrieval of the index change from the measured

change of the effective index. The sensitivity of the measurement of the physical or

chemical quantity present in the cover depends on the strength and the distribution of

the evanescent field in the cover. The main design task is therefore to find the

monomode waveguide structure which maximizes the sensitivity on the quantity to be

measured [42].

In the case of the rib waveguide, the geometrical parameters of the waveguide

section shown in figure (2-1) are chosen in order to fulfill the monomode propagation

condition. This can be achieved if the geometry of the waveguide fulfils the following

relationship using the effective index method (EIM) [43].

2

r

< c +

1 - r

t (2-1)

where:

eff

w

tH

=, eff

eff

h

r

H

=,

q

,

eff

hh=+ eff

H

Hq

=

+,

2

2c

eff 2

f

c

B

ww

kn n

=+ −, (2-2)

22 2

cs

2

f

cf

BB

qkn n kn n

=+

−s

−

, (2-3)

2 Theoretical considerations 37

f

n ,

s

n and are refractive indices of the guiding region, the substrate and the cover,

respectively.

c

n

,cs

B

= 1 for TE modes and

2

,cs

f

n

⎛⎞

⎜⎟

⎝⎠

for TM modes, k = 2

π

λ

,

λ

is the wavelength. The

factor c can be either 0 or 0.3.

w

H

h

Figure (2-1): Schematic of a rib waveguide

The results of the calculations using core refractive index 1.55 are given

graphically in figure (2-2). In the region below the line the structure is monomode. It is

evident that according to the theory, the waveguide width is more critical parameter

than the rib height.

At present, the beam propagation method (BPM) is the most widely used for the

study of light propagation in waveguides and therefore it is used to calculate the

thickness of the different layers. Nowadays, there are a great number of versions of

BPM [44]-[48]; BPM based on the finite difference method (FD-BPM), BPM based on

the fast Fourier transform (FFT-BPM), BPM based on the McKee-Mitchell Scheme and

BPM based on the finite element method (FE-BPM). Especially, FE-BPM is superior to

FFT-BPM in the sense that the former can be applied to strongly guiding waveguides

and strongly polarization dependent waveguides. In addition, FE-BPM can arbitrarily

select the order and the number of elements, depending on the required computational

accuracy.

Recently, the so-called imaginary distance beam propagation method (ID-BPM)

[49] has been reported as an analysis method of eigenmodes which in it the propagation

direction is selected along the imaginary axis, and selecting the appropriate propagation

step size, we can extract the specific eigenmode from the initial input field expressed by

arbitrarily superposing the eigenmodes. The main advantages of the ID-BPM as an

eigenmode solver are as follows:

- High-efficient calculation algorithms developed for the BPM analysis can

be directly utilized

2 Theoretical considerations

38

- Matrices derived from the BPM formulation are essentially complex, hence,

lossy optical waveguides can be easily treated with no additional effort

- Eigenmodes can be obtained successively from the fundamental to higher

order modes

- Employing the appropriate boundary conditions, not only guided modes but

also leaky modes may be treated

0

1

2

3

4

5

6

7

0,4 0,6 0,8 1 1,2 1,4 1,6 1,8

Rib height (μm)

Width (μm)

Multimode

Monomode

Figure (2-2): Monomode condition for SiON waveguide

Here, in order to treat leaky modes, the perfectly matched layer (PML) boundary

condition, the validity of which has been already confirmed in the real distance beam

propagation method, are employed as boundary conditions for artificial boundaries.

A 3-D optical waveguide surrounded by PML regions I, II, and III with thickness

as shown in figure (2-3) is considered, where x and y are the transverse directions, z is

the propagation direction, PML regions I and II are faced with the x and y directions,

respectively, region III corresponds to the four corners, and Wx and Wy are the

computational window sizes along the x and y directions, respectively.

2 Theoretical considerations 39

Using the transversely scaled version of PML [49], Maxwell’s equations can be

written as:

2

0

Hj nsE

ωξ

∇× = , (2-4)

and

0

E

jsH

ω

μ

∇× =− (2-5)

III x II III

z

y x x

I I

Wy

d

d Wx d

Figure (2-3): Optical waveguide is surrounded by PML regions

where,

xx yy z

is is is

x

yz

∂∂

∇= + +

∂∂

∂

, (2-6)

where ix, iy, and iz are the unit vectors in the x, y and z directions, respectively, and the

values of sx and sy are summarized in table (2-1).

2 Theoretical considerations

40

Region sx sy

I 1 s

II s 1

III 1 1

Table (2-1): The definition of sx and sy

with:

(2-7)

2

0

11

em

sj j

n0

σ

ω

ξω

=− =−

μ

where:

E and H are the electric and the magnetic field vectors, respectively,

ω

is the angular frequency,

0

ξ

and μ0 are the permittivity and the permeability of free space, respectively,

n is the refractive index and

e

σ

and m

σ

are the electric and the magnetic conductivities of PML, respectively.

The relation (2-7) is required to satisfy the PML impedance matching condition:

2

00

e

n

m

σ

ξ

μ

= (2-8)

which means that the wave impedance of a PML medium exactly equals that of the

adjacent medium with refractive index n in the computational window:

0

2

0n

μ

ξ

,

2 Theoretical considerations 41

Regardless of the angle of propagation or frequency. In the PML medium, we

assume a q–power profile of the electric conductivity as:

max

q

ed

ρ

σσ

⎛⎞

=⎜⎟

⎝⎠

(2-9)

where

ρ

is the distance from the beginning of PML. Using the theoretical reflection

coefficient R at the interface between the computational window and the PML medium:

max

00

exp 2

q

d

R

cn d

σρ

d

ρ

ξ

⎡

⎛⎞

=−

⎢

⎜⎟

⎝⎠

⎢⎥

⎣⎦

∫⎤

⎥

(2-10)

The maximum conductivity max

σ

can be determined as:

0

max

1

ln

2

qcn

dR

ξ

σ

+

⎛⎞⎛⎞

=⎜⎟⎜⎟

⎝⎠⎝⎠

1

(2-11)

where c is the light velocity of free space. Usually, a parabolic profile is assumed for the

conductivity, q = 2, and thus, s in (2-4) and (2-5) is finally written as:

2

3

1

4

sj

nd d R

λρ

π

⎛⎞

=− ⎜⎟

⎝⎠

1

ln,

in the non-PML region

s = 1, in the PML region

where 2c

π

λ

ω

= is the free-space wavelength.

Under the scalar approximation, from (2-4) and (2-5) the following basic equation

can be obtained:

2

00

y

x

xy

s

sp sp sp kgs

xsx ysy z z

⎛⎞

∂∂Φ∂∂Φ∂∂Φ

⎛⎞ ⎛⎞

+

++

⎜⎟ ⎜⎟

⎜⎟

∂∂∂∂∂∂

⎝⎠ ⎝⎠

⎝⎠ Φ=

(2-12)

where k0 is the free space wavenumber, and

Φ

, p and are given by: g

2 Theoretical considerations

42

Φ = Ex, p = 1 and = n2 for quasi-TE modes and g

Φ = Hx, p = 1 / n2 and = 1 for quasi-TM modes. g

substituting a solution of the form:

()()(

00

,, ,, exp

)

x

yz xyz jknzΦ=Ψ −

(2-13)

into (2-12), and assuming Fresnel approximation, the following beam propagation

equation for the varying complex amplitude

Ψ

can be obtained:

()

22

00 00

20

y

x

xy

s

jk n ps s p s p k s g n p

zxsx ysy

⎛⎞

∂Ψ ∂ ∂Ψ ∂ ∂Ψ

⎛⎞

−+ + +−

⎜⎟

∂∂ ∂ ∂ ∂

⎝⎠

⎝⎠ Ψ=

(2-14)

re n0 is the reference refractive index.

By solving the last equation the effective index of the desired mode can be

This was one mathematical way to design a monomode optical waveguide, and

Figure (2-4) is representing a slab waveguide refractometric sensor in which the

Evanescent wave sensing of a chemical quantity which is homogeneously

whe

calculated, and then the field distribution and effective index of the fundamental mode

are obtained. The value of reference refractive index, n0 may be arbitrarily chosen.

now the design task will be transferred to find the waveguide structure which

maximizes the sensitivity on the quantity to be measured.

measurand is homogeneously distributed in the cover (homogeneous sensing) and

assuming that the cover medium is a fluid, which implies that the contact zone between

the cover and the waveguide is of zero thickness.

distributed in the waveguide cover refers to a different electromagnetic condition. The

sensitivity is now related to the integral of the squared evanescent field in the cover

material. A waveguide refractometric sensor can be used in liquid concentration

monitoring, for measuring traces of chemicals by means of a thick selective membrane

and, more generally, for measuring all chemical quantities whose variation corresponds

to a change of refractive index.

2 Theoretical considerations 43

Figure (2-4): Schematic representation of a slab waveguide refractometric sensor

The sensing operation consists of the measurement of changes of the real or

Nevertheless, for practical application the sensitivity should be as high as possible

here:

, ,

imaginary part of the effective refractive index caused by measurand-induced changes

of the refractive index of the cladding. Hence, the sensitivity is defined as the ratio of

the effective index change of the guided mode to the refractive index change of the core

layer environment, i.e., both cladding and substrate. This ratio is 1 for a free-space

beam.

and therefore the waveguide parameters should be chosen with care.

w

2

eff eff

N

ε

=2

cc

n

ε

=2

s

n

ε

=

, and

her

2

ww

n

ε

=

w e eff

ε

,c

ε

,

s

ε

and w

ε

are the effective, cover, substrate and waveguide permittivities

ective, cover, substrate and waveguide indices respectively.

By starting the analysis with the well-known characteristic equation for a three-

respectively.

Neff, nc, ns and nw are the eff

layer slab waveguide [50]:

Substrate ns

Waveguide nw

Cover nc

w

2 Theoretical considerations

44

2

22

2

22

2arctan

arctan 0

p

eff s w

weff

s

weff

p

eff c w

c

weff

Nn

wn

nN n

nN

Nn

nm

n

nN

π

λ

π

⎡⎤

−⎛⎞

⎢⎥

−− ⎜⎟

−

⎢⎥

⎝⎠

⎣⎦

⎡⎤

−⎛⎞

⎢⎥

−=

⎜⎟

−

⎢⎥

⎝⎠

⎣⎦

−

(2-15)

where w is the thickness of the waveguide core layer, and m is an integer defining the

mode order, p = 0 for TE- and p = 1 for TM-polarization while

λ

is the free-space

wavelength.

Let:

22

2

eff s

s

weff

Nn

xnN

−

=−2

, 22

22

effw

ceff

cNn

nN

x−

−

= (2-16)

and

2

w

s

n

α

= ,

2

w

c

n

α

= (2-17)

by using equations (2-16) and (2-17), equation (2-15) can be rewritten in a more

normalized form:

2

1

2arctan arctan 0

1

ws

ss cc

s

wn xxm

x

πααα

λ

−

−−−

+

π

=

(2-18)

with

s

α

= c

α

= 1 for TE modes.

The sensitivity S of homogeneous sensing is defined as the rate of change of the

modal effective index Neff with respect to the change in the cover index nc.

1

eff c

ceff

Nn

SnN

−

⎛⎞

∂∂

==

⎜⎟

⎜

∂∂

⎝⎟

⎠

, (2-19)

2 Theoretical considerations 45

22

1

11

{1

11

arctan arctan }

TE c c c

cc

sc

cs

Sxxx

xxm

xx

αα

π

−

=++

⎡⎤

++ + +

⎢⎥

⎣⎦

(2-20)

and,

2

22

2

22

2

21

1

s

scs

s

cs

s

TM

ccc c

c

x

S

xx x

αα

αβ

α

+

−⎛⎞

++

⎜⎟

⎛⎞

⎝⎠

+

⎜⎟

⎝⎠

=⎛⎞

++ +

⎜⎟

⎝⎠

(2-21)

where:

2

1

arctan arctan 1

s

ss cc

ss s

s

x

xxm

xx

βα απ

αα

⎡⎤

⎛⎞

⎢⎥

⎜⎟

+

⎢⎥

⎜

=+++

⎢⎥

⎜⎟

⎛⎞

⎢⎥

+

⎜⎟

⎢⎥

⎝⎠

⎣⎦

⎟

(2-22)

Searching for the condition of the maximum sensitivity in a structure of constant

nw, ns and nc amounts one can get the exact maximum condition after some

mathematical operations:

for TE modes:

11

arctan arctan

cs

x

xm

xx

π

⎡⎤

++ + +

⎢⎥

⎣⎦

23

2

1

11

31

s

cc

s

3

1

s

x

xx

x

α

⎡⎤

−

⎢⎥

++ = +

⎢⎥

+

⎢⎥

⎣⎦

(2-23)

2 Theoretical considerations

46

and for TM modes:

()

2

22

11

21 12

12

1

ss

c

xx

xr x

αα

2

1c

s

x

β

θ

αθ

⎡⎤

⎛⎞

−

⎢⎥

+

⎜⎟

⎛⎞

⎢⎥

⎝⎠

⎜⎟

++ + −−

⎜⎟

⎢⎥

+

θα

⎝⎠

⎜⎟

⎛⎞

⎢⎥

⎜⎟

+

⎜⎟

⎝⎠

⎢⎥

⎝⎠

⎣⎦

2

22

11

23

11

cc c

s

c

sscc

sc

t

xt t x r

xx

γ

βα α

ααα

αα

⎡⎤

⎢⎥

⎛⎞

⎢⎥

+++ + +

⎜⎟

⎢⎥

⎛⎞⎛⎞

⎝⎠

⎢⎥

++

⎢⎥

⎜⎟⎜⎟

⎢⎥

⎢

⎝⎠⎝⎠

⎣

⎣⎦

0=

⎥

⎦

(2-24)

where:

2

1

cc c

c

rx x

αα

⎛

=+

⎜

⎝⎠

⎞

⎟

(2-25)

1

s

c

s

x

t

x

α

−

=

−

(2-26)

2

1

s

cs

s

x

θ

αα

+

=⎛⎞

+

⎜⎟

⎝⎠

(2-27)

()

222

22

2

22

2

11

21

1

sss

s

ss

s

2

3

s

x

xx

αα

γ

α

⎛⎞ ⎛

+−+ +

⎜⎟ ⎜

⎝⎠ ⎝

=⎛⎞

+

⎜⎟

⎝⎠

x

⎞

⎟

⎠

(2-28)

The normalized waveguide thickness achieving maximum sensitivity can be

obtained by substituting the xs solutions of equations (2-23), (2-24) into the

characteristic equation (2-18) of the three layer structure for TE0 and TM0 modes:

2 Theoretical considerations 47

[]

2

11 arctan arctan

ws

1

21

s

scc

nw x

s

x

αα

+

=+

λπ

α

−

(2-29)

In the case of surface sensing the measurand in the slab waveguide refractometric

sensor is an ultrathin film at the waveguide-cover interface as shown in figure (2-5).

The sensitivity in surface sensing is related to the squared field magnitude at the

waveguide-cover interface [50].

Waveguide nw

Cover nc

w

d

nf

Substrate ns

Figure (2-5): Schematic representation of a surface sensor

By starting the analysis with the normalized characteristic equation for the surface

sensing case:

()

2

1

2arctan arctan 1 0

1

ws

ss c c

s

wn xx

x

παααδ

λ

−

−−−−

+m

π

=

(2-30)

where for TE modes:

(

2

)

f

c

d

π

δ

εε

λ

=−

(2-31)

and for TM modes:

2 Theoretical considerations

48

()

111

2eff

cf

fc

eff c

d

εεε

π

δεε

λεε

⎛⎞

+−

⎜⎟

⎝⎠

=− − (2-32)

The sensitivity S of a surface sensing structure is defined as the rate of change of

the model effective index Neff with respect to the dielectric load term that is defined as:

(

2

)

f

c

d

π

η

εε

λ

=−

, (2-33)

1

eff

N

SN

η

−

⎛

∂∂

==

⎜

⎜⎟

∂∂

⎝⎠

⎞

⎟

, (2-34)

()

22

1

11

1 arctan arctan

s

TE

sc s c

1

s

cs

S

xx x xm

x

α

π

α

−

=⎡⎤

++ + +++

⎢⎥

⎣⎦

,

(2-35)

and

2

222

11

1

111

1

c

cc

ccc c

cc

TM

x

px

x

S

α

αα

ϑ

⎛⎞⎛⎞

+−

⎜⎟⎜⎟

−

+

⎜⎟⎜⎟

++

⎜⎟⎜⎟

⎝⎠⎝⎠

= (2-36)

where:

(

)

()

2

1

arctan arctan 1

1

s

ss cc

ss s

s

c

cc c

c

x

xxm

x

xx

ϑα απ

αα

+

=+++

⎛⎞

+

⎜⎟

⎝⎠

+

+⎛⎞

+

⎜⎟

⎝⎠

(2-37)

and,

2 Theoretical considerations 49

2

w

c

f

pn

ε

α

=+

(2-38)

For maximum surface sensitivity for TE modes:

2

33

1

11

arctan arctan 3 1

11

0

s

cs

cs s

s

cs

xxm

xx

x

xx

α

π

α

⎛⎞

−

⎜⎟

⎡⎤

⎜⎟

+++++

⎢⎜

⎣+

⎜⎟

⎝⎠

−−=

⎥

⎟

⎦

(2-39)

and for TM modes:

()

2

22

2

22

2

22 2

22

2

11

111

12

1

11

1

21

3.

1

2

1

11

s

sc 2

s

c

sc

s

cc

cs

ss

ss s

s

x

px

x

xx

xxxx

xx

px

x

αα

α

αα

α

⎧

⎛⎞⎛⎞

+−

⎪

⎜⎟⎜⎟

⎡

⎤

⎛⎞

⎪

⎜⎟⎜⎟

−−+Γ

⎢

⎥

⎨⎜⎟

+

⎜⎟⎜⎟

⎢

⎥

⎝⎠

⎪

⎣

⎦

−

⎜⎟⎜⎟

⎪

⎝⎠⎝⎠

⎩

⎫

⎛⎞ ⎪

+

⎜⎟

⎛⎞

⎪

⎝⎠

+−−

⎬

⎜⎟

⎛⎞

⎝⎠

⎪

+

⎜⎟ ⎪

⎝⎠ ⎭

⎡⎤

⎛⎞

−−

⎢⎥

⎜⎟

⎝⎠

⎢⎥

+

⎢⎥

+

⎢

⎣⎦

−

2

10

cc

c

x

ϑα

α

⎛⎞

+=

⎜⎟

⎝⎠

⎥

(2-40)

where,

2

11

1

c

cc c

c

x

ϑα

α

⎛⎞

⎜

+

⎜

Γ= − ⎜⎟

+

⎜⎟

⎝⎠

⎟

(2-41)

The normalized waveguide thickness achieving maximum sensitivity can be

obtained as before in the homogeneous sensitivity:

2 Theoretical considerations

50

[]

2

11 arctan arctan

1

21

ws

s

scc

s

nw x

x

αα

λπ

α

+

=+

−

(2-42)

By taking all considerations above into account during the design of a monomode

waveguide which operates with signal wavelength 633 nm (in the visible range) the

strip-loaded waveguide shown in figure (2-6) was designed.

Silicon

n = 1.46 SiO2

n = 1.55 SiON

TEOS

3 μm

n = 1.46

2.0 μm

0.5 μm

0.1 μm

0.4 μm

Figure (2-6): Schematic cross section and target design parameter of a strip-loaded

waveguide

2.3 Design of Mach–Zehnder interferometer (MZI)

The Mach–Zehnder interferometer (MZI) is one of the work-horses in the field of

IO chemical sensors. On the one hand because of its performance and on the other hand

because it is relatively simple to implement its basic structure in integrated optics and to

provide it with a chemo-optical interface layer. And indeed many papers have been

published in which it is shown that an IO MZI loaded with a certain type of interface

enables the determination of low concentrations of certain chemicals.

The IO MZI sensor will be treated in greater detail in order to illustrate the

approach which is needed to develop stable, reproducible IO chemical sensing systems

2 Theoretical considerations 51

with high resolution and low detection limit, for the market demands of applications

such as determining contaminant concentrations in food or protein concentrations in

health care. First the principle of the IO MZI sensor will be explained. Next an approach

will be given for arriving at an MZI implementation, which shows a very high

resolution of the refractive index measurement and finally some other interesting MZI

implementations will be discussed.

The structure of the IO MZI consisting of two monomode waveguides is given in

figure (2-7) [51]. At the input side the mode is excited by coherent monochromatic light

from e.g. a fiber. Next the modal power Pin is equally divided into two parallel branches

by a Y-splitter, and after traveling these branches the branches merge again at a second

Y-junction, where the modes from both waveguides interfere with each other. Their

phase difference, Δφ, defines the power of the mode in the output waveguide Pout.

Figure (2-7): Mach – Zehnder interferometer (MZI) [51].

Assuming a lossless MZI device with symmetrical Y-junctions it can be derived

that:

Pout / Pin = (1 / 2) (1 + cos Δφ) (2-43)

Using the previous equation, the transfer function of the MZI, i.e., the power ratio

Pout / Pin, is depicted as a function of Δφ in figure (2-8).

2 Theoretical considerations

52

0

1

0 45 90 135 180 225 270 315 360 405 450 495 540

Phase difference (dgree)

The power ratio

Figure (2-8): Plot of the transfer function of the MZI versus the phase difference Δφ

Such an MZI device can be applied for many purposes. By applying an electro-

optical material and electrodes in one of the branches, the MZI can be used as an on/off

modulator, while if it was provided with one or more additional input channels and if

opto-optical materials were inserted it could also function as an optical transistor or a

logic function. For sensing, however, one applies locally (within the so-called sensing

window) on the top of one of the branches (the sensing branch) a chemo-optical layer.

The MZI used as a sensor utilizes the transduction chain:

ΔC → Δn → ΔNeff → Δφ → Δ (Pout / Pin) (2-44)

where Δx denotes a change of the parameter x.

In all IO sensors a change of the value of the measurand (ΔC) affects the optical

properties or position of one or more of the materials forming the waveguide. The

extent of this change depends on the value of the measurand. The effects of these

changes on modal parameters are in most cases not only nearly proportional to the

change of optical properties but also to the fraction of the modal power which

propagates through that material, and in fact it is this fraction which interrogates with

the material. The most relevant optical properties which may be influenced are the

refractive index, the absorption coefficient or the emission properties (e.g.

2 Theoretical considerations 53

luminescence) of one of the materials. A change of the refractive index (Δn) expresses

itself as a change of two parameters: the real part of Neff and the field profile. A change

of an absorption coefficient manifests itself in the imaginary part of Neff and hence in

the modal attenuation. Some of the absorbing material may show luminescent

properties: part of the absorbed power is emitted as photons of another wavelength. In

general the major part of the emitted power leaves the system as free space radiation,

while a minor part is captured by the waveguide generating a new mode of another

wavelength.

Changes of the (complex) refractive index distribution in a region which is

relevant for the waveguide can be caused by physical (e.g. thermo-optical, electro-

optical, mechano-optical) or chemo-optical cross effects and in addition by partial

substitution of one material by another one [52]. For obtaining large sensitivities it is

preferred that all waveguide materials show the relevant cross effect in order to utilize

the modal power most effectively; in addition, it would be advantageous if all these

materials possessed large cross coefficients.

The third step in the transduction chain can be quantified by the expression:

Δφ = ΔNeff L (2π/λ0) (2-45)

where L is the interaction length (the length of the channel section provided with the

interface or simply the sensor area) and λ0 is the vacuum wavelength of the

monochromatic light source.

Using the inverse transfer functions the concentration change can be derived from

the change of the output power. This sounds simple, but in fact a too simple picture has

been presented; reality is more unruly: many other physical phenomena appear to

interfere in such a transduction chain: the perturbing factors. Maximizing resolution

implies on the one hand maximizing all transfer functions and hence maximizing all

partial sensitivities, such as (∂Neff / ∂n), [∂ (Δφ) / ∂Neff] and [∂ (Pout / Pin) / ∂ (Δφ)] but

on the other hand minimizing the effects of all perturbing factors.

Starting with maximizing the sensitivities, it can be immediately concluded that [∂

(Pout / Pin) / ∂ (Δφ)] is at maximum at the quadrature points: Δφ = (2m + 1) π / 2, and

from (2-45) that [∂(Δφ) / ∂Neff] will increase the longer the interaction length and the

lower the wavelength. Both latter parameters are limited however: the first one mainly

because the chip area is restricted, limiting L usually to about 1 cm; the second one as a

consequence of material properties: the increase of absorption- and scatter-losses with

decreasing wavelength. Hence generally a wavelength in the visible region is chosen

while its precise value may depend on the requirements of the application aimed at.

Also imperfectness of technology is a complicating factor. It hampers e.g. achieving

perfect mirror symmetry of the Y-junctions, which had been implicitly assumed in

deriving expression (2-43); also it hampers reaching the quadrature points at a preset

2 Theoretical considerations

54

value (e.g. zero) of the concentration. Fortunately, these imperfections can be easily

accounted for by inserting additional parameters (the visibility factor V and the phase

offset Δφ0 whose values have to be determined experimentally) into expression (2-43):

Pout = (Pin / 2) (1 + V cos (Δφ + Δφ0)) (2-46)

There are a number of perturbing factors, which will affect the Δφ → P

out

transfer:

- Scatter light will be generated in the IO circuitry of which a part Pscat may be

detected at the output port of the MZI

- Expressions (2-43) and (2-46) are correct if one mode of one given

wavelength propagates through the system only. While launching the zeroth-

order TE mode from a fiber generally also the zeroth-order TM mode is

weakly excited and the power of this TM00-mode has to be removed from the

system

- A laser never shows a constant output: wavelength-, phase- and output

power-fluctuations may occur, e.g., as a consequence of temperature

fluctuations. Also mechanical instabilities can influence the Pout / Pin ratio

- Note also that if incidentally starting from (Δφ + Δφ0) = m π (m being an

integer) it is unknown whether Δφ is increasing or decreasing (directional

ambiguity) while also near this state the partial sensitivity [∂Pout / ∂ (Δφ)] is

very low (sensitivity fading).

100

μ

m

φ

r

Sensin

g

are

a

Figure (2-9): Top view of the basic configuration of an IO Mach-Zehnder

Interferometer

In the integrated version of MZI, an input optical waveguide is split into two arms

which, after a certain distance, recombine again in an output optical waveguide as

shown in figure (2-9). Several MZI configurations were designed with varying

parameters but there are some fixed principles which must be taken into consideration

in each design in order to have small propagation losses. These principles are:

2 Theoretical considerations 55

- The Y junction is formed with straight arms and small opening angles (i.e. °1)

- The Y junction is shaped with circular bends with radii ranging between (200

μ

m to 5 mm)

- Separation between the sensor and the reference branches must exceed 50

μ

m

to avoid coupling between modes traveling through both branches and the width

of the branch is in the range of 3

μ

m to get monomode waveguide.

- In one arm, a (0.5, 1 or 2 mm) length and from 10 to 20

μ

m wide sensor area is

created.

- The total length of the device is ranging between 7 to 9 mm to get a small chip

area.

- The visibility signal (difference between the maximum and minimum intensity)

depends on the coupling factor of the divisor and on the propagation losses of

the guided mode in the interferometer arms. It must be at maximum and

therefore it is important to design a divisor or Y junction with a coupling factor

of 3 dB, that is, input light is equally divided in each branch of the

interferometer. Moreover, propagation losses in the sensor and reference branch

should be identical.

2.4 The heater design

In an optical waveguide, the optical path of the propagation light can be modified

by thermal changes. A change in temperature of the waveguide by modifies the

optical phase of the light traveling through it by Δφ. This is due to two effects: the

variation in length of the waveguide due to a thermal expansion or contraction, and the

temperature-induced change in the refractive index [54], [55].

TΔ

These changes affect directly the optical phase of the light, given by:

λ

π

ϕ

nL4

= (2-47)

where n is the refractive index,

λ

is the wavelength of the light and L is the waveguide

length.

The temperature effects on phase variations, TΔ

Δ

ϕ

, can be calculated by obtaining

the derivative of φ with respect to temperature T, and is given by:

bL

dT

d

λ

π

ϕ

4

= (2-48)

where:

2 Theoretical considerations

56

dn n dL

bdT L dT

⎛

=+

⎜

⎝⎠

⎞

⎟

(2-49)

The right hand terms of the last equation depend on the chemical composition of

the waveguide.

As shown in figure (2-10), one design for a heater which is made from titanium

and has two contacts from aluminum which will be used to inject electrical current to

heat the reference arm of the MZI waveguide in order to create the required difference

in refraction index that will control the propagation of light through the sensor and

increase the sensitivity of the sensor.

The heater is 300 nm thick, 5

μ

m width, 35

μ

m height, 10

μ

m distance

between vertical parts and 2 mm length. This heater will be placed at a distance equal to

3

μ

m away from the reference arm.

The thermal conductivity between the heater and the waveguide should be

maximized, but this is normally limited by the need for sufficient optical insulation. In

low-index contrast waveguides the distance between the core and the heater element

must be several micrometers. Increasing the thermal conductivity around the heated

waveguide makes the response faster, but it also increases the power consumption [55],

[56]. This leads to a trade-off between the speed and the power consumption. In many

waveguide structures the requirement for sufficient optical confinement also limits the

possibility to increase the thermal conductivity between the waveguide core and the

underlying substrate.

In order to calculate the total resistance of the heater to compare it with the

measured value, the total length L and the cross sectional area of the heater must be

calculated as shown in figure (2-11) and then the total resistance of the heater can be

calculated from the next equation:

L

RA

ρ

= (2-50)

where

ρ

, is the material resistivity of the heater. The total length of the heater is 11 mm,

its cross sectional area is 1.5 pm2, the resistivity of the titanium is and

hence the total resistance of the heater can be calculated as 3.08 K

8

42*10 .m

−Ω

Ω

.

2 Theoretical considerations 57

10

μ

m

35

μ

m

Aluminium

contacts

100

μ

m

Titanium heate

r

Figure (2-10): A top view of the titanium heater and the aluminium contacts

L

A

Figure (2-11). The heater as one part

With respect to the contacts, the resistivity of aluminium is and the

total resistance of the two contacts is 1.6

8

2.8*10 .m

−Ω

Ω

and therefore the resistance of the two

contacts is negligible with respect to the resistance of the heater.

Now the whole sensor is designed and is shown in figure (2-12). The next design

task is transferring this sketch to the computer using Cadence program in order to

fabricate the mask which will be used to fabricate the desired sensor.

2 Theoretical considerations

58

Figure (2-12): Optical waveguide based chemical sensor

Sensin

g

are

a

r

φ

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 59

3 Deposition of silicon nitride and silicon

oxynitride by LPCVD

3.1 Introduction

In order to optimize the deposition processes by LPCVD to obtain high

homogeneity layers of silicon nitride (Si3N4) and silicon oxynitride (SiON) to use them

as guiding layers in optical waveguide chemical sensors the appropriate temperature and

deposition rate had been chosen. Also the wafers must be distributed uniformly along

the boat with leaving two free places between the wafers. These deposited layers must

be suitable for both masking the local thermal oxidation and for coal etching. This

process gives layers with high homogeneity and is better than the physical deposition

processes.

The disadvantage of this process is that LPCVD is done at high temperatures

(these high temperatures enhance deposition processes). To obtain Si3N4 and SiON the

triethylsilane liquid (C6H16Si) was used instated of dichlorsilane gas because it is safe

and harmless.

Three processes were done with a temperature of 740 °C, 760 °C and 780 °C.

These different temperatures were chosen to help in optimizing the LPCVD process (to

determine the temperature and deposition rate which give higher homogeneity layers).

3.2 Description of LPCVD process

The deposition of silicon nitride (Si3N4) and silicon oxynitride (SiON) was done in

a three-zone oven for low pressure chemical vapor deposition shown in figure (3-1).

The LPCVD is a vacuum deposition process and it is suitable for the production of thin

layers. In the case of Si3N4, the ammonia (NH3) and triethylsilane (C6 H16 Si) were let to

flow into the tube and reacted with each other to develop Si3N4 layer on the wafers as

shown in the next equation:

4 NH3 + 3 C6 H16 Si Si3N4 + Secondary products

In the case of SiON, additionally to NH3 and C6 H16 Si, oxygen O2 also was let to

flow into the tube as shown in the next equation:

NH3 + C6 H16 Si + O2 SiON + Secondary products

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

60

The quantity of gases can be adjusted by mass flow controllers. The deposition

rate depends on the temperature and the pressure. At low temperatures the deposition

rate is very low because more reactive gas molecules go away than deposit on the

wafers. With increasing the temperature, the deposition rate increases on the surface of

the wafers.

Figure (3-1): 3-zone-oven used to produce SiO2 by thermal oxidation, TEOS

decomposition, deposition of Si3 N4 and SiON layers by LPCVD

The developed Si3N4 is stable mechanically and has a refractive index of (2.0 –

2.1). The developed SiON has a refractive index of (1.5 – 1.6). The boat and the carrier

used to carry the wafers into the tube are shown in figure (3-2) without the wafers and

in figure (3-3) with the wafers showing on it the wafers position, the direction of gas

inlet (1) and the direction of gas outlet (2).

The boat is made from quartz and can carry maximum 9 wafers since there are

two free places between wafers. This improves the homogeneity of the deposited layers.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 61

Figure (3-2): The carrier (1) and the boat (2) used to carry the wafers into the tube

Position 4 Position 6

Position 2 Position 8

Figure (3-3): The carrier and the boat with the wafers showing the wafers position, the

direction of gas inlet (1) and gas outlet (2)

The boat before going into the tube is shown in figure (3-4) where there is a

second quartz carrier (1) placed above the wafers to improve the heat distribution and

gas diffusion.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

62

Figure (3-4): The boat with the wafers just before going into the tube

Figure (3-5): The quartz tube of LPCVD

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 63

The wafers are placed in the carrier such that their surface is toward the tube inlet

hole (toward the direction of gases inlet). To improve the homogeneity of the deposited

layers, the boat must be fully occupied with each deposition; therefore blind wafers are

used to fill the places which have no wafers.

Then the boat is brought into the tube very slowly and very carefully since inside

the tube is very hot as shown in figure (3-5) and with moving the boat quickly the

wafers can be bended or broken due to the high temperature difference between inside

and outside the tube. After that, the tube will be evacuated and then heated to the

desired temperature and the deposition can be started.

3.3 Measuring of layer thickness and refractive index by ellipsometer

An ellipsometer which is shown in figure (3-6) enables to measure the refractive

index and the thickness of semi-transparent thin films. The instrument relies on the fact

that the reflection at a dielectric interface depends on the polarization of the light while

the transmission of light through a transparent layer changes the phase of the incoming

wave depending on the refractive index of the material. An ellipsometer can be used to

measure layers as thin as 1 nm up to layers which are several microns thick.

Applications include the accurate thickness measurement of thin films, the identification

of materials and thin layers and the characterization of surfaces.

Figure (3-6): The ellipsometer

The principle of operation of an ellipsometer is illustrated by the schematic

drawing of the ellipsometer shown in the figure (3-7) below:

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

64

Figure (3-7): Schematic drawing of an ellipsometer

It consists of a laser (commonly a 632.8 nm helium/neon laser), a polarizer and a

quarter wave plate which provide a state of polarization which can be varied from

linearly polarized light to elliptically polarized light to circularly polarized light by

varying the angle of the polarizer. The beam is reflected off the layer of interest and

then analyzed with the analyzer. The operator changes the angle of the polarizer and

analyzer until a minimal signal is detected. This minimum signal is detected if the light

reflected by the sample is linearly polarized, while the analyzer is set so that only light

with a polarization which is perpendicular to the incoming polarization is allowed to

pass. The angle of the analyzer is therefore related to the direction of polarization of the

reflected light if the null condition is satisfied. In order to obtain linearly polarized light

after reflection, the polarizer must provide an optical retardation between the two

incoming polarizations which exactly compensates for the optical retardation caused by

the polarization dependent reflections at each dielectric interface. Since the amplitude of

both polarizations was set to be equal, the ratio of the amplitudes after reflection equals

the tangent of the angle of the analyzer with respect to the normal [57].

3.4 Preparation of the silicon wafers

For the deposition of Si3N4 and SiON 45 wafers were required (24 wafer for

silicon nitride and 21 wafer for silicon oxynitride). The first step was marking the

wafers on their back surfaces. This was done by a special cutter with diamond edge.

After that, a cleaning process was done which remove any organic materials from the

surface of the wafers. Cleaning process steps are shown in table (3-1).

3.5 Deposition process

3.5.1 Silicon nitride (Si3N4) deposition

The deposition was done at temperatures of 740 °C, 760 °C and 780 °C and the

deposition time was 20 min. The wafer was placed at the center of the boat (position 4)

during deposition.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 65

Table (3-1): Cleaning procedures

3.5.1.1 Deposition of Si3N4 at 740 °C

The process parameters of this deposition are shown in table (3-2) and the results

of thicknesses and refractive indices are distributed on the wafer shown in figure (3-8).

1 Blow off with N2 To remove small particles from the wafers

2 Ultrasonic bath It consists of water and a wetting agent and it separates the

particles which are strongly bonded with the wafers

3 Washing in DI To remove the wetting agent of the previous cleaning step

water from the surface of the wafers

4 Piranhia-acid The acid consists of a H2 SO4 / H2 O2 solution and it solves

a any organic residues from the surface of the wafers

5 Washing in DI To remove the acid residues from the surface of the wafers

water

6 Drying process At the end, the wafers are washed in DI water and they

d dried using N2 in drying machine

Ammonia (NH3) flow rate 21 %

Triethylsilane (C6 H16 Si) flow rate 97 %

Process temperature 740 °C

Process pressure 0.6 mbar

Process time 20 min

Table (3-2): Process parameters for silicon nitride deposition at 740 °C

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

66

20.9 nm

(2.054)

20.2 nm 19.6 nm

(2.044) (2.043)

21.8 nm 21.1 nm 24.1 nm

(2.048) (2.047) (2.061)

23.3 nm 24.1 nm

(2.060) (2.061)

22.5 nm

(2.067)

Figure (3-8): Thicknesses and refractive indices distribution at 740 °C (wafer 1)

The deposition layer has a thickness ranging between 20 nm and 24 nm. The

homogeneity of the layer is not uniform since there is a thickness variation along the

wafer equal to 4.5 nm (18.7 %). The layer thickness increases at the parts which the

gases are moving and reacting faster. The average value for the thickness is 21.9 nm and

for the refractive index is 2.054. The deposition rate was approximately 0.9 nm/min

(low rate and the desired layers can not be deposited in an acceptable time).

3.5.1.2 Deposition of Si3N4 at 760 °C

The process parameters of this deposition are shown in table (3-3) and the results

of thicknesses and refractive indices are distributed on the wafer shown in figure (3-9).

Ammonia (NH3) flow rate 21 %

Triethylsilane (C6 H16 Si) flow rate 97 %

Process temperature 760 °C

Process pressure 0.6 mbar

Process time 20 min

Table (3-3): Process parameters for silicon nitride deposition at 760 °C

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 67

23.1 nm

(2.092)

23.5 nm 24.1 nm

(2.117) (2.097)

25.1 nm 23.7 nm 26.3 nm

(2.107) (2.107) (2.105)

25.9 nm 26.9 nm

(2.107) (2.107)

26.1 nm

(2.107)

Figure (3-9): Thicknesses and refractive indices distribution at 760 °C (wafer 2)

At a temperature of 760 °C, the deposition rate increased to 1.2 nm/min. The

homogeneity also improved because the thickness variation along the wafer decreased

to 3.8 nm (14.1 %). The gas deposition at the edge is more than at the center of the

wafer (the thickness is smaller at the center). The average value for the thickness is 24.7

nm and for the refractive index is 2.105.

3.5.1.3 Deposition of Si3N4 at 780 °C

The process parameters of this deposition are shown in table (3-4) and the results

of thicknesses and refractive indices are distributed on the wafer shown in figure (3-10).

Ammonia (NH3) flow rate 21 %

Triethylsilane (C6 H16 Si) flow rate 97 %

Process temperature 780 °C

Process pressure 0.6 mbar

Process time 20 min

Table (3-4): Process parameters for silicon nitride deposition at 780 °C

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

68

27.0 nm

(2.111)

26.5 nm 26.8 nm

(2.0977) (2.107)

28.3 nm 28.1 nm 29.1 nm

(2.120) (2.057) (2.107)

30.0 nm 29.6 nm

(2.123) (2.105)

30.0 nm

(2.088)

Figure (3-10): Thicknesses and refractive indices distribution at 780 °C (wafer 3)

The deposition shows the same result as the deposition at 760 °C. The thickness

variation is 3.5 nm (11.7 %). The deposition rate increased to 1.5 nm/min. The average

value for the thickness is 28.3 nm and for the refractive index is 2.102.

3.5.1.4 Temperature optimization

From the results of the three deposition processes, it is clear that, deposition at 740

°C is the worst one because the deposition rate is small and the thickness variation along

the wafer is large (18.7 %). This deposition can not deposited thick layers at short time

since the deposition rate is only 0.9 nm/min. The depositions at 760 °C and 780 °C

show better results with respect to the deposition rate and the homogeneity. Therefore,

the next deposition processes will be done at only 760 °C and 780 °C using three wafers

in each deposition to optimize the homogeneity along the boat. The wafers will be

placed at positions 2, 4 and 8 in the boat.

The results for depositions of wafers (4, 5 and 6) in the same boat at temperature

of 780 °C are shown in table (3-5) and of wafers (7, 8 and 9) in the same boat at

temperature of 760 °C are shown in table (3-6). It is found that the deposition rate

increased at the front place of the boat (position 2). The thickness variation along the

boat of wafers (4, 5 and 6) was 23.8 nm (52.3 %).

The thickness variation along the boat of wafers (7, 8 and 9) was 15.7 nm (47.43

%). The thickness variation along this boat is smaller than that of the first boat. This

means that the homogeneity was improved. Hence, the deposition at 760 °C had better

results and it is also preferred because the temperature is lower. Therefore, the next

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 69

depositions will be done only at 760 °C and to decrease the thickness variation

along the boat at temperature of 760 °C, the temperatures of the heating zones along the

tube must be optimized.

Temperature Wafer 4 Wafer 5 Wafer 6

780 °C Position 2 Position 4 Position 8

Measurement Thickness Refractive Thickness Refractive Thickness Refractive

No. (nm) index (nm) index (nm) index

1 38.4 2.066 29.2 2.115 23.0 2.098

2 42.4 2.022 31.2 2.125 25.0 2.115

3 45.5 2.076 32.0 2.016 25.0 2.107

4 41.5 2.076 30.1 1.980 23.3 2.075

5 36.0 2.044 27.9 1.966 21.2 2.057

6 35.9 2.066 28.2 1.967 21.8 2.068

7 37.8 2.057 28.5 1.967 21.5 2.063

8 39.8 2.076 29.9 1.986 23.1 2.070

9 42.7 2.081 31.1 2.006 24.5 1.966

The average 39.9 2.063 29.7 2.000 23.1 2.069

Deposition rate 1.995 nm/min 1.485 nm/min 1.155 nm/min

Thickness vari. 7 nm (16.7 %) 4.1 nm (12.8 %) 4 nm (16 %)

Table (3-5): The results for depositions of wafers (4, 5 and 6) in the same boat at

temperature of 780 °C

The next deposition was done on wafers (10, 11 and 12) in the same boat with

changing the temperatures of the heating zones along the tube. The temperature at the

front zone was 720 °C, at the middle zone was 753 °C and at the back zone was 759 °C.

The results of this deposition are shown in table (3-7).

By changing the temperature of the heating zones along the boat, it is found that

the gas deposition decreased at front of the boat since the front temperature is low. So

the temperature in the front side of the boat must be increased. The thickness variation

along the boat decreased to 9.6 nm (28.6 %). The homogeneity of the individual wafers

had been improved.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

70

Temperature Wafer 7 Wafer 8 Wafer 9

760 °C Position 2 Position 4 Position 8

Measurement Thickness Refractive Thickness Refractive Thickness Refractive

No. (nm) index (nm) index (nm) index

1 29.6 2.001 22.9 2.099 18.1 2.067

2 33.1 2.046 24.3 2.078 19.3 2.068

3 32.8 2.039 24.3 2.126 19.3 2.084

4 30.4 2.026 23.4 2.104 18.7 2.017

5 27.9 1.972 21.4 2.090 17.7 2.007

6 28.2 1.976 21.9 2.092 17.4 2.047

7 28.9 1.990 22.0 2.077 17.5 1.997

8 31.4 2.019 23.5 2.100 18.8 2.047

9 33.2 2.046 24.9 2.126 19.7 2.077

The average 30.6 2.013 23.5 2.099 18.4 2.046

Deposition rate 1.530 nm/min 1.175 nm/min 0.920 nm/min

Thickness vari. 5 nm (15.1 %) 3.5 nm (14 %) 2 nm (10.5 %)

Table (3-6): The results for depositions of wafers (7, 8 and 9) in the same boat at

temperature of 760 °C

The next deposition was done on wafers (13, 14 and 15) in the same boat with

changing the temperature of the heating zones along the tube. The temperature at the

front zone was 732 °C, at the middle zone was 754 °C and at the back zone was 759 °C.

The results of this deposition are shown in table (3-8).

By increasing the temperature of the front heating zone, the deposition rate

increased at the front of the boat. The thickness variation along the boat decreased to 5.6

nm (16.62 %). The wafers 14 and 15 have nearly the same layer thicknesses. The

temperature of the front heating zone must be increased in order to improve the

homogeneity of the layers.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 71

Front: 720 °C Middle: 753 °C Back: 759 °C

Wafer 10 Wafer 11 Wafer 12

Position 2 Position 4 Position 8

Measurement Thickness Refractive Thickness Refractive Thickness Refractive

No. (nm) index (nm) index (nm) index

1 24.7 2.093 30.6 2.097 31.3 2.109

2 27.1 2.083 31.7 2.105 32.8 2.113

3 26.6 2.103 31.7 2.109 33.6 2.121

4 25.9 2.067 31.6 2.103 33.1 2.113

5 25.1 2.087 30.1 2.097 31.5 2.097

6 24.0 2.086 30.1 2.102 30.3 2.116

7 25.1 2.087 30.1 2.109 29.8 2.097

8 26.1 2.085 31.2 2.106 31.7 2.107

9 26.1 2.088 31.5 2.097 33.5 2.112

The average 25.6 2.087 30.9 2.103 31.8 2.109

Deposition rate 1.280 nm/min 1.545 nm/min 1.590 nm/min

Thickness vari. 3 nm (11.1 %) 1.6 nm (5.05 %) 3 nm (9.0 %)

Table (3-7): Deposition of wafers (10, 11 and 12) in the same boat with changing the

temperatures of the tube to [Front zone: 720 °C, Middle zone: 753 °C and Back zone:

759 °C]

The next deposition was done on wafers (16, 17 and 18) in the same boat with

changing the temperatures of the heating zones along the tube. The temperature at the

front zone was 750 °C, at the middle zone was 756 °C and at the back zone was 760 °C.

The results of this deposition are shown in table (3-9).

The results in this deposition had been improved. The homogeneity of wafer 16 is

not optimized yet but it is increased along the boat. The thickness variation along the

boat decreased to 4.8 nm (15.1 %) and this was the best result could be obtained. The

homogeneity was optimized in this deposition and to test the ability of reproduction,

this deposition will be repeated.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

72

Front: 732 °C Middle: 754 °C Back: 759 °C

Wafer 13 Wafer 14 Wafer 15

Position 2 Position 4 Position 8

Measurement Thickness Refractive Thickness Refractive Thickness Refractive

No. (nm) index (nm) index (nm) index

1 28.1 2.093 32.2 2.087 32.2 2.121

2 30.3 2.101 33.3 2.117 33.4 2.146

3 31.3 2.110 33.1 2.112 33.3 2.127

4 29.0 2.087 32.6 2.116 33.7 2.117

5 29.6 2.080 31.3 2.101 30.4 2.114

6 27.5 2.090 31.0 2.103 30.9 2.116

7 28.1 2.097 30.9 2.097 30.9 2.114

8 30.2 2.107 32.7 2.114 33.1 2.118

9 30.9 2.077 33.2 2.112 33.6 2.117

The average 29.4 2.094 33.2 2.107 32.2 2.121

Deposition rate 1.470 nm/min 1.660 nm/min 1.610 nm/min

Thickness vari. 3.2 nm (10.22 %) 2.4 nm (7.21 %) 3.3 nm (9.79 %)

Table (3-8): Deposition of wafers (13, 14 and 15) in the same boat with changing the

temperatures of the heating zones along the tube to [Front zone: 732 °C, Middle zone:

754 °C and Back zone: 759 °C]

The last deposition was repeated but with 4 wafers instead of three and the fourth

wafer was placed at position 6 in the boat. The results of this deposition are shown in

table (3-10).

This deposition shows the reproducibility of the process since the thickness

variation along the boat is 4.7 nm (15.1 %) which is almost the same as before. The

homogeneity in this deposition is sufficient for the intended purposes.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 73

Front: 750 °C Middle: 756 °C Back: 760 °C

Wafer 16 Wafer 17 Wafer 18

Position 2 Position 4 Position 8

Measurement Thickness Refractive Thickness Refractive Thickness Refractive

No. (nm) index (nm) index (nm) index

1 27.6 2.087 29.6 1.977 30.3 2.097

2 30.2 2.095 30.6 2.023 31.6 2.104

3 30.2 2.097 31.6 2.017 31.4 2.116

4 29.3 2.077 30.3 2.017 30.3 2.106

5 27.2 2.079 28.7 2.006 29.1 2.097

6 27.0 2.083 28.4 2.006 29.0 2.087

7 27.2 2.086 30.1 2.019 28.7 2.089

8 28.9 2.084 29.0 2.096 30.8 2.099

9 30.2 2.077 30.5 2.036 31.8 2.101

The average 28.6 2.085 29.8 2.022 30.3 2.100

Deposition rate 1.430 nm/min 1.490 nm/min 1.515 nm/min

Thickness vari. 3.2 nm (10.60 %) 3.2 nm (10.13 %) 3.1 nm (9.75 %)

Table (3-9): Deposition of wafers (16, 17 and 18) in the same boat with changing the

temperatures of the tube to [Front zone: 750 °C, Middle zone: 756 °C and Back zone:

760 °C]

3.5.1.5 Ammonia flow optimization

In the next depositions the amount of ammonia will be changed to see its effect on

homogeneity and refractive index. The first deposition (wafer 23) was done with flow

rate equal to 30 % (63.18 SCCM) of ammonia. The parameters of this process are

shown in table (3-11) and the results are shown in figure (3-11).

It is found that the deposition rate is 1.385 nm/min, the thickness variation along

the wafer is 4.6 nm (14.84 %), the average thickness is 27.7 nm and the average

refractive index is 2.085.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

74

Front: 750 °C Middle: 756 °C Back: 760 °C

Wafer 19 Wafer 20 Wafer 21 Wafer 22

Position 2 Position 4 Position 6 Position 8

Meas. Thick. Refr. Thick. Refr. Thick. Refr. Thick. Refr.

No. (nm) index (nm) index (nm) index (nm) index

1 27.4 2.097 28.7 1.097 29.1 2.097 28.5 2.057

2 29.0 2.112 29.9 2.083 30.1 2.119 29.8 2.069

3 29.3 2.087 29.6 2.107 30.7 2.107 30.6 2.107

4 29.0 2.100 29.3 2.100 29.7 2.107 30.3 2.060

5 26.3 2.105 27.6 2.103 27.9 2.101 27.6 2.041

6 26.3 2.087 27.4 2.100 27.9 2.088 27.4 2.048

7 26.4 2.087 27.2 2.097 28.3 2.098 26.5 2.058

8 29.0 2.087 29.2 2.104 29.6 2.090 29.5 2.060

9 29.6 2.110 29.7 2.109 30.0 2.098 31.0 2.077

AV 28.0 2.097 28.7 2.100 29.6 2.101 29.0 2.064

DR 1.400 nm/min 1.435 nm/min 1.480 nm/min 1.45 nm/min

TV 3.3 nm (11.15 %) 2.7 nm (9.03 %) 2.8 nm (9.12 %) 4.5 nm (14.5 %)

Table (3-10): Deposition on wafers (19, 20, 21 and 22) in the same boat with changing

the temperature of the tube to [Front zone: 750 °C, Middle zone: 756 °C and Back zone:

760 °C] where AV is the average value, DR is the deposition rate and TV is the

thickness variation

Ammonia (NH3) flow rate 30 %

Triethylsilane (C6 H16 Si) flow rate 97 %

Process temperature 760 °C

Process pressure 0.6 mbar

Process time 20 min

Table (3-11): Process parameters for Si3N4 deposition with ammonia flow rate of 30 %

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 75

26.4 nm

(2.111)

26.6 nm 27.0 nm

(2.077) (2.067)

27.9 nm 27.4 nm 28.8 nm

(2.086) (2.067) (2.086)

31.0 nm 28.5 nm

(2.107) (2.077)

28.5 nm

(2.088)

Figure (3-11): Thicknesses and refractive indices distribution of Si3N4 deposition with

flow rate of 30 % ammonia (wafer 23)

The next deposition (wafer 24) was done with ammonia flow rate 15 % (31.59

SCCM). The parameters of this process are shown in table (3-12) and the results are

shown in figure (3-12).

Ammonia (NH3) flow rate 15 %

Triethylsilane (C6 H16 Si) flow rate 97 %

Process temperature 760 °C

Process pressure 0.6 mbar

Process time 20 min

Table (3-12): Process parameters for Si3N4 deposition with ammonia flow rate of 15 %

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

76

25.2 nm

(2.005)

26.4 nm 26.3 nm

(2.077) (2.007)

27.5 nm 26.8 nm 27.4 nm

(2.023) (2.068) (2.044)

28.0 nm 27.8 nm

(2.067) (2.069)

28.0 nm

(2.067)

Figure (3-12): Thicknesses and refractive indices distribution of Si3N4 deposition with

flow rate of 15 % ammonia (wafer 24)

It is found that the deposition rate is 1.35 nm/min, the thickness variation along

the wafer is 2.8 nm (10 %), the average thickness is 27.0 nm and the average refractive

index is 2.047.

From these results there is no much difference in the refractive index in both cases

but the homogeneity is better with ammonia flow rate 15 %.

3.5.2 Silicon oxynitride (SiON) deposition

The optimized parameters which were obtained from the deposition of silicon

nitride were used here to deposit silicon oxynitride in order to obtain a homogenous

layer to use it as a guiding layer in the waveguide which will be used as a transducer for

the desired chemical sensor. Every deposition was done with front zone temperature

750 °C, middle zone temperature 756 °C, back zone temperature 760 °C and pressure

0.6 mbar using one wafer and this wafer was placed at position 4 (the best position) in

the boat.

The process parameters used in the first deposition on wafer 1 are shown in table

(3-13) and the results of this deposition are shown in table (3-14).

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 77

Ammonia (NH3) flow rate 34.66 %

Triethylsilane (C6 H16 Si) flow rate 85 %

Process temperature (750-756-760) °C

Process pressure 0.6 m

bar

Process time 50 min

Oxygen (O2) flow rate 0.178 %

Table (3-13): Process parameters used in SiON deposition (O2 flow rate is 0.178 %)

0.178 % of O2 M1 M2 M3 M4 M5 Average

Thickness (nm) 598.330 598.910 597.003 602.001 599.423 599.133

Table (3-14): The results of SiON deposition with O2 flow rate of 0.178 %

From the results of this deposition it is found that the average refractive index

(1.504). The results show that the layer is homogenous since the thickness variation

along the wafer is approximately 5 nm (0.83 %) and the homogeneity is better than with

silicon nitride. In the next depositions the oxygen flow rate was decreased to increase

the refractive index of the silicon oxynitride with fixing the other parameters. The

results of the depositions on wafers (2 – 6) are shown in table (3-15) and on wafers (7 –

11) in table (3-16).

The results show that the layers of SiON are more homogenous than the layers of

Si3N4 since the thickness variation along the wafer in SiON is very small. From the

results shown in both tables, it is clear that, the lower the oxygen flow rate the lower the

deposition rate and the higher the refractive index of the silicon oxynitride. The

relationship between the oxygen flow rate and the refractive index of the silicon

oxynitride is shown in figure (3-13).

Refractive index 1.504 1.501 1.499 1.510 1.511 1.504

(Wafer 1) Deposition rate is 11.982 nm/min

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

78

0.166 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 4) Deposition rate is 11.102 nm/min

Thickness (nm) 551.312 553.479 554.019 558.989 557.676 555.095

Refractive index 1.533 1.539 1.533 1.524 1.537 1.531

0.162 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 5) Deposition rate is 10.818 nm/min

Thickness (nm) 540.991 542.432 539.987 538.762 542.319 540.898

Refractive index 1.549 1.547 1.541 1.539 1.532 1.542

0.174 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 2) Deposition rate is 11.689 nm/min

Thickness (nm) 583.111 582.963 584.073 585.601 586.496 584.449

Refractive index 1.501 1.533 1.540 1.499 1.491 1.513

0.170 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 3) Deposition rate is 11.426 nm/min

Thickness (nm) 569.039 569.998 571.438 573.111 572.867 571.291

Refractive index 1.511 1.517 1.524 1.531 1.534 1.523

0.158 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 6) Deposition rate is 10.525 nm/min

Thickness (nm) 525.374 525.989 527.231 526.317 526.412 526.265

Refractive index 1.551 1.553 1.556 1.548 1.549 1.551

Table (3-15): The results of deposition processes of SiON on wafers (2 – 6) with

decreasing the oxygen flow rate with each wafer

From figure (3-13), it can be seen that the higher the oxygen flow rate the lower

the refractive index. From this figure also the required oxygen flow which gives the

desired refractive index can be obtained.

3 Deposition of silicon nitride and silicon oxynitride by LPCVD 79

Table (3-15). The results of deposition processes of SiON for wafers 3 – 10 with

changing the oxygen flow rate with each wafer

Table (3-16): The results of deposition processes of SiON on wafers (7 – 11) with

decreasing the oxygen flow rate with each wafer

Deposition of wafers (12 – 21) were done with 0.158 % oxygen flow rate at

different times and the results were given graphically as shown in figure (3-14). The

figure (3-14) shows the relationship between the thicknesses of the deposited layers and

0.138 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 11) Deposition rate is 9.433 nm/min

Thickness (nm) 474.321 468.031 469.781 473.073 473.146 471.67

Refractive index 1.622 1.599 1.581 1.607 1.571 1.596

0.146 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 9) Deposition rate is 9.871 nm/min

Thickness (nm) 490.081 492.783 495.141 494.938 494.777 493.544

Refractive index 1.571 1.573 1.573 1.582 1.579 1.576

0.150 % of O2 M1 M2 M3 M4 M5 Average

(Wafer 8) Deposition rate is 10.082 nm/min

Thickness (nm) 501.999 503.246 504.491 504.663 506.081 504.096

Refractive index 1.561 1.564 1.572 1.577 1.570 1.569

0.154 % of O2 M1 M2 M3 M4 M5 Average

Thickness (nm) 511.931 513.702 514.678 519.317 516.321 515.190

Refractive index 1.561 1.559 1.554 1.562 1.558 1.559

(Wafer 7) Deposition rate is 10.304 nm/min

0.142 % of O2 M1 M2 M3 M4 M5 Average

Thickness (nm) 482.631 480.374 484.990 485.371 481.012 482.876

Refractive index 1.574 1.591 1.592 1.599 1.582 1.587

(Wafer 10) Deposition rate is 9.658 nm/min

3 Deposition of silicon nitride and silicon oxynitride by LPCVD

80

the deposition times. From this figure the required time which gives the desired

thickness can be obtained.

Figure (3-13): The relationship between the oxygen flow rate and the refractive index of

the deposited layers

1.440

1.460

1.480

1.500

1.520

1.540

1.560

1.580

1.600

1.620

0.138 0.142 0.146 0.15 0.154 0.158 0.162 0.166 0.17 0.174 0.178

Oxygen flow rate (%)

Refractive index

0

100

200

300

400

500

600

700

25 30 35 40 45 50 55 60 65 70

Time (min)

Thickness (nm)

Figure (3-14): The relationship between the thicknesses of the deposited layers and the

deposition times for SiON layers with flow rate of oxygen equal to 0.158 %.

4 Silicon oxynitride waveguide based ammonia sensor 81

4 Silicon oxynitride waveguide based ammonia

sensor

4.1 Introduction

As an application of low pressure chemical vapor deposition of silicon oxynitride,

the deposited silicon oxynitride layer will be used in this part as a guiding layer in the

designed Mach-Zehnder interferometer waveguide which will be used as a transducer in

an optical waveguide based ammonia sensor.

This sensor consists of an integrated Mach-Zehnder interferometer which has two

monomode waveguides, one of them represents the reference arm and the other

represents the sensing arm. A titanium heater with aluminium contacts will be

integrated near to the reference arm in order to increase the sensitivity of the sensor by

using the thermo-optical effect. A chemo-optical sensitive material to ammonia (its

refractive index changes with changing the amount of ammonia that diffuses into it

from the ambient air) will be spin coated on the sensing window in the sensing arm.

4.2 Sensor fabrication

The sensor fabrication processes are divided to three main processes which are the

MZI waveguide fabrication process, the heater fabrication process and the sensitive

layer fabrication process.

4.2.1 MZI waveguide fabrication

With respect to the MZI waveguide, the extensively studied structure is Si / SiO2 /

SiON / TEOS where silicon oxynitride (SiON) is used as core layer. All the information

required for SiON deposition using low pressure chemical vapor deposition (LPCVD)

were obtained from the last chapter. The first buffer layer is usually grown by thermal

oxidation (TO) and must be at least a 2 μm thick oxide layer (the buffer layer must be

thick enough to prevent absorption of the guided light by the Si substrate). Finally, a 0.5

μ

m Tetraethylorthosilicate (TEOS) layer will be deposited by LPCVD.

Fabrication of the waveguides was performed using our clean room facilities. The

first step in the fabrication process of the device is the growth of a 2

μ

m thick SiO2

lower cladding layer of refractive index 1.46 on a silicon wafer by thermal oxidation in

wet atmosphere at an oven temperature of 960 °C and a water temperature 95 °C for 30

4 Silicon oxynitride waveguide based ammonia sensor

82

h and then oxygen for 5 min [see table (4-1)]. Figure (4-1) shows the relation between

the time and the layer thickness during the oxidation process.

Oven temperature 960 °C

Water tem

p

erature 95 °C

Process time 30 h

Process

p

ressure 1 atm

.

Table (4-1): Process parameters used in thermal oxidation

Then the silicon oxynitride SiON core layer with a refractive index 1.55 and a

thickness of 0.5

μ

m is deposited by LPCVD at (750-756-760) °C and a pressure of

0.6 mbar. The SiON core layer of refractive index 1.55 can be deposited using flow of

Ammonia (NH3), Triethylsilane (C6 H16 Si) and Oxygen (O2) as shown in table (4-2).

0

50

100

150

200

250

300

10 20 30 40 50 60 70 80 90 100

Time (min)

Thickness (nm)

Figure (4-1): The thickness of SiO2 layers by thermal oxidation at 960 °C / 95 °C

4 Silicon oxynitride waveguide based ammonia sensor 83

Ammonia

(

NH3

)

flow rate 34.66 %

Trieth

y

lsilane

(

TES

)

flow rate 85 %

Ox

yg

en

(

O2

)

f

low rate 0.158 %

Process

p

ressure 0.6 mba

r

Table (3-2). Process parameters for SiON deposition.

Table (4-2): Process parameters for SiON deposition

Finally, a 0.5

μ

m Tetraethylorthosilicate [(C2H5)4SiO4] layer with a refractive

index 1.46 was deposited by LPCVD at temperature of 721 °C with pressure of 0.3

mbar as shown in table (4-3). Figure (4-2) shows the thicknesses of different TEOS

layers with time.

Table (4-3): Process parameters used in TEOS deposition

After each step of the last three processes, the average thickness of film deposited

on the silicon wafer is measured using an ellipsometer which also was used to calculate

the refractive index of the film.

A five point scan was done on the sample, in which the laser picked up five points

on the sample and measured the thickness and calculate the refractive index for each of

these points. The average of the five points was then taken to get the average thickness

and refractive index of the sample.

Tetraeth

y

lorthosilicate

(

TEOS

)

flow rate 24.57 %

Process

p

ressure 0.3 mba

r

Tem

p

erature 721 °C

De

p

osition time 85 min

De

p

osition rate 5.882 nm

/

min

Tem

p

erature

(

750-756-760

)

°C

De

p

osition time 47.5 min

Deposition rate 10.526 nm/min

4 Silicon oxynitride waveguide based ammonia sensor

84

0

100

200

300

400

500

600

700

10 20 30 40 50 60 70 80 90 100

Time (min)

Thickness (nm)

Figure (4-2): The thicknesses of deposited TEOS layers with different times

Then the photoresist was spin coated and the waveguide pattern is defined by

optical contact photolithography [see table (4-4)] and transferred to the TEOS layer by

reactive ion etching (RIE). This etching process is anisotropic, and it produces vertical

sidewalls that guarantee precise control and good reproducibility of the geometry and,

accordingly, of the effective lateral index contrast of the waveguide structure as shown

in figure (4-3).

Oven 45 min 150 °C

Photoresist adhesion 5 min N2

Blow 5 s 1000 rpm

Low spin 7 s 800 rpm

High spin 45 s 4000 rpm

Hot plate 60 s 110 °C

UV exposure 2.5 s Air humidity is 36 %

Development 40 s (1 l DI water + 8 g NaOH) in H2O [2:1]

Oven 45 min 130 °C

Table (4-4): Process parameters used in spin coating and photolithography for TEOS

4 Silicon oxynitride waveguide based ammonia sensor 85

μ

m 0.5

μ

m 1

Figure (4-3): SEM images of a cross section of the waveguide revealing the resulting

etch profile

This etching is done in a width of 3

μ

m by reactive ion etching (RIE) to form a

rib of 0.4

μ

m in the TEOS for light confinement in the SiON film below as shown in

figure (4-4).

4 Silicon oxynitride waveguide based ammonia sensor

86

The etching process is done using power equal to 1.4 KW, pressure equal to 25

mTorr and flow of two gases, the first is Argon (Ar) and the second is Trifluoromethane

(CHF3) as shown in table (4-5).

Ar

g

on

(

Ar

)

flow rate 51.724 %

Trifluoromethane

(

CHF3

)

flow rate

86.3 %

Process

p

ressure 25 m

Table (4-5): Process parameters used in RIE for TEOS layer (the upper cladding)

Figure (4-4): SEM image for reactive ion etching (RIE) in TEOS layer

After this, the mask had been removed and a cleaning step had been done to the

wafer. The fabrication process flow for SiON planar optical waveguide, as described

before, is illustrated in figure (4-5).

Tor

r

RF

p

ower 1.4 KW

Process time 23.4 min

Etchin

g

rate 17.14 nm/min

1

μ

m

4 Silicon oxynitride waveguide based ammonia sensor 87

SiO2

(a)

Si

SiON

SiO2

(b)

Si

TEOS

SiON

SiO2 (c)

Si

4 Silicon oxynitride waveguide based ammonia sensor

88

Photoresist

TEOS

SiON

SiO2 (d)

Si

Photomask Glass

Chrome

Photoresist

TEOS

SiON

SiO2 (e)

Si

4 Silicon oxynitride waveguide based ammonia sensor 89

Photoresist

TEOS

SiON

SiO2 (f)

Si

TEOS

SiON

SiO2 (g)

Si

Figure (4-5): Fabrication process flow for high-refractive index contrast SiON/SiO2

waveguide structures; (a) grown of SiO2 layer by using thermal Oxidation; (b)

depositing of SiON layer by using LPCVD; (c) depositing of TEOS layer by using

LPCVD; (d) spin coating the photoresist; (e) defining the waveguide pattern by using

optical contact lithography and development processes; (f) forming the rib in the TEOS

by using reactive ion etching (RIE); (g) removing the mask and cleaning the wafer

4 Silicon oxynitride waveguide based ammonia sensor

90

Figure (4-6) shows two SEM images of the MZI waveguide structure. Figure (4-6)

(a) represents a cross-section image for the waveguide structure and figure (4-6) (b)

represents an image for the splitting point of the MZI waveguide.

(a)

μ

m 2

(b)

Figure (4-6): (a) Represents a SEM image for the waveguide structure; (b) Represents a

SEM image for the splitting point of the MZI waveguide

4 Silicon oxynitride waveguide based ammonia sensor 91

Then the waveguide was tested for coupling and propagation loss. The light

coming from a red He-Ne laser (633 nm) is coupled into the waveguide by means of

end fire coupling and a photodiode detects the light coming out from the waveguide.

4.2.2 Heater fabrication

In order to construct the heater and its contacts the following steps were done:

- A layer of thickness equal to 300 nm of titanium was thermally evaporated on

the wafer [see table (4-6)]

- After spinning the photoresist and photolithography process the titanium was

wet etched using the materials shown in table (4-7). The etching rate was 50

nm/min and after that the wafer was cleaned

Power increasing 20 s until 5 % power

Wait 20 s

Power increasing 30 s until 8 % power

Delay 5 s

Rate 1 nm/s

Process time 5 min

Table (4-6): Process parameters of Titanium evaporation

DI water 300 ml

Hydrogen peroxide (H2O2) 180 ml

Ammonium hydroxide (NH4OH) 60 ml

Table (4-7): The materials which were used in titanium etching

- As before, a 300 nm aluminium layer was evaporated [see table (4-8)]. After

that, the photoresist was spin coated and photolithography was done. Then the

aluminium was wet etched using the materials shown in table (4-9). The etching

was done at temperature of 60 °C and had a rate of 1

μ

m/min.

- At the end, the rest of photoresist was removed and a cleaning step was done

Figure (4-7) shows a SEM image for the titanium heater with aluminium

contacts. The Mach-Zehnder interferometer, the titanium heater and aluminium contacts

all together can be seen in figures (4-8) and (4-9).

4 Silicon oxynitride waveguide based ammonia sensor

92

Power increasing 40 s until 10 % power

Wait 2 min

Power increasing 60 s until 18 % power

Delay 5 s

Rate 1 nm/s

Process time 5 min

Table (4-8): Process parameters of aluminium evaporation

1 2

840 ml Phosphoric acid (H3PO4) 85 % 1120 ml (H3PO4)

164 ml DI water 224 ml DI water

38 ml Nitric acid (HNO3) 65 % 50 ml HNO3

3 ml Antarox

[700 ml of (1) + 1400 ml of (2)] at 60 °C

Table (4-9): The materials which were used in aluminium etching

μ

m 10

Figure (4-7): A SEM image for the titanium heater with aluminium contacts

4 Silicon oxynitride waveguide based ammonia sensor 93

μ

m 100

Figure (4-8): A microscopic image for the structure consisting of MZI, the heater and

the contacts

μ

m 40

Figure (4-9): A SEM image showing the structure of the sensor

4.2.3 Sensitive layer fabrication

To build a sensor, sensitive material had to be selected that changes its refractive

index in response to the presence of a targeted chemical substance, such as ammonia in

ambient air. One of the choices was a thin film of polymer doped with an appropriate

indicator dye. Dye-doped polymers are traditionally used in optical chemical sensors, or

optodes, based on a change in optical absorption. In our case, however, the sensitive

optical parameter to be measured was not the absorbance but the refractive index.

4 Silicon oxynitride waveguide based ammonia sensor

94

As a sensitive layer, the polymer polymethyl methacrylate (PMMA) was used,

which can easily be dissolved in a number of commercial solvents and mixed with

various indicator dyes. It can also easily be processed into a thin layer by spin coating.

As an indicator dye that is sensitive to ammonia, Bromocresol Purple (BCP) was

selected [58].

The BCP was added to a solution of PMMA at a concentration of 5.5 % weight

concentration of the BCP in PMMA. The solution was filtered by a filter with 0.5 μm

pore size. Then it was spin coated at 4000 rpm for 60 s to give a layer thickness 200 nm

[see figure (4-10)] and backed on a hot plate for 60 s at 180 °C.

The following defines is the photolithography process. A UV lamp exposed the

samples for 3 min. The UV beam is aligned on the pattern from the mask. After all the

samples were exposed to UV, the samples were developed using methylisobutyl ketone

(MIBK) in isopropanol (IPA) [1:3 MIBK to IPA] for 30 s. Following the development

process, the samples were rinsed with DI water (de-ionized water). Baking the samples

in a convection oven at 95 °C for 30 min [see table (4-10)]. The final thickness of the

PMMA-BCP layer was close to 200 nm and its refractive index was 1.48.

0

100

200

300

400

500

1000 1500 2000 2500 3000 3500 4000

Spin Speed (rpm)

Film Thickness (nm)

Figure (4-10): The film thickness vs. spin-speed of the sensitive layer

4 Silicon oxynitride waveguide based ammonia sensor 95

Oven 45 min 150 °C

For layer adhesion 5 min N2

Blow 5 S 1000 rpm

High spin 60 S 4000 rpm

Hot plate 60 S 180 °C

UV exposure 3 min Air humidity 39 %

Development 30 S [1:3 MIBK to IPA]

Oven 30 min 95 °C

Table (4- 10): Process parameters used in fabricating the PMMA-BCP sensitive layer

4.3 Experimental results

The results are split into three main areas of study. The first is studying if the

waveguide has an expected behavior concerning insertion losses, propagation losses,

wavelength response and light confinement.

The second is studying the modulation of light using thermo-optical effect by the

heater in order to increase the sensitivity.

The third is studying the detecting of ammonia by studying the change in intensity

of the output port of the MZI.

4.3.1 Waveguide characterization

If the cladding layers are properly tuned, both in refractive index and thickness,

losses have a minimum at the operating wavelength (633 nm). If there were any

variation of these parameters, minimum losses will shift from the operating wavelength.

Using end-fire coupling as in the experimental setup shown in figure (4-11), losses

in test waveguides were measured. The losses sharply increase as waveguide gets

narrower. It has to be taken into account that total losses here include attenuation and

4 Silicon oxynitride waveguide based ammonia sensor

96

insertion losses. Care has been taken to minimize injection losses due to modal

mismatch, reflections and edge defects or irregularities. The insertion loss

measurements have been calculated equal to 1.4

±

0.2.

Micrometer

Photodiode

(a)

(b)

Figure (4-11): Experimental setup for coupling light to the waveguide by using end fire

method (a) Sketch; (b) Real image

Micropositioner

MZI

waveguide

Laser

source

Holder

Lens

Mirror

Microscope

4 Silicon oxynitride waveguide based ammonia sensor 97

Propagation loss measurements have been performed by measuring the intensity

of the transmitted light in waveguides at different lengths. From the fit results a

propagation loss coefficient of about 1.5

±

0.3 dB/cm. These losses are independent on

the coupling losses. Figure (4-12) shows the normalized intensity with respect to the

horizontal direction of the waveguide.

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

Normalized Intensity

-2 -1 0 1 2

Horizontal direction (μm)

Figure (4-12): The normalized intensity vs. the horizontal direction

4.3.2 Light modulation

By heating near the reference arm of the MZI waveguide using the metal heater

(titanium heater) the refractive index of the reference arm will change, and hence, there

will be a phase difference between the two optical branches, and hence, the output

intensity will be changed, and the heating process is done by using a voltage controlled

power supply. In general, this power is inversely proportional to the response time. The

response time of this device was calculated to be in the range of several milliseconds.

If the phase change of the waves in the two arms is reaching

π

, the optical output

power will be zero. The phase change

ϕ

Δ

can be expressed as follows [59], [60]:

4 Silicon oxynitride waveguide based ammonia sensor

98

TL

dT

dn Δ=Δ

λ

π

ϕ

2 (4-1)

Where

λ

is the wavelength of the transmitted light, dn/dT is the thermo-optic

coefficient of the silicon substrate, T

Δ

is the temperature difference between the two

waveguides and L is the heater length.

Taking into consideration the high thermal conductivity of the silicon substrate

(150 W/mK) and the small distance between the heater and the reference arm of the

Mach-Zehnder interferometer (3 μm) and knowing that the thermo-optic coefficient for

the SiON-based waveguide is equal to

5

102.1 −

×1

c

−

°, the transmitted light wavelength

is 633 nm and the heater length is 5.5 mm long, the temperature difference required to

give the different phase difference between the signals traveling in the two arms of

Mach-Zehnder interferometer can be calculated. Then a temperature difference of 5

is needed for having a phase difference equal to

C°

π

(zero output optical power) and a

temperature difference of 10 is needed for having a phase difference equal to zero

(2

C°

π

) and this gives maximum output optical power.

For most materials, the resistivity changes with temperature. If the temperature

range is not too large, the resistivity is a linear function of the temperature.

As the temperature of the heater is varied, its resistance changes due to:

- The temperature dependence of the resistivity

- The thermal expansion of the heater

For titanium, which is used here as a heater, the thermal expansion effects can be

neglected in comparison with the resistivity change. Since only the change in resistance

of the material is important, the overall resistance of the heater has a similar dependence

on the temperature:

(

0

1

)

0

R

RTT

α

=+−⎡

⎣⎤

⎦

(4-2)

where

α

is the temperature coefficient of resistance (TCR) and has a value of 0.0038

1

K

−

° for titanium (the material of the heater), T0 is the reference temperature, T is

temperature of interest, R0 is the resistance at reference temperature and R is the

resistance at temperature of interest where the influence of the temperature on the

resistance of the heater is plotted in figure (4-13). According to the plot, the resistance is

almost linearly increasing at this temperature interval.

4 Silicon oxynitride waveguide based ammonia sensor 99

1

1,005

1,01

1,015

1,02

1,025

1,03

1,035

1,04

012345678910

Temperature difference (K)

(Resistance ratio)

Figure (4-13): The resistance ratio (R/R0) as a function of temperature difference (T –

T0) for the titanium heater

2

2,2

2,4

2,6

2,8

3

3,2

3,4

3,6

0 5 10 15 20 25 30 35 40 45 50 55 60

Dissipated power (mW)

Resistance (kΩ)

Figure (4-14): Resistance of the heater against the dissipated power

However, changes in the resistance of the heater are observed during heating as

shown in figure (4-14). The resistance increases rapidly when more than 25 mW is

4 Silicon oxynitride waveguide based ammonia sensor

100

dissipated. Heating above 60 mW is destructive (the heater can not withstand more than

60 mW).

Figure (4-15) shows the effect of applying the heating power on the transmission

intensity of the light at 633 nm wavelength. As the power is increased, there is an

increase in transmission that peaks at 7 mW (constructive interference). A destructive

interference in the Mach-Zehnder device was achieved at 39 mW as indicated by the

transmission minimum. The maximum sensitivity was achieved at 24 mW.

0

0,2

0 8 16 24 32 40 48

Power (mW)

Tr

0,4

0,6

0,8

1

1,2

ansmission Intensity (a.u.)

Figure (4-15): Transmission intensity against electrical heating power

0.5

Maximum

Sensitivity

Constructive

Interference

Destructive

4.3.3 Ammonia sensing

The sensor under study relies on detection of a refractive index change at the

ammonia sensitive layer. This device falls into the so-called evanescent field sensing

technology wherein an electric field penetrates into the layer above the waveguide and

serves as a probe beam. The reversible reaction has associated with a refractive index

change to which the evanescent field is quite sensitive.

4 Silicon oxynitride waveguide based ammonia sensor 101

When the sensitive layer was exposed to ammonia a change in its absorption was

observed. An initially almost colorless film appeared to be yellow (the absorption

increase) after exposure and it was accompanied by an increase in the refractive index at

633 nm of maximum value equal to 0.004 [as shown in figure (4-16)], originating a

change in the effective refractive index of the guided mode and, therefore, in the MZI

sensor output.

0

0,0005

0,001

0,0015

0,002

0,0025

0,003

0,0035

0,004

0,0045

0 0,8 1,6 2,4 3,2 4 4,8

Ammonia concentration (%)

Refractive index change

Figure (4-16): The calculated refractive index change in the sensitive layer during

exposure to ammonia

The experimental setup shown in figure (4-17) is used for ammonia detecting

measurements. It consists of an external red He – Ne laser source, single mode silicon

optical fiber is used to couple the light directly into the waveguide inside the box, lens

for light beam converging, glass box with 7 holes (one of them is used to pump the

ammonia into the box, the second hole is used to pump the nitrogen into the box, the

third hole is used to pump the gases out the box, two holes are used to connect the wires

of the heater to the power supply and the last two holes are used to connect the

photodiode to the micrometer), mass flow controller to control the ammonia

concentration, photodiode to detect the light coming at the output port of the Mach-

Zehnder interferometer, microscope, micrometer, micropositioner and power supply.

All measurements were made at room temperature.

The hole which is used to pump the gases out the box is located at the opposite

side of the inlet tube which is used to pump the ammonia into the box so that ammonia

flows parallel to the sensitive surface. The gas must flow parallel to the surface to

minimize sample vibrations.

4 Silicon oxynitride waveguide based ammonia sensor

102

Power supply

Microscope

Optical fiber

Micro

p

ositione

r

Photodiode

Lens

Ammonia

N2

Sensor

Glass box

Pump

out

Micromete

r

Laser source

(a)

(b)

Figure (4-17): The experimental setup used for ammonia sensing measurements; (a)

sketch; (b) image

4 Silicon oxynitride waveguide based ammonia sensor 103

Light from an external laser source was coupled into the input port of the Mach-

Zehnder interferometer through a single mode optical fiber by using a lens, a

microscope objective and a micropositioner. The intensity at the output port was

detected by using a photodiode and a micrometer.

An electrical power equal to 24 mW (which gives maximum sensitivity) was

given to the heater for thermal tuning. And then, the glass box was closed carefully and

ammonia was pushed into the box. The total length of the sensitive layer was exposed to

the ammonia stream.

The recovery time is large in the case of ammonia with dry air and acceptable with

wet air. The best way to test the sensor is mixing the ammonia with air and with extra

water vapor and transfer the mixture into the glass box which initially had been filled

with air at atmospheric pressure (the higher relative humidity in the air the shorter the

time required for recovering change in the refractive index of the sensitive layer after

exposure).

0,75

0,8

0,85

0,9

0,95

1

1,05

0 0,8 1,6 2,4 3,2 4 4,8 5,6 7,2 8

Ammonia concentration (%)

Output intensity (a.u.)

Figure (4-18): Calibration plot for the sensor response represented by the output

intensity as a function of ammonia concentration

4 Silicon oxynitride waveguide based ammonia sensor

104

0

0,2

0,4

0,6

0,8

1

1,2

1,4

0 0,001 0,002 0,003 0,004

Refractive index change

Phase change (rad)

Figure (4-19): The phase change

ϕ

Δ as a function of the change in the refractive index

of the sensitive layer

0

10

20

30

40

50

60

70

80

90

1. Qrtl. 2. Qrtl. 3. Qrtl. 4. Qrtl.

Ost

West

Nord

0,75

0,8

0,85

0,9

0,95

1

1,05

0 40 80 120 160 200 240 280 320 360 400 440

time (s)

Output intensity (a.u.)

0,75

0,8

0,85

0,9

0,95

1

1,05

0 40 80 120 160 200 240 280 320 360 400 440

time (s)

Output intensity (a.u.)

0,75

0,8

0,85

0,9

0,95

1

1,05

0 40 80 120 160 200 240 280 320 360 400 440

time (s)

utput intensity (a.u.)O

Wet N2

0 % NH3

4.8 % NH3

Figure (4-20): The sensor response as a function of response time

4 Silicon oxynitride waveguide based ammonia sensor 105

Figure (4-18) shows the sensor response represented by the output light intensity

of the MZI as a function of the ammonia concentration. From this figure, the sensitivity

of the sensor could be calculated to be close of 2.4 % (ammonia concentration below

2.4 % could not be detected).

ϕ

The calculated phase change

Δ

between the light beams propagating in the two

MZI arms as a function of the change in the refractive index of the sensitive layer is

shown in figure (4-19).

With respect to the response and recovery times, it was found that the response

time is close to 120 sec and the recovery time is close to 240 sec as shown in figure (4-

20).

106

5 Conclusions 107

5 Conclusions

Low pressure chemical vapor deposition was optimized. This optimization is

concerning with the deposition rate and the homogeneity of the layers. The deposition at

740 °C was not good because the deposition rate was small and the thickness variation

along the wafer was large (18.7 %). This deposition can not deposit thick layers at short

time since the deposition rate was only (0.9 nm/min). The deposition at 760 °C was the

best because it had better results and it is also preferred on deposition at 780 °C since

the temperature is lower. To decrease the thickness variation along the boat at a

temperature of 760 °C the temperatures of the heating zones were optimized. The front

zone temperature was optimized to 750 °C, the middle zone temperature was optimized

to 756 °C and the back zone temperature was optimized to 760 °C. The homogeneity of

the silicon nitride was better at an ammonia flow rate of 15 % than at an ammonia flow

rate of 30 % but there was no much difference of the refractive index in both cases. The

thickness variation along the boat was optimized to 15.1 % in the case of silicon nitride

and the deposited layers on the wafers were mechanically stable. In the case of silicon

oxynitride the homogeneity was better than with silicon nitride since the thickness

variation along the wafer with silicon oxynitride was 0.83 %. It is found that the higher

the oxygen flow rate the higher the deposition rate and the lower the refractive index of

the silicon oxynitride.

An optical microsensor based on silicon microelectronics technology was

developed. It makes use of an integrated Mach-Zehnder interferometer configuration

fabricated with a silicon oxynitride waveguide. The main function of this sensor was the

detection of ammonia gas. The sensor consisted of an integrated MZI and an integrated

titanium heater with aluminum contacts. The titanium heater was integrated near one

arm from the two arms of MZI (the reference arm) and on the second arm (sensing

arm), a sensing window was created. A chemo-optical sensitive material to ammonia

(PMMA-BCP) was spin coated on the sensing window in the sensing arm.

Due to total internal reflection at the waveguide interfaces, light coupled into the

waveguide core is confined and guided. The coupled light processes an evanescent field

distribution that decays exponentially into both the substrate and cover layers.

Interaction of a target analyte with the chemo-optical selective coating on the

sensing window produces a phase change in this guided optical wave. The phase delay

introduced to the guided wave as a result of interaction with a target analyte is easily

detected by means of an optical circuit known as an interferometer such as Mach-

Zehnder interferometer.

For sensing applications, waveguide structure must verify two conditions: it has to

be monomode and it should show a high surface sensitivity. To obtain a single mode

behavior, large attenuation losses for the higher order modes must be done by

calculating the parameters of the waveguide carefully using beam propagation method

5 Conclusions

108

(BPM). Concerning the second condition, sensitivity increases by increasing the

strength and the distribution of the evanescent field. In this work, designing the

monomode silicon oxynitride waveguide was done using imaginary distance beam

propagation method (ID-BPM).

The planar monomode MZI waveguide structure was fabricated using standard

technology (Si / SiO2 / SiON / TEOS) and had been designed in order to obtain a high

intensity of the evanescent field close to the waveguide surface (high surface

sensitivity).Silicon oxynitride with low optical losses can be made with good uniformity

and reproducibility of refractive index and layer thickness.

As a sensitive layer, a PMMA film is doped with bromocresol purple (BCP), an

indicator dye, that causes the index of refraction of the film to vary with the amount of

ammonia that diffuses into the film from the ambient air.

This thesis was developed in five chapters which covered a literature review on

the low pressure chemical vapor deposition and its application in optical waveguide

based chemical sensors. The first chapter was an introduction. The second chapter was

the theoretical considerations concerning the design of the sensor. The third chapter was

optimization of low pressure chemical vapor deposition of silicon nitride and silicon

oxynitride. The fourth chapter was concerning the sensor fabrication and the

experimental results. The last chapter was the conclusions of this study.

The integrated optical device used here may serve as a key component of

integrated optical sensors, and can be produced at low cost. By directly attaching the

light source and the detectors to the integrated optical device, a very compact sensing

element will be obtained. Alternatively, optical fiber may be coupled to the input port of

the device. In this case, the electronic and optoelectronic components can be placed at

some distance from the actual sensor head.

The sensor can be used as low cost component of a distributed optical network of

chemical sensors for monitoring presence of hazardous air pollutants in the exhaust of

aircraft / spacecraft propulsion.

In this sensor, ammonia concentration below 2.4 % could not be detected by this

sensor while the response and recovery times were in the range of 120 and 240 sec

respectively.

By making use of larger numbers of dye molecules per unit volume of polymer

will result in a stronger change in the refractive index and correspondingly in more

phase change between the two light beams of MZI for the same concentration of

ammonia and hence increase the sensitivity. The sensitivity of the sensor may further be

improved by increasing the interaction length.

5 Conclusions 109

This sensor has a very important advantage, that this sensor can be used to detect

different chemical gases by removing the sensitive material after measurements, and

spin coating another material which can be used to detect another gas. Therefore, the

device can be used again for other applications.

As a future work, more work is needed to enhance the sensor sensitivity and

response.

5 Conclusions

110

Abbreviations 111

Abbreviations

Ar Argon

BCP Bromocresol purple

BPM Beam propagation method

CHF3 Trifluoromethane

C2 H16 Si Triethylsilane

(C2 H5)4 SiO4 Tetraethylorthosilicate

CVD Chemical vapor deposition

DCS Dichlorosilane

DI De-ionized

EIM Effective index method

Fe2O3 Iron oxide

FD-BPM Finite difference beam propagation method

FE-BPM Finite element beam propagation method

FET Field-effect transistor

FFT-BPM Fast Fourier transform beam propagation method

He-Ne Helium-neon

H2O2 Hydrogen peroxide

H3PO4 Phosphoric acid

HNO3 Nitric acid

IC Integrated circuits

ID-BPM Imaginary distance beam propagation method

InP Indium phosphide

IO Integrated optics

IO MZI Integrated optical mach-zehnder interferometer

IPA Isopropanol

ISFET Ion sensitive field effect transistor

LiNbO3 Litium niobate

LOCOS Local-oxidation-of-silicon

LPCVD Low pressure chemical vapor deposition

MIBK Methylisobutyl ketone

Abbreviations

112

MOSFET Metal-oxide-semiconductor field-effect transistor

MZI Mach-zehnder interferometer

NH3 Ammonia

NH4OH Ammonium hydroxide

O2 Oxygen

OIC Optical integrated circuits

OLIVE Oil-latching interfacial-tension variation effect

PECVD Plasma enhanced chemical vapor deposition

PML Perfect matched layer

PMMA Polymethyl methacrylate

PMMA-BCP Polymethyl methacrylate-bromocresol purple

RIE Reactive ion etching

SAFET Surface accessible field effect transistor

SAW Surface acoustic wave

SGFET Suspended gate field effect transistor

Si Silicon

Si3N4 Silicon nitride

SiO2 Silicon dioxide

SiON Silicon oxynitride

SnO2 Tin oxide

SOI Silicon-on-insulator

TCR Temperature coefficient of resistance

TE Transverse electric

TEOS Tetraethylorthosilicate

TES Triethylsilane

TIR Total internal reflection

TM Transverse magnetic

TO Thermal oxidation

UV Ultraviolet

References 113

References

[1] G. T. Roman, “Process characterization of LPCVD silicon nitride and the

consequential fabrication of low stress microcantilevers”, CNF, Cornell University,

2003.

[2] http://www.timedomaincvd.com/CVD_Fundamentals/films/SiN_properties.html,

02.05.2007.

[3] http://galileo.phys.virginia.edu/outreach/8thgradesol/InternalReflectionFrm.htm,

20.03.2007.

[4] P. V. Lambeck, “Integrated optical sensors for the chemical domain”, Meas. Sci.

Technol. 17, PP. R93 – R116, 2006.

[5] http://www.monos.leidenuniv.nl/smo/index.html?basics/light.htm, 07.01.2007.

[6] R. Ulrich, “Theory of the prism–film coupler by plane wave analysis”, J. Opt. Soc.

Am., 1970.

[7] T. Tamir and S. T. Peng, “Analysis and design of grating couplers”, Appl., Phys.,

Vol. 14, PP. 235 – 254, 1977.

[8] O. Mitomi, K. Kasaya and H. Miyazawa, “Design of a single–mode tapered

waveguide for low loss chip–to–fiber coupling”, IEEE, J. Quantum Electron, Vol. 30,

PP. 1787 – 1793, 1994.

[9] W. Göpel and J. N. Zemel, “Sensors: A comprehensive survey in chemical and

biochemical sensors”, Weinheim, Germany: VCH-Verlagsgesellschaft, vol. 2/3, 1991.

[10] J. Janata, “Principles of chemical sensors”, New York: Plenum, 1989.

[11] J. Janata, and R. J. Huber, “Solid state chemical sensors”, San Diego, CA:

Academic, 1985.

[12] A. Hierlemann, O. Brand, C. Hagleitner and H. Baltes, “Microfabrication

techniques for chemical / biosensors”, IEEE, Vol. 91, No. 6, PP. 839 – 863, June 2003.

[13] A. L. Spetz, “Chemical sensor technologies”, S-SENCE / IFM, Linköping

University, Sweden, 2006.

[14] R. P. H. Kooyman and L. M. Lechuga, “Immunosensors based on total internal

reflectance”, in handbook of biosensors and electronic noses, E. Kress-Rogers. Ed.

Boca Raton, FL: CRC, 1997.

[15] P. V. Lambeck, “Sensors actuators”, B 8, PP. 108 – 116, 1992.

[16] H. Porte, V. Gorel, S. Kiryenko, J. P. Goedgebuer, W. Daniau and P. Blind,

“Imbalanced Mach-Zehnder Interferometer in micromachined silicon substrate for

pressure sensor”, Journal of Lightwave Technology, Vol. 17, No. 2, PP. 229 – 233,

February 1999.

[17] F. Prieto, B. Sepulveda, A. Calle, A. Llobera, C. Dominguez, A. Abad, A. Montoya

and L. M. Lechuga, “An integrated optical interferometeric nanodevice based on

References

114

silicon technology for biosensor applications”, Nanotechnology 14, PP. 907 – 912,

2003.

[18] F. Prieto, L. M. Lechuga, A. Calle, A. Llobera, and C. Dominguez, “Optimized

silicon antiresonant reflecting optical waveguides for sensing applications”, Journal of

Lightwave Technology, Vol. 19, No. 1, PP. 75 – 83, January 2001.

[19] P. Äyräs, S. Honkanen, K. M. Grace, K. Shrouf, P. Katila, M. Leppihalme, A.

Tervonen, X. Yang, B. Swanson and N. P eyghambarian, “Thin film chemical sensors

with waveguide Zeeman interferometery”, Pure Appl., Opt. 7, PP. 1261 – 1271, 1998.

[20] J. F. McAleer, P. T. Moseley, J. O. Norris, and D. E. Williams, “Tin dioxide gas

sensors”, Journal Chem. Soc., Faraday Trans. I; Vol. 83, PP. 1323-1346, 1987.

[21] P. N. Bartlett, P. B. M. Archer and S. K. Ling-Chung, “Conducting polymer gas

sensors, part 1: fabrication and characterization”, Sensors and Actuators B, vol. 4, PP.

365-372, 1991.

[22] Y. Nakatani, M. Matsuoka and Y. Iida, “

γ

-Fe2O3 ceramic gas sensor”, IEEE

Trans. on Components, Hybrids and Manufacturing Tech., Vol. CHIMT-5, No. 4,

December 1992.

[23] K. Ihokura and J. Watson, “The stannic oxide gas sensor: principles and

applications”, CRC Press: Ann Arbor, Michigan, PP. 2-7, 1994.

[24] P. Bergveld, “Development of an ion-sensitive solid-state device for

neurophysiological measurements,” IEEE Trans Biomedical Eng., Vol. 17, PP. 70-71,

January 1970.

[25] K. I. Lundstrom, M. S. Shivaraman and C. M. Svenson, “A hydrogen-sensitive Pd

gate MOS transistor”, Journal App Physics, Vol. 46, No. 9, PP. 3876-3881, September

1975.

[26] M. Stenberg and B. Dahlenback, “Surface-accessible FET for gas Sensing”,

Sensors and Actuators, Vol. 4, PP. 273-281, 1983.

[27] M. Joseowics and J. Janata, “Suspended gate field effect transistor modified with

polypyrrole as alcohol sensor”, Anal Chem., Vol. 58, PP. 514-517, 1986.

[28] I. Lundstrom and C. Svensson, “Gas-sensitive metal gate semiconductor devices”,

Solid Sate Chemical Sensors, Jiri Jonata and Robert J. Huber, Eds, Academic Press:

New York, PP. 2-50, 1985.

[29] H. D. Wiemhofer and W. Gopel, “Fundamentals and principles of potentiometric

gas sensors based upon solid electrolytes”, Sensors and Actuators B, Vol. 4, PP. 365-

372, 1991.

[30] G. Hoetzel and W. Weppner, “Potentiometric gas sensors based on fast solid

electrolytes”, Sensors and Actuators, Vol. 12, PP. 449-453, 1987.

[31] J. Janata, “Principles of chemical sensors”, Plenum Press: New York, Chapters 2,

V3, and 5, 1989.

[32] C. Nylander, “Chemical and biological sensors”, J. Phys. E: Sci. Instrum., Vol.

18, PP. 736 – 750, 1985.

References 115

[33] T. Koster and P. V. Lambeck, “An integrated optical platform for absorptive

sensing of chemical concentrations using chemo-optical monolayers”, J. Meas., Sci.,

Technol., PP. 1230 – 1238, 2002.

[34] J. Tschmelak, G. Proll and G. Gauglitz, “Optical biosensor for pharmaceuticals,

antibiotics, hormones, endocrine disrupting chemicals and pesticides in water: assay

optimization process for estrone as example”, Talanta, Elsevier Science BV, PP. 313 –

323, 2005.

[35] K. Tiefenthaler and W. Lukosz, “Sensitivity of grating couplers as integrated-

optical chemical sensors”, J. Opt., Soc., Am., Vol. 6, No. 2, PP. 209 – 216, 1989.

[36] O. Parriaux and G. Veldhuis, “Normalized analysis for the sensitivity optimization

of integrated optical evanescent wave sensors”, J. Lightwave Technol., PP. 573 – 582,

1998.

[37] G. Veldhuis, O. Parriaux, H. J. W. M. Hoekstra and P. V. Lambeck, “Sensitivity

enhancement in evanescent optical waveguides sensors”, J. Lightwave Technol., PP.

677 – 682, 2000.

[38] J. V. Lith, “Novel integrated optical sensing platforms for chemical and immuno-

sensing”, PhD Thesis, Uni. of Twente, Enschede, The Netherlands, 2005.

[39] J. Comley, “Label-free detection; new biosensors facilitate broader range of drug

discovery applications”, Drug Discovery World winter, 2004.

[40] C. Malins, A. Doyle, B. Maccraith, F. Kvasnik, M. Landl, P. Simon, L. Kalvoda, R.

Lukas, K. Pufler and I. Babusik, “Personal ammonia sensor for industrial

environments”, J. Environ. Monit., PP. 417 – 422, 1999.

[41] F. Ay, “Silicon oxynitride layers for applications in optical waveguides”, Master

Thesis, Bilkent Uni., September 2000.

[42] G. J. Valdhuis, “Bent-waveguide devices and mechano-optical switches”, PhD

thesis, Uni. of Twente, 1998.

[43] S. P. Pogossian, L. Vescan, and A. Vonsovici, “The single-mode condition for

semiconductor rib waveguides with large cross section”, Journal of Lightwave

Technology, Vol. 16, No. 10, PP. 1851 – 1853, Oct. 1998.

[44] S. S. A. Obayya, B. M. Azizur, K. T. V. Grattan, and H. A. El-Mikati, “Full

vectorial Finite-Element based imaginary distance beam propagation solution of

complex modes in optical waveguides”, Journal of Lightwave Technology, Vol. 20, No.

6, PP. 1054 – 1060, June 2002.

[45] J. Yamauchi, G. Takahashi, and H. Nakano, “Full-Vectorial beam propagation

method based on the McKee-Mitchell scheme with improved finite difference formulas”,

Journal of Lightwave Technology, Vol. 16, No. 12, PP. 2458 – 2464, Dec. 1998.

[46] T. Fujisawa and M. Koshiba, “Full vector Finite-Element beam propagation

method for three dimensional nonlinear optical waveguides”, Journal of Lightwave

Technology, Vol. 20, No. 10, PP. 1876 – 1884, Oct. 2002.

[47] J. Shibayama, T. Takahashi, J. Yamauchi, and H. Nakano, “Efficient time domain

finite difference beam propagation methods for the analysis of slab and circularly

symmetric waveguides”, Journal of Lightwave Technology, Vol. 18, No. 3, PP. 675 –

687, March 2000.

References

116

[48] Y. Tsuji, M. Koshiba, and N. Takimoto, “Finite Element beam propagation method

for anisotropic optical waveguides”, Journal of Lightwave Technology, Vol. 17, No. 4,

PP. 723 – 728, April 1999.

[49] Y. Tsuji and M. Koshiba, “Guided-Mode and Leaky-Mode Analysis by Imaginary

Distance Beam Propagation Method Based on Finite Element Scheme”, Journal of

Lightwave Technology, Vol. 18, No. 4, PP. 618 – 623, 2000.

[50] O. Parriaux and P. Dierauer, “Normalized expressions for the optical sensitivity

of evanescent wave sensors: erratum”, Optics Letters, Vol. 19, No. 20, PP. 508 – 510,

1994.

[51] http://www.imm.cnm.csic.es/biosensores/mzi.htm, 23.04.2007.

[52] O. Parriaux, “Fiber optic chemical sensors and biosensors”, Italy, O. Wolfbeis,

Ed. (CRC, Boca Raton), Vol. 1, PP. 111-192, 1991.

[53] W. Lukosz, “Integrated optical nanomchanical devices as modulators, switches

and tunable wavelength filters as acoustical sensors”, Proc., SPIE, Vol. 3, PP. 402 -

406, 1992.

[54] J. Angel-Valenzuela, A. Sanchez, R. Cardoso, L. Villegas-Vicencio, D. Salazar

and H. Marquez, “A fabry-perot-type integrated optical temperature transducer”,

Instrumentation and Development, Vol. 5, No. 3, PP. 184 – 189, 2001.

[55] E. Camargo, H. Chong and R. Rue, “2D photonic crystal thermo-optic switch

based on AlGaAs/GaAs epitaxial structure”, Optics Express, Vol. 12, No. 4, PP. 588 –

592, 2004.

[56] T. Aalto, M. Kapulainen, S. Yliniemi, P. Heimala and M. Leppihalme, “Fast

thermo-optical switch based on SOI waveguides”, Integrated Optics, Vol. 4987, PP.

149 – 159, 2003.

[57] http://ece-www.colorado.edu/~bart/book/ellipsom.htm, 01.05.2007.

[58] S. Sarkisov, D. Diggs, G. Adamovsky and M. Curley, “Single-arm double-mode

double-order planar waveguide interferometric sensor”, Applied Optics, Vol. 40, No. 3,

PP. 349 – 359, 2001.

[59] C. Wu, P. Lin, R. Huang, W. Chao and M. Lee, “Design optimization for

micromachined low power Mach-Zehnder thermo-optic switch”, Applied Physics

Letters 89, 121121, 2006.

[60] G. Treyz, “Silicon Mach-Zehnder waveguide interferometer operating at 1.3 μm”,

Electron. Lett. 27, 118-120, 1991.

List of Figures 117

List of Figures

1.1 Block diagram of a LPCVD system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.2 Total internal reflection of light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.3 Different light rays with different incidence angles are arriving

to the boundary of two mediums from a higher refractive index medium

to a lower refractive index medium where the total internal reflection

occurs when the angle of incidence is greater than the critical angle . . . . . . . 8

1.4 Multiple total internal reflection in an optical fiber . . . . . . . . . . . . . . . . . . . . 8

1.5 The rib waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.6 The channel waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.7 The strip loaded waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.8 Schematic cross-section of waveguide structures with the location of

the mode in each case (a) Rib waveguide (b) Strip loaded waveguide . . . . . 11

1.9 The propagation of light as arunning wave of electromagnetic field . . . . . . 12

1.10 Prism coupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.11 Grating coupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.12 Basic structure of the thin-film sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1. 13 ISFET (ion selective FET) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1.14 ChemFET Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

1.15 SAFET structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

1.16 SGFET structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

1.17 Fraction F, of the sites which are associated with a target molecule as a

function of the concentration c of the target molecules . . . . . . . . . . . . . . . . . 31

1.18 Evanescent field penetrates the cover layer . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

1.19 Mach Zehnder interferometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.1 Schematic of a rib waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.2 Monomode condition for SiON waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.3 Optical waveguide is surrounded by PML regions . . . . . . . . . . . . . . . . . . . . 39

List of Figures

118

2.4 Schematic representation of a slab waveguide refractometric sensor . . . . . 43

2.5 Schematic representation of a surface sensor . . . . . . . . . . . . . . . . . . . . . . . . 47

2.6 Schematic cross section and target design parameter of a strip-loaded

waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

2.7 Mach-Zehnder interferometer (MZI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

2.8 Plot of the transfer function of the MZI versus the phase difference

ϕ

Δ

. . . . 52

2.9 Top view of the basic configuration of an IO MZI . . . . . . . . . . . . . . . . . . . . . 54

2.10 A top view of the titanium heater and the aluminum contacts . . . . . . . . . . . . 57

2.11 The heater as one part . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

2.12 Optical waveguide based chemical sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

3.1 3-zone-oven, is used to produce SiO2 by thermal oxidation, TEOS

decomposition and SiON layers by LPCVD . . . . . . . . . . . . . . . . . . . . . . . . . . 60

3.2 The carrier (1) and the boat (2) used to carry the wafers into the tube . . . . . . 61

3.3 The carrier and the boat with the wafers showing the wafers position, the

direction of gas inlet (1) and gas outlet (2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.4 The boat with the wafers just before going into the tube . . . . . . . . . . . . . . . . . 62

3.5 The quartz tube of LPCVD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

3.6 The ellipsometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

3.7 Schematic drawing of an ellipsometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

3.8 Thicknesses and refractive indices distribution at 740 °C (wafer 1) . . . . . . . . 66

3.9 Thicknesses and refractive indices distribution at 760 °C (wafer 2) . . . . . . . . 67

3.10 Thicknesses and refractive indices distribution at 780 °C (wafer 3) . . . . . . . . 68

3.11 Thicknesses and refractive indices distribution of Si3N4 deposition with

30 % flow rate of ammonia (wafer 23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

3.12 Thicknesses and refractive indices distribution of Si3N4 deposition with

15 % flow rate of ammonia (wafer 23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

3.13 The relationship between the oxygen flow rate and the refractive index

of the deposited layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

3.14 The relationship between the thickness of the deposited layer and the

deposition time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

4.1 The thickness of SiO2 layers by thermal oxidation at 960 °C / 95 °C . . . . . . . . 82

4.2 The thickness of TEOS layers by using LPCVD with different time . . . . . . . . . 84

List of Figures 119

4.3 Representative images of cross-section of the waveguide revealing the

resultindg each profile. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

4.4 Reactive ion etching (RIE) in TEOS layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

4.5 Fabrication process flow for high-refractive index contrast

SiON/SiO2 waveguide structures; (a) Grown of SiO2 layer by using

thermal Oxidation, (b) Depositing of SiON layer by using

LPCVD, (c) Depositing of TEOS layer, (d) Putting the photoresist

Layer by using spin coating, (e) Defining the waveguide pattern

by using optical contact lithography, (f) Forming the rib in the

TEOS by using reactive ion etching (RIE) and (g) Removing the

mask and making cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.6 (a) Represents a cross-section image for the waveguide (b) Represents

an image for the splitting point of the MZI waveguide . . . . . . . . . . . . . . . . . . . 90

4.7 A microscopic image for the titanium heater with an aluminum contact . . . . 92

4.8 A top view for the structure consisting of MZI, the heater and the

Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

4.9 A view showing the structure of the sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

4.10 The film thickness vs. spin-speed for sensitive layer . . . . . . . . . . . . . . . . . . . . 94

4.11 Experimental setup for coupling light to the waveguide by using

end fire method (a) sketch (b) image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

4.12 The normalized intensity vs. the horizontal direction . . . . . . . . . . . . . . . . . . . 97

4.13The resistance ratio (R/R0) as a function of temperature difference

(T – T0) for titanium heater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

4.14 Resistance of the heater against the dissipated power . . . . . . . . . . . . . . . . . 99

4.15 Transmission intensity against electrical heating power . . . . . . . . . . . . . . . . 100

4.16 The refractive index change in the sensitive layer during exposure to

ammonia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

4.17 The experimental setup used for ammonia sensing measurements . . . . . . . . 102

4.18 Calibration plot for the sensor response represented by the output

intensity as a function of ammonia concentration . . . . . . . . . . . . . . . . . . . . . 103

4.19 The phase change

ϕ

Δ as a function of the change in the refractive

Index change of the sensitive layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

4.20 The sensor response as a function of response time . . . . . . . . . . . . . . . . . . . 104

List of Figures

120

List of tables 121

List of tables

2.1 The definition of sx and sy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.1 Cleaning procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.2 Process parameters for silicon nitride deposition at 740 °C . . . . . . . . . . . . . . 65

3.3 Process parameters for silicon nitride deposition at 760 °C . . . . . . . . . . . . . . 66

3.4 Process parameters for silicon nitride deposition at 780 °C . . . . . . . . . . . . . . 67

3.5 The results for depositions of wafers (4, 5 and 6)in the same boat

at 780 °C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

3.6 The results for depositions of wafers (7, 8 and 9)in the same boat

at 760 °C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3.7 Deposition of wafers (10, 11 and 12) at 760 °C with changing the

Temperatures of the tube to [front: 720 °C , middle: 753 °C and

back: 759 °C ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

3.8 Deposition of wafers (13, 14 and 15) at 760 °C with changing the

Temperatures of the tube to [front: 732 °C , middle: 754 °C and

back: 759 °C ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

3.9 Deposition of wafers (16, 17 and 18) at 760 °C with changing the

Temperatures of the tube to [front: 750 °C , middle: 756 °C and

back: 760 °C ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

3.10 Deposition of wafers (19, 20, 21 and 22) at 760 °C with changing the

Temperatures of the tube to [front: 750 °C , middle: 756 °C and

back: 760 °C ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

3.11 Process parameters for Si3N4 deposition with ammonia flow rate

of 30 % . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

3.12 Process parameters for Si3N4 deposition with ammonia flow rate

of 15 % . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

3.13 Process parameters used in SiON deposition (O2 flow rate is 0.178 %) . . . . 77

3.14 The results of SiON deposition with O2 flow rate of 0.178 % . . . . . . . . . . . . 77

3.15 The results of deposition processes of SiON on wafers (2 – 6) with

decreasing the oxygen flow rate with each wafer. . . . . . . . . . . . . . . . . . . . . . . 78

List of tables

122

3.16 The results of deposition processes of SiON on wafers (7 – 11) with

decreasing the oxygen flow rate with each wafer. . . . . . . . . . . . . . . . . . . . . . . 79

4.1 Process parameters 3used in thermal oxidation . . . . . . . . . . . . . . . . . . . . . . . . . 82

4.2 Process parameters for SiON deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

4.3 Process parameters used in TEOS deposition . . . . . . . . . . . . . . . . . . . . . . . . . . 83

4.4 Process parameters used in spin coating and photolithography for TEOS layer 84

4.5 Process parameters used in RIE for TEOS layer (the upper cladding) . . . . . . . 86

4.6 Process parameters of titanium evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

4.7 The materials which were used in titanium etching . . . . . . . . . . . . . . . . . . . . . . 91

4.8 Process parameters of aluminum evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4.9 The materials which were used in aluminum etching . . . . . . . . . . . . . . . . . . . . 92

4.10 Process parameters used in fabricating PMMA-BCP sensitive layer . . . . . . . 95