Introducing Photonics

The essential guide for anyone wanting a quick introduction to the fundamental ideas
underlying photonics, the science of modern optical systems. The author uses his
40 years of experience in photonics research and teaching to provide intuitive explan-
ations of key concepts and demonstrates how these relate to the operation of photonic
devices and systems. The reader can gain insight into the nature of light and the ways in
which it interacts with materials and structures and learn how these basic ideas are
applied in areas such as optical systems, 3D imaging and astronomy. Carefully designed
end-of-chapter problems enable students to check their understanding. Hints are given
at the end of the book and full solutions are available online. Mathematical treatments
are kept as simple as possible, allowing the reader to grasp even the most complex of
concepts. Clear, concise and accessible, this is the perfect guide for undergraduate
students taking a first course in photonics and anyone in academia or industry wanting
to review the fundamentals.

B R I A N C U L S H A W is an Emeritus Professor of Optoelectronics at Strathclyde Univer-

sity, having previously served as Head of Department and Vice Dean. He is also a
Director of OptoSci Ltd and was the 2007 President of the International Society for
Optics and Photonics (SPIE).
“The field of photonics is the study of the generation, manipulation, and detection of
electro-magnetic radiation (which we call light) whose energy is carried by what are
known as photons. Professor Brian Culshaw has written a non-mathematical
textbook that introduces the fundamentals of how light interacts with matter and
reviews many of the diverse applications of optical systems in everyday life. This
book is thus written at a level appropriate for undergraduate science or engineering
students. Although advanced mathematical concepts are invoked (such as imaginary
numbers and complex integration), mathematical derivations are avoided and simple
explanations are used throughout the book. The author reviews the interactions
between light and matter, including both absorption and emission of light, and then
considers how the length scale of structures or features in matter can influence this
interaction. Finally, the highlight of Professor Culshaw’s book is a review of the
many applications of photonics, ranging from lighting, digital communication, to
even metal welding and cutting. For example, students are often surprised to learn
that the internet and its functionality is enabled by the same form of energy that
allows us to illuminate our homes after dark. Hence Introducing Photonics manages
to successfully convey the diversity, excitement, and impact of photonic
applications in modern society, without the mathematical details.”
Kent D. Choquette, University of Illinois at Urbana-Champaign
“The high point of this inspiring little book is the 40-odd stimulating and thought-
provoking questions at the end of each chapter.”
Chris Dainty, National University of Ireland
“Brian Culshaw of Strathclyde University has been a long-time contributor to both
photonics and ultrasonics. This book, Introducing Photonics, shares with the reader
some of the knowledge about photonics the author has accumulated over his long
career. The book contains much more on the optical properties of materials than
many other books on this subject. I recommend the book to anyone who would like
to benefit from the author’s experience teaching photonics and performing cutting-
edge research in this field.”
Joseph Goodman, Stanford University
“The book that we were waiting for! The book that only a few very experienced
people world-wide are capable to write! The book that we were waiting for to
understand and/or clarify key photonic concepts in very easy, efficient and effective
way! A key book for beginnings into the science and technology of light
(photonics). It is an excellent tool for undergraduate students that aim to get a rapid,
clear and effective view of photonic essential concepts, that also works for anyone to
“refresh”, efficiently, the fundamentals and technologies of light!”
Jose Miguel Lopez Higuera, University of Cantabria
“This book aims to introduce undergraduates to photonics and does it in a different
and interesting format compared to typical textbooks. The author discusses many
diverse topics in photonics together with achievements made possible by today’s
optical technology and what will be possible in the future photonics world. This
book will encourage students to explore the ever-expanding field of photonics.”
David A. Jackson, University of Kent
“Professor Brian Culshaw has been dealing with these issues for over 40 years,
hence his knowledge transfer is extremely intuitive, deeply thought out not only
from a scientific point of view but, above all, from the point of view of the recipient
for whom photonics may be a new challenge. The minimal mathematical apparatus
present in the book allows the reader to focus on an intuitive understanding of the
discussed issue, thus creating an open path for him to search for new own
applications. The concepts of ending each chapter with a set of problems for
individual decision by the reader should be accepted as extremely valuable here.
I think that their correct resolution is a challenge not only for the main recipients of
the book - undergraduate students, but also for a wide range of active scientists
involved in photonics. Hence, the book Introducing Photonics is an extremely
valuable textbook appearing on the market.”
Leszek R. Jaroszewicz, Military University of Technology
“Introducing Photonics is an enjoyable and thought provoking read for all with an
interest in this field. This book is an excellent means to augment your existing
knowledge and offers stimulating viewpoints on photonics from a long-standing
expert in this area”
Kyriacos Kalli, Cyprus University of Technology
“An outstanding book that grasps the essence of photonics. Important concepts,
from basic ideas to applications and future possibilities, are perfectly explained with
easy language, clear descriptions and concise illustrations. University students in
photonics as well as the more mature professionals in related fields will benefit from
reading it.”
Wei Jin, The Hong Kong Polytechnic University
“Professor Culshaw has written a must-read prime text for anyone considering
entering the exciting field of modern-day photonics. . . all bases are covered. . . a
comprehensive treatment from a fundamental introduction of the concepts behind
light theory through to future photonic technologies and trends.”
Elfed Lewis, University of Limerick
“This book provides a detailed explanation of the whole of the subject. It presents a
new approach, concentrating on physical ideas, and developing them to explain
more complex phenomena. This book is essential reading for practitioners,
researchers, or students of physics and engineering whose work involves any of the
large number of application areas touched by Introducing Photonics.”
Yanbiao Liao, Tsinghua University
“The refreshing, intuitive style of this book makes it an excellent first text on light-
based science and technology. Without sacrificing rigor, Introducing Photonics
makes the field easily approachable; students with little more than basic science
training will find this an enlightening and enjoyable text. Thought-provoking
exercises challenge one to solve complete problems, rather than just use formulas to
arrive at numerical answers by rote. In the final chapters, a sweeping panorama of
current and future applications and innovations inspires the reader with the
excitement and promise of the burgeoning field covered by this book: photonics,
‘the electronics of the 21st Century’.”
R. A. Lieberman, Lumopix, LLC
“We are just at the beginning of the age of light, the most sustainable and
challenging age ever. Nothing like photonics has already changed and is still going
to change for centuries our way of living, interacting, communicating, moving,
caring and feeding ourselves. This book is a valuable and timely guide to review the
key concepts of photonics and related applications through intuitive explanations
and straightforward problems to check the understanding. From the basics of light-
matter interaction, the reader is guided through consolidated photonic tools to the
most advanced and intriguing devices, envisioning that the best of photonics has yet
to come.”
Anna Grazia Mignani, National Research Council (CNR), Italy
“Drawing on his vast experience in photonics research and innovation, Brian
Culshaw presents the fundamental principles of photonics in an accessible manner,
with end-of-chapter problems to reinforce the learning objectives. This is certain to
be the go-to book for all students starting out in the field of photonics.”
Sinéad O’Keeffe, University of Limerick
“Introducing Photonics is a surprising book! The profound and appealing way in
which fundamental concepts are presented and what follows from them on
applications in a multitude of fields of optics and photonics is remarkable, as well as
the elegance and proximity with which the text is written.
At the end of each chapter a set of enlightening problems is proposed that follow
the same framework, that is, its purpose is to induce thought and reflection on a
particular phenomenon in the field of optics and photonics, as well as possible
technological applications that may result from it.
The book is also prospective in a crystalline way regarding the future of photonics
in its multiple areas, revealing from his author a deep knowledge and insight of the
field.
In summary, I consider Introducing Photonics an inestimable book to guide and
inspire a generation of students and practitioners in the field of optics/photonics.”
José Luís Santos, University of Porto
“Brian Culshaw, in Introducing Photonics, succeeds in presenting a wide range of
important topics in a clear, concise and effective manner that will be welcomed by
students and technical workers. The linkages to real world applications throughout
the text, problems and recommended readings serve as an effective entry point into
this important field that has changed the world.”
Eric Udd, Columbia Gorge Research, LLC
Introducing Photonics

BRIAN CULSHAW
University of Strathclyde
 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781107155732
DOI: 10.1017/9781316659182
© Cambridge University Press 2020
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2020
Printed in the United Kingdom by TJ International Ltd, Padstow, Cornwall
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Culshaw, B., author.
Title: Introducing photonics / Brian Culshaw.
Description: Cambridge ; New York, NY : Cambridge University Press, 2020. |
Includes bibliographical references and index.
Identifiers: LCCN 2019036970 (print) | LCCN 2019036971 (ebook) | ISBN 9781107155732
(hardback) | ISBN 9781316609415 (paperback) | ISBN 9781316659182 (epub)
Subjects: LCSH: Photonics.
Classification: LCC TA1520 .C85 2020 (print) | LCC TA1520 (ebook) |
DDC 621.36/5–dc23
LC record available at https://lccn.loc.gov/2019036970
LC ebook record available at https://lccn.loc.gov/2019036971

ISBN 978-1-107-15573-2 Hardback

ISBN 978-1-316-60941-5 Paperback
Additional resources for this publication at www.cambridge.org/culshaw.
Cambridge University Press has no responsibility for the persistence or accuracy
of URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
 Contents

Preface page xi

1 Photonics – An Introduction 1
1.1 What is Photonics? 1
1.2 Exploring Some Concepts 3
1.3 Applications 5
1.4 About This Book 5
1.5 Some Related Background Ideas 8
1.6 Problems 9
2 The Nature of Light 11
2.1 Light in a Vacuum 11
2.2 Light in Isotropic and Anisotropic Materials 12
2.3 Light Interacting with Structures 19
2.4 Summary 23
2.5 Problems 23
3 Light Interacting with Materials 25
3.1 Linear Optical Materials 25
3.2 Non-Linear Optical Materials 29
3.3 The Diversity of Optical Absorption 33
3.4 Light Emission 43
3.5 Materials with Controllable Optical Properties 51
3.6 Summary 54
3.7 Problems 55
4 Light Interacting with Structures 58
4.1 Structures with Dimensions Much Greater than the Optical
Wavelength 58

vii
viii Contents

4.2 Structures with Dimensions of the Order of the Optical

Wavelength 64
4.3 Structures with Dimensions Much Smaller than
the Optical Wavelength 77
4.4 Summary 81
4.5 Problems 82
5 Photonic Tools 85
5.1 Optical Systems 85
5.2 Spectrometers 88
5.3 Display Technologies 90
5.4 Images and Imaging 91
5.5 Three-Dimensional Imaging 91
5.6 Precision Measurement 93
5.7 Optical Fibres and Communication Systems 96
5.8 Lasers in Machining and Surgery 98
5.9 Environmental Analysis 99
5.10 Light as a Therapy 102
5.11 Astronomy and Very Large Telescopes 102
5.12 Optical Tweezers 104
5.13 Summary 106
5.14 Problems 106
6 The Future 109
6.1 Super-Resolution Imaging 110
6.2 Photonic Integrated Circuits 113
6.3 Nanophotonics and Plasmonics 115
6.4 Metamaterials 117
6.5 Photonics and Materials – A Brief Summary 119
6.6 Entangled Photons 120
6.7 Graphene and Other Exotic Materials 122
6.8 Manipulating Light Sources 124
6.9 Summary 126
6.10 Problems 127

Appendix 1 More About the Polarisation of Waves 129

Appendix 2 The Fourier Transform Properties of Lenses 132
 Contents ix

Appendix 3 The Clausius–Mossotti Equation and the Frequency

Dependence of the Refractive Index 136
Appendix 4 The Concepts of Phase and Group Velocity 139
Appendix 5 Some Fundamental Constants 142
Appendix 6 Comments and Hints on Chapter Problems 143
Further Reading 157
Index 161
 Preface

Light has immediate and obvious appeal – the vast majority of us see light and
communicate through it daily. We process images – facial expressions and
traffic signs. We know light has an impact on people and places, emotionally
and physically. Different colours affect our moods, sunshine lifts the spirit but
with side effects for the excessively indulgent.
The word ‘photonics’ emerged into the scientific and technical vocabulary
towards the end of the twentieth century and has come to mean much more than
simply visible light extending into the redder than red and the bluer than blue.
Photonics is the science of light. It is the technology of generating, controlling,
and detecting lightwaves and photons, which are particles of light. The characteris-
tics of the waves and photons can be used to explore the universe, cure diseases and
even to solve crimes. Scientists have been studying light for hundreds of years. The
colours of the rainbow are only a small part of the entire lightwave range called the
electromagnetic spectrum. Photonics explores a wide variety of wavelengths, from
gamma rays to radio including x-rays, ultraviolet, and infrared light.
Photonics is also gaining much respect within the technical community;
indeed 2015 was the International Year of Light and from 2018 onwards the
International Day of Light will be marked on 21 May. But what makes photon-
ics special? Indeed, what is it and how does it contribute to our emerging world?
These questions are easy to answer. Fibre optic communications are critical to
the success of the internet, the biomedical community uses ever more versatile
visible imaging systems exploiting subtle photonic concepts and phototherapy is
now routine. Mobile displays and digital cameras are ubiquitous and there is
much, much more to come.
The aim of this book is primarily to provide answers to the above two
questions in concise and accessible manner. Yes, there are one-thousand-page
tomes full of mathematics which expound every detail of the scientific
principle. However, here we seek to highlight basic concepts and ideas and,
most of all, attempt to develop some intuitive insight into what is going on.
xi
xii Preface

Such insights, whilst not necessarily always totally accurate, are essential to
understanding and formulating ideas and making that essential first guess
(rough estimate) as to whether an idea is indeed feasible.
Some readers will be intrigued by photonics for its fascination alone as a
technical discipline. For many the thought that this fascination may evolve into
a diversity of application sectors over the coming decades will stimulate
further curiosity and encourage broader participation. The final couple of
chapters look very briefly into the diversity of prospects, both in applications
and in scientific advances, to trigger yet more application potential. One
aspiration here is to motivate yet further interest in this intriguing domain.
The book also includes problems at the end of each chapter, and the principal
aim of these is to encourage readers to dig further and discuss the possible
solutions among themselves. Indeed, many may respond best to teamwork
approaches where different readers approach different pre-agreed tasks and then
come together to evaluate a complete solution. Many of the problems are some-
what like ‘real engineering’ in that there is no perfect answer but the optimum
approach needs to be fully understood and critically assessed by the readers. There
will be discussion spaces on the web available as the book progresses.
For those wishing for more detail there are also many suggestions for further
reading; there is a great deal of good material available on the web. Indeed,
perhaps the real hope is that this book will stimulate the reader’s curiosity
towards discovering more about this fascinating topic.
Approaching and hopefully consolidating these insights has, from the
author’s perspective, emerged over many years with enormous thanks to
countless encounters with groups ranging from students to internationally
recognised experts; hopefully this book will consolidate at least the gist of
these many experiences. Any list of these encounters is too long to mention
and any attempt will inevitably have omissions.
There are also other acknowledgments to make, notably to the ever tolerant
Sarah Strange at CUP whose constructively critical and wise comments on the
manuscript have been remarkably helpful and thought provoking. All errors and
omission, though, are purely the author’s fault and comments from readers will
always be welcome. Thanks too to Elizabeth Horne and Julie Lancashire for
constructive support and understanding when yet another delay in the delivery of
the manuscript appeared. Then there was Susan Parkinson, a splendidly under-
standing copy editor. Last, but by no means least, a very big thank you to the
long suffering Aileen, who not only put up with the author in the household but
also turned the manuscript from unintelligible ramblings into (we hope) a
comprehensible text. We shall certainly celebrate when the book emerges.

Brian Culshaw
Glasgow
 1
Photonics – An Introduction

Photonics is probably the outstanding candidate for the principal enabling

technology of the twenty-first century. The term emerged around half a century
ago and is now slowly creeping into everyday vocabulary just as ‘electronics’
did in the twentieth century. The aim in this short book is to introduce the
principal concepts and ideas which underpin photonics and to explore its
current and evolving technological significance.

1.1 What is Photonics?

‘Photonics is the science of light. It is the technology of generating, controlling
and detecting lightwaves and photons, which are particles of light. The char-
acteristics of the waves and photons can be used to explore the universe, cure
diseases and even to solve crimes. Scientists have been studying light for
hundreds of years. The colours of the rainbow are only a small part of the
entire lightwave range called the electromagnetic spectrum. Photonics explores
a wider variety of wavelengths, from gamma rays to radio including X-rays,
ultraviolet and infrared light.’ This succinct definition comes from the website
of the International Year of Light (IYL) celebrated in 2015.
The concepts through which we explore and understand light are indeed
common throughout the electromagnetic wave spectrum, as indicated in the
IYL definition. However, it is straightforward to appreciate that the scale of the
wavelengths across this spectrum, extending from hundreds of metres for radio
waves to subnanometres at X-ray frequencies, implies that different features
within this common set will become more or less dominant in the understand-
ing and application of a particular part of the spectrum. The transition from
‘electronics’ to ‘photonics’ reflects this gradual transfer, as indicated in
Figure 1.1.

1
 Electronics increasingly important Photonics increasingly important
Visible

Radio Tera- Infra- Ultra- X-rays Gamma

Microwaves hertz rays
waves red violet

Medium VHF UHF Radio Astronomy Security Global Warming Sunburn Medical Astronomy
Wave TV Satellite Links Imaging Photosynthesis Imaging Mössbauer
MRI
Radar X-ray Spectroscopy
Mobile phones Astronomy
1 MHz 1 GHz 1 THz 1 PHz 1 EHz Frequency

Wavelength
1 km 1m 1 mm 1 mm 1 nm

Figure 1.1 The principal regions of the electromagnetic spectrum. The boundaries between the various regions are fuzzy (except for the tiny
visible part), though most regions have a presence in everyday life.
 1.2 Exploring Some Concepts 3

The focus in what follows is on those concepts which are most important in
and around the visible spectrum, from the far infrared to the far ultraviolet.
This has emerged as the more conventional domain for the term ‘photonics’.
The aspiration in this book is to convey an intuitive understanding of the
essential photonics concepts and to develop a critical insight into applying
them in order to appreciate the potential and, indeed, the limitations of new and
emerging photonic technologies.

1.2 Exploring Some Concepts

The basic ideas of electronics and of photonics concern the interactions
between electromagnetic radiation and materials. These interactions in turn
depend fundamentally on both the material itself and also its dimensions
compared with the wavelength of the electromagnetic radiation (Figure 1.1).
This is exemplified through the observation that television is broadcast on
radio waves with ultra-high frequencies – a few hundred MHz – where the
wavelength is around the dimensions of a domestic antenna. From the antenna
the signal proceeds into an electronic circuit – in your TV – which ‘holds’ the
arriving electromagnetic wave within its much smaller dimensions. There is
also an intermediate phase, in which the electromagnetic wave captured by
your TV antenna is trapped within a cable before arriving at the electronic
circuit. The flow of electrons in materials is common to the whole process,
since the transmitted TV wave is generated from electrons flowing through a
suitable transforming device to produce a propagating electromagnetic wave.
The source of the term ‘photonics’ lies in the concept of the photon. Photons,
discrete quanta of light, were inferred through the ‘ultraviolet catastrophe’ in the
black body spectrum and later more clearly defined through Einstein’s descrip-
tion of the photoelectric effect, in which electrons are emitted from a material
only if sufficient energy is acquired from the incident light. The required energy
depends not on the incident power but on the optical frequency. This is in
contrast with what is happening in a TV antenna which modifies the movement
of already free electrons in a way that is directly dependent on the total power.
However, it was already known that in other structures, lightwaves behaved in
exactly the same way as radio waves, for example in transmitting through the
atmosphere, being reflected at dielectric boundaries or interfering with them-
selves after transmission through different paths.
In photonics the threshold of ‘where photonics matters’ is now associated
with the energy of the photon. The term ‘photonics’ becomes applicable when
the photon energy becomes comparable with or exceeds the thermal energy of
4 Photonics – An Introduction

the particles in the material with which the electromagnetic wave is interacting.
The energy of the photon, namely Planck’s constant multiplied by the fre-
quency of the electromagnetic wave, and the energy of a freely moving
particle, which is of the order of Boltzmann’s constant times the absolute
temperature, are compared in Figure 1.2 at a temperature of 300 K. (The
photon energy here is presented in electron volts – the energy change as an
electron changes its electrical potential by 1 V – see Appendix 5 for the values
of these constants.) When the photon energy comfortably exceeds the inherent
thermal background energy of particles (at a wavelength of about 30 microns
at 300 K – the far infrared), then electromagnetic radiation is captured or,
indeed, generated as a particle. At lower energies (wavelengths above about
300 microns, 1 THz) we can regard the capture or generation of electromag-
netic radiation as being due to a flow of electrons produced by the electric field
in an electromagnetic wave. In other words, at lower frequencies a wave model
for the interaction becomes appropriate; the ideas of ‘electronics’ rather than
‘photonics’ apply. This will depend on temperature: at 3 K a photonic descrip-
tion of an interaction will begin to be applicable at frequencies two decades
less than at 300 K.
Similar observations apply to structural dimensions, though the boundaries
between the photonic and electronic descriptions are based on different criteria.
Materials behave ‘electronically’ when their dimensions are significantly less

1 MV
Photonics Electronics
concepts concepts
to the fore to the fore
1 kV

Photon
energy 1V Visible
(eV) spectrum
Thermal particle motion energy
at room temperature – 300 K
1 mV

1 mV

1 pm 1 nm 1 mm 1 mm 1m
Wavelength (metres)

Figure 1.2 The plot shows photon energy as a function of wavelength. The
broken line shows the energy of thermal random motion at 300 K. At wavelengths
below about 30 microns (10 THz), the photon energy begins to dominate. At about
300 microns (~1 THz) the ideas of electronics become more applicable. The THz
region falls in the transition zone.
 1.4 About This Book 5

than a wavelength. A contrasting aspect of photonics has been that very small
scale structures with dimensions in the photonics region have only recently
become potentially achievable; that is, to dimensional precision measured in
nanometres. It is remarkable that gold nanospheres have been used to provide a
spectrum of permanent colour in glassware since ancient times. Finally, there is
the dimensional range of structures which is comparable with the wavelength
and it is here that diffraction effects, in particular, become critically important
and waveguiding – as in fibre optics – comes to the fore.
These principles and their context in potential applications will be explored
in more detail later in the book. They have, however, prompted considerable
debate as to whether light is ‘really’ a wave or a particle. The answer to this
question is that it depends on the context. Our aspiration here is to help
develop an understanding of which approach is relevant and when. An attempt
to portray these differing situations is presented in Figure 1.3, which highlights
the underlying phenomena that constitute photonics.

1.3 Applications
Applications, achieved and potentially feasible, are becoming a major stimulus
for the interest in and development of photonics. We shall explore briefly a
number of case studies later in the book, but among the everyday presence of
photonics sits the internet, which could not function at all without fibre optic
communications. Your camera, your smart phone and the increasing efficiency
and versatility of lighting technologies all rely on recent advances in photon-
ics. Many aspects of medical diagnosis, microscopy and medical therapeutics,
rely on photonics. Optical absorption is the initiator for climate change; the
monitoring processes measuring changes in our atmosphere, as well as the
long-term solutions aimed at driving our energy sources directly from light, all
need an understanding of photonics and its exploitation.

1.4 About This Book

The essential concepts which are covered in the book are shown in Figure 1.4.
The next three chapters are devoted to the principles of the subject. Having
established the principles, we shall explore some current applications through
brief case studies. Finally comes a speculative look at the future in terms of
both the enabling technologies and the technological advantages, together with
some social needs to which photonics may contribute to in the future either
 Free space,
dimensions >> l:
in space wave model in space

Waveguides,
Transmit the dimensions ~ l: Detect and
in time in time
Manipulate photon as wave model decode the
the photons waves or arriving
Generate currents photons
the
photons:
lasers, in Nanostructures, in received
LEDs, transmitted dimensions << l: properties
thermal properties current model – intensity,
sources – intensity, wavelength,
wavelength, polarisation
polarisation etc.
etc.
Figure 1.3 A picture of ‘what is photonics’: the generation, manipulation, transmission and detection of electromagnetic radiation in that region of
the spectrum where the photon picture dominates the behaviour at generation and detection, but where transmission can be described in terms of
waves, in large-scale dielectrics or in medium-scale waveguides. At the smallest scale, over very short distances, currents in conductors or in nano-
scale synthetic dielectric structures, become important.
 Chapter 1 Chapter 2 Chapter 3 Chapter 4
Introduction The Nature of Light Light and Materials Light and Structures

Overall aim Wave–particle duality Linear materials, the Free space

dielectric constant,
resonances and plasma
What is photonics? Nanophotonics, circuits and Dimensions ~ λ, waveguides
structures Non-linear materials and
fluorescence
Dimensions << λ, metamaterials
Application motivation and circuits
Quantum mechanics Spectroscopy
(linear and heat)
Basic generic concepts
Other absorption

Light-emission
LED lasers

Light-emission
LED lasers

Materials controlling light:

EO, AO, spatial, temporal

Chapters 5 and 6 Applications and the Future

Figure 1.4 The essential concepts of photonics and their organisation in this book.
8 Photonics – An Introduction

Applications and the future

Chapter 5 Chapter 6

Using What’s Here What’s Coming:

Technology and Applications

Spectroscopy Photonics – the great

enabler: 21st century
electronics
Lasers, cutting,
surgery, photoacoustics
Emerging technologies:
future applications
Interferometry and
precision instruments
Super-resolution
imaging
Displays

Entanglement
Fibrecomms
Plasmonics,
nanophotonics,
Microscopy, telescopes optical circuits

3D, holography, OCT Metamaterials

Phototherapies

Solar energy

Environmental etc.

Figure 1.5 The present and future role of photonics in our developing society.

through further exploitation of what we already know or through the use of the
new tools, as encapsulated in Figure 1.5.

1.5 Some Related Background Ideas

The best way of describing the interaction between light and a material
depends on the nature and structure of the material and its dimensions.
Resonance plays a large part in photonics; this occurs when the structural
 1.6 Problems 9

dimensions match the wavelength. The same phenomenon is familiar in

musical instruments; long organ pipes resonate at lower frequencies than short
ones. Resonant structures have entirely different propagation properties from
an unstructured equivalent space and this applies in every day optics too, from
the coatings on camera lenses to the colours of butterfly wings. The phenom-
enon whereby the dimension of a structure, in relation to the wavelength of an
incident wave, affects the material properties seen by the wave is ubiquitous
and powerful; it applies to every form of wave from mechanical resonances
and acoustic waves to visible light and into the X-ray region. The same
concept also explains much of quantum mechanics, particularly the ideas of
energy levels and band structure. The concept of the ‘universal wave’ will be
used extensively in what follows and, for a deeper and more thorough exam-
ination of this very powerful intuitive tool, there is little if anything to match
J. R. Pierce’s book (Further Reading A). Other conceptual tools are also
extremely useful for an understanding of photonics; of these perhaps the
Fourier transform is the most important. There is an extensive literature
available, including a classic text by R. Bracewell and one by J. F. James
(both listed in Further Reading A). Much of what is discussed here is also
underpinned by Maxwell’s equations, which themselves describe many of the
interactions between electromagnetic waves and materials. Daniel Fleisch has
produced a straightforward guide to these important relationships (see Further
Reading A). At the other extreme, whilst the present text aims to introduce the
basic concepts of photonics, the book by Saleh and Teich (see Further Reading
C) presents over 1000 pages of comprehensive detail for those who may wish
to go further.
Finally, the International Year of Light produced some inspiring texts
(see Further Reading B) highlighting the critical contributions which light
makes to our world and exploring the many photonic tools we can use to aid
this exploration ranging from a mirror through a smart phone to the Hubble
telescope.

1.6 Problems
Some topics to investigate:
1. How might the boundary lines for photonics, as indicated in Figure 1.2,
change in deep space? Could there be circumstances where, for example,
the photon concept is useful for describing an electromagnetic wave of 3 cm
wavelength?
10 Photonics – An Introduction

2. The term ‘microwave photonics’ appears fairly often in current scientific

literature. What is likely to be the prime condition for the observation of
microwave photons? Figures 1.1 and 1.2 might give some hints on this.
Note that microwave photonics can also apply to the situation where a light
beam is modulated by a microwave frequency signal.
3. Now think about a microwave communications satellite, in synchronous
orbit about 22,400 miles above the earth where the ambient temperature is a
few kelvins. Would you consider a communications satellite a photonic
device and why? Some points to consider are that it is subject to the sun’s
rays for half the day, the internal circuits generate heat and the heat loss in
what is essentially a vacuum will be primarily by radiation. (It may be
helpful to look up the satellite communications frequency band using a
search engine. A great deal of useful background on many, many topics can
be found in this way!)
4. At the other end of the spectrum – when might X-rays with photon energies
in the 100 V to 100 kV range behave as waves rather than particles?
 2
The Nature of Light

The question ‘is light a ray, a wave or a particle?’ has long fascinated
philosophers and has intensified since the concept of the photon first
emerged at the beginning of the twentieth century. The pragmatic answer
to the question is ‘it depends what you are looking for’ and is really about
how best to conceptualise light in particular circumstances. We have
already mentioned such distinctions briefly in Chapter 1. This chapter will
extend these basic concepts to introduce the material which follows in later
chapters.

2.1 Light in a Vacuum

This is the simplest case. When no other materials impede its progress or
modify its path (Figure 2.1), propagation in a vacuum can be accurately
described through a wave propagation model. Ray optics also works since in
an idealised infinite vacuum there are no edges or multiple paths through
which the light can interfere or diffract.
By the same token, however, a particle of light will also proceed unim-
paired along the same path as the ray. And, already, we have an apparent
contradiction with the wave propagation model, in that a single photon
cannot spread its presence over a surface area increasing as the square of
the distance from the source. This dilemma continues to
produce philosophical debate but really only becomes pertinent when we
produce single identifiable photons and watch them in motion. More on
this later!
Perhaps the principal useful aspect of propagation in a vacuum is that light
then proceeds at the velocity commonly approximated as c = 3
108 m/s.

11
12 The Nature of Light

Spherical
Ray propagation wavefronts

Wave amplitude
indicating 1/r spreading
(1/r2 power density)

Distance
marked
in l

Figure 2.1 Propagation of rays and waves from a point source in a vacuum. The
wave amplitude representation indicates a reduction in power density with the
inverse square of the distance, but the total power remains constant as the wave
progresses.

2.2 Light in Isotropic and Anisotropic Materials

In a material, light, as it propagates, interacts with the molecules within the
material, typically by inducing some form of electronic motion within the
molecular structure; this may result in dipole moments which change in
response to the lightwave’s electric field (Figure 2.2).
We can regard the way in which this dipole moment follows the incoming
electric field as like a classical mass–spring–damper oscillator excited by the
electromagnetic wave. This process (Figure 2.3) can be viewed in more detail
through the Clausius–Mossotti equation, discussed in Appendix 3, to give a
surprisingly useful and accurate insight into the behaviour of the material’s
dielectric constant ε (see Appendix 3) and, through that, its refractive index n
pffiffi
as a function of frequency, n ¼ ε, is probably the most useful concept in
optics and photonics and is defined by equation 2.1, where v is the velocity of
light within the material:
c
n¼ (2.1)
v
The slowing down of light in a material can be viewed as due to the energy
storage and release through the electron motion induced in the material, which
introduces a delay in propagation. Through similar reasoning (and confirmed
through the Clausius–Mossotti equation) the more energy that is stored the
larger the delay, so the more molecules per unit volume the higher the
 2.2 Light in Isotropic and Anisotropic Materials 13

Effective overall centre of

electron charge distribution shifted from Force due to incoming wave
nucleus by applied field

Incoming
-
nucleus
optical +
wave

Figure 2.2 A schematic of how the oscillating electric field in an incident

electromagnetic wave can distort the electronic charge distribution around an
atom, producing an oscillating dipole.

Equivalent
spring
constant k
(electron
attracted to
nucleus)

Mass of equivalent field-

m excited particle (electron)

C
Damper constant C
representing
material losses

Figure 2.3 The mass–spring–damper model of an atomic electron forced into

oscillations about the nucleus by an optical wave, thus forming an oscillating
dipole.

refractive index. By the same token, in the case of anisotropic materials,

exemplified in many crystalline structures, the refractive index seen by an
incoming electromagnetic wave will be higher when its electric field oscillates
in a crystalline plane with a higher density of molecules. An electric field
oscillating in a plane with lower density will experience a lower refractive
index and so will travel faster. (This phenomenon is known as birefringence.)
Hence the different polarisation states (Figure 2.4) of an incident light beam
with respect to the material’s orientation will see a range of refractive indices.
14 The Nature of Light

Atoms or molecules in
an anisotropic crystalline
structure

Electric field sees

higher refractive
index

Electric field sees lower refractive index

Direction
of propagation

Figure 2.4 The origin of birefringence in asymmetrical materials – the effective

molecular density perceived by the incoming wave depends upon its electrical
polarisation direction.

Another important observation here is that the refractive index for polarisa-
tion states other than the two states aligned with the crystalline structure of the
material will be a mix of two values – the input state will be split into these two
components along the so-called principal axes of the material. Each compon-
ent will consequently exit with a different phase delay, so that the output
polarisation state will in general be different from the input state. Appendix 1
explores this in more detail.
There are many other insights which are enabled by the mass–spring–
damper model. Any mass–spring–damper arrangement involves a frequency
response to a given applied force. At low frequencies the deflection introduced
by the applied force has a particular characteristic value and is in phase with
the applied force. At the resonant frequency, the deflection is a maximum but
is now in quadrature with the applied force. After passing through the reson-
ance the displacement becomes small and in antiphase (Figure 2.5).
By the same token, then the refractive index will have exactly the same general
resonance characteristics, as the mass–spring–damper model though in molecular
materials there are many resonances and therefore the refractive index behaviour
follows the trends indicated in Figure 2.6. Consequently, since all these reson-
ance effects are taking place, the refractive index does vary with frequency and so
the velocity of light within the material is also a function of frequency – a
phenomenon known as dispersion. Looking at an everyday example, at low
frequencies water has a dielectric constant of 80, corresponding to a refractive
index of about 9, but the refractive index in the visible region is about 1.3.
Figure 2.6 also imples a number of other important photonic features,
common to all materials. The refractive index generally decreases with
 2.2 Light in Isotropic and Anisotropic Materials 15

100
Displacement,
arbitrary units
50

0
0 10 20 30 40 Frequency, 50
–50 arbitrary units

–100

–150 Phase, degrees

–200

Figure 2.5 The generalised displacement and phase of a mass–spring–damper

oscillator as functions of the frequency of the applied force. Note the change in
phase around resonance and the fact that the displacement is in antiphase above the
resonant frequency, resulting in a negative contribution to the dielectric constant.

100 Rf and
microwaves
Infrared
Real part
of the
refractive UV
index
1 Nuclear
vibrations
Molecular Atomic
vibrations vibrations Real part < 1

frequency
Imaginary part

Figure 2.6 The generalised behaviour of the refractive index as a function of

frequency indicating resonances due to various molecular and atomic vibrations.

frequency except around the resonances, after which there is another step
downwards. Also, the refractive index can drop below unity. But does the
velocity of light ever exceed its velocity in a vacuum? The implications of this
apparent dilemma are examined briefly in Appendix 4, which considers the
phase and group velocity concepts.
Thus far we have ignored another important aspect of our mass–spring–
damper model – namely, the damper. This represents the sources of loss, and
this lost energy is converted typically into heat. So, strictly speaking, the
16 The Nature of Light

refractive index, which describes the transfer of energy through the material,
should include this loss term. This is expressed through the representation n0 =
n + jk, where j is the square root of 1 and k is the known as the imaginary
component of the refractive index. Additionally, as indicated in Figure 2.6, the
loss terms increase significantly at resonance since this is where the radiation is
more readily absorbed, typically by a large factor. This absorbance feature is
one source of the colour in the world around us.
The propagation of light within a homogenous material is then in general
described through the propagating wave equation
   
2πnz 2πkz
E ðz; t Þ ¼ E 0 exp j ω t   (2.2)
λ λ
Here E is the electric field amplitude of the propagating wave, z the axis along
which the wave propagates, ω the optical angular frequency in radians/second
and λ the wavelength in vacuum; and the final term represents the attenuation
of the electromagnetic wave. The first term gives the time dependence, and the
second term the distance dependence, of the optical phase. Note that the
imaginary part k of the refractive index determines the level of attenuation.
There is yet another aspect to this – suppose the electrons are no longer
bound to individual molecules but free to move. This happens, for example, in
conducting metals and also in the ionosphere around the earth. In this case the
restoring force is due to the electrostatic forces exerted by the positive ion
lattice, as indicated schematically in Figure 2.7. Remember, too, in the case of

+ +
-
+
- -
Fixed nuclei (+)
- + +
-
+

+ - +
-
- + -
+ + +
-
- - -
+
- + Electrons ( – )
- +
- - free to roam

-
+ + - +
-
+ + +
-
+ - +
- +
-
-
+
- + -
+
+ -

Figure 2.7 The plasma case where free electrons experience a restoring force from
stationary positive ions. Note there is minimal damping for frequencies above the
inverse of the electron lifetime in the material.
 2.2 Light in Isotropic and Anisotropic Materials 17

conductors, the losses are due to the collisions between the moving electrons
and the stationary ions, a phenomenon encapsulated in the electron relaxation
time for the material concerned.
There is again a moving mass, namely that of the electrons, and so again we
have a resonance dependent upon the density of the free electron cloud and
given by (see Appendix 3):
ω2p ¼ q2e N 0 =me (2.3)

where qe and me are the electronic charge and mass, N0 is the free electron
density per cubic metre and ωp is known as the plasma resonance (angular)
frequency.
This plasma resonance has many interesting ramifications. Everyone is
aware that X-rays travel through metal but light does not. Satellite broadcast-
ing in the GHz region comes down from outer space through the ionosphere.
However, much lower-frequency radio waves in the short wave band bounce
off the ionosphere in the same way that light bounces off metals.
The impact of the relaxation time in plasmas (see below) is also important.
As mentioned above, the conduction losses in a metal are basically determined
by the frequency of collisions between the freely flowing conduction electrons
and the stationary ions in the metal. However, if we apply an electric field at a
frequency which exceeds this collision frequency, the inverse of which is the
relaxation time, then there are many fewer collisions between the conduction
electrons and the stationary particles in the metal, so the losses decrease
dramatically – in other words, the metal becomes transparent.
The refractive index of gold as a function of wavelength is indicated in
Figure 2.8. Among other things the rapid variation of the complex index with
wavelength in the visible region (roughly 0.4–0.8 µm) accounts for the differ-
ence in the colour of gold as compared, for example, to the colour of silver.
Most metals in and around the optical and ultraviolet region exhibit an intri-
guing transition between relatively low-frequency behaviour as a conductor
and the higher-frequency behaviour as a relatively transparent dielectric. In
other words, the electrons in the metal constantly turn around so that there is no
current flow and likewise no collisions to cause absorption losses. This
fascinating transition has, within the past couple of decades, evolved into so-
called plasmonics, which seeks to understand and exploit this somewhat
complex evolution.
This discussion has been about light as a wave. There has been no mention
of the photon. These classical wave approaches do give a very reliable guide
and an indispensable insight into the interaction between light and materials.
The photon model also has its place, however. This is most apparent with
18 The Nature of Light

18

Refractive
index Imaginary
part
10

Visible
range

Real part
0
0 1000 2000 3000
Wavelength (nm)

Figure 2.8 The refractive index of gold indicating its change in behaviour
emerging at frequencies above the plasma resonance, corresponding to about
200 nm wavelength.

Excited
states

Energy

Stable ground state

Figure 2.9 An energy level representation for a single atom – which approximates
to the behaviour of a gas at low pressure.

regard to the energy levels evident in a material (Figure 2.9), a concept which
has evolved as the ‘conventional’ approach to explaining the wavelength-
selective absorption of light in materials.
The differences between these energy levels and the ground state in a
molecule equal the photon energy required to excite the molecule from the
stable ground state. This absorbed energy then typically (but not always) re-
emerges as heat when the molecule relaxes to its ground state. The same
behaviour can be ascribed to the resonant absorption of mass–spring–damper
 2.3 Light Interacting with Structures 19

systems if we are looking purely to match the frequency of the input light with
the frequency of the mechanical resonance. This merging of the wave (and
mechanical resonance) domain and the photon domain (via energy level
differences) can provide useful insights into material behaviour.
There are lower-frequency transitions in molecules as well, corresponding to, for
example, exciting an entire molecule into rotational rather than electron-orbit
resonances (see Figure 2.6 for the general trends). A microwave cooker which
operates at around 2.5 GHz exploits exactly this effect, tuning to resonances in
water, fats and sugars. Here, however, we are by no means concerned with a photon
energy in the order of a few microvolts; this is well below the photonics region
discussed in Chapter 1. The scientific community also talks of these frequencies in
terms of molecular resonances, implying an analogous electromagnetic concept.
Much of the wave and photon dilemma is encapsulated in the above
discussion. Whilst numerous weighty texts have been published on the subject
there is really no definitive answer. For us it amounts to finding the approach
which gives the most useful model to analyse and understand the photonics
scenario of interest.

2.3 Light Interacting with Structures

A material structure, in our present context, is an arrangement which comprises
two or more materials with different optical properties. Once again, the
question of a structure’s scale compared with the optical wavelength enters
the discussion.
So, we start with reflection and refraction at a large-scale structure, namely an
interface between two dielectric materials (Figure 2.10). We will simplify the
discussion by assuming that the interface is planar. The light that passes through
the interface is refracted by an amount determined by Snell’s law, namely:
nr sin θi
¼ (2.4)
ni sin θr
The reflected light obeys the normal ‘angle of incidence equals angle of
reflection’ relationship, but what is interesting here is the amount which is
reflected and the way in which this amount depends upon the incident angle.
Indeed, at one specific angle – the Brewster angle – only one polarisation state,
namely that with the electric vector aligned with the surface of the interface
(sometimes called S (in-plane) polarisation and sometimes called TE for
transverse electric polarisation) is reflected. The reflection coefficient is not
100%, so that the refracted ray still contains some of this polarisation state. The
reflection coefficient as a function of angle for the two linear polarisation states
20 The Nature of Light

qr Incident ray Reflected ray

n2
qi
n1
From low index
From high index to high index n2
qi to low index qr
n1

Brewster angle from air to medium of index 1.5

1
Power reflection coefficient

0.5
P polarisation

S polarisation
0
0 30 60 90
Incident angle (degrees)

Figure 2.10 Reflection and refraction at a planar interface between two materials
differing in refractive index. The diagram at top left shows a ray striking the
interface at angle θi and being partly transmitted at angle θr and partly reflected
back into the higher-index material. The ray shown with a broken line is incident
at the critical angle, and so would travel along the interface. At incident angles
greater than thus, total internal reflection occurs. The Brewster angle plot applies
to an index of 1.5 in the higher-index material for light incident from the
low-index side of the interface.

is also shown in Figure 2.10. The partial removal of the S polarisation state
(with electric field vibrations perpendicular to the P state vibrations and to the
direction of the reflected ray) manifests itself in daily life, amongst other things
in the polarising sunglasses designed to reject the (partially) polarised reflected
light from our surroundings.
The discussion thus far concerns a light beam travelling from a low-index
medium to a higher-index medium. When the situation is reversed, there will
be a case when the angle of incidence is high enough for refraction to occur
into the interface plane (i.e. the angle of refraction is 90), and then the incident
angle is referred to as the critical angle. If the incident angle goes beyond that,
total internal reflection occurs, as implied in Figure 2.10, which also indicates
some reflection for incident angles below the critical angle.
Incidentally, the derivation of Snell’s law (see the Chapter 2 problems) exem-
plifies a very important principle in understanding the coupling of light from one
material or structure into another. This principle is that the projected components
of the wavelengths along the interface must be the same – a concept known as
 2.3 Light Interacting with Structures 21

phase matching. This idea applies to any wave propagating through an interface,
whether the wave is acoustic, electromagnetic or even a water wave.
There are many other possible interfaces. In the above we have assumed a
perfectly planar interface. However, slight roughness will introduce corres-
ponding slight changes in the angles of incidence and refraction until, in the
limit, a perfectly scattering surface would send the light in all directions. Again
we encounter this daily – sunlight entering your room is scattered throughout
the entire volume of the room, by the walls and furniture.
Moving on to a medium-scale structure reaches into diffraction and interfer-
ence: we move from the ray optic approach, which works at large scales, into the
wave optics approach. Much is encapsulated in the classic Young’s slits experi-
ment, shown in Figure 2.11. Light from a single-parallel-beam single-wavelength
source impinges on two slits, separated by a distance d and each of a width
conveniently assumed to be much less than the wavelength. The input light then
reradiates in all directions and produces an interference pattern as indicated. The
interference pattern is determined by the vector sum of the fields directed through
the two paths. Full constructive interference occurs when the path difference dA –
dB is an integer number n of wavelengths, which gives, for y » d,
dx
nλ  (2.5)
y
where x is the distance on the screen from the axis of the system and y is the
distance along the axis between the slits and the screen. There is a time delay

Observation
Incident plane wave,
Slits screen x
wavelength l
separation d
dA
A
dB

dA – dB = n l for maximum constructive

interference on observation screen
Interference
fringes

Figure 2.11 Young’s slits – the simplest geometry to exhibit interference

phenomena.
22 The Nature of Light

between arrivals at a particular point on the screen from the two paths and there
is an implicit assumption in equation 2.5 that the phase delay on the optical
signal is completely predictable for this time difference. Sometimes it is not,
which leads us into the concept of coherence; ultimately this coherence
concept applies to both the spatial and temporal behaviour within an optical
source, a subject to which we shall return later.
Much of photonics concerns this medium-scale structure, ranging from
camera lenses and image sensing arrays to DVD players and the display on
your telephone handset. Furthermore, almost all of this discussion is concerned
with wave optics.
At small scales we move into the domain of subwavelength structure in a
material. This domain has only recently become available, with the notable
exception of materials such as colloidal gold, which has been used in pottery for
centuries and for which the colour can range through reds to blue depending
upon the diameter of the tiny gold spheres within the colloid. The resonance
wavelength for the electron plasma circulating the spheres varies as indicated in
Figure 2.12, but gold spheres throughout this dimensional range also absorb
more in the blue than the red, owing to material absorption, the property which
makes ‘normal’ gold appear golden. In current terminology these gold spheres
would be regarded as an example of quantum dots. As colloidal gold they were
simply a way of introducing permanent colour into artefacts. The tiny structural
dimensions take the ‘normal’ colour far from its golden origins.

580
Peak absorption wavelength (nm)

560

540

520

10 50 100
Gold nanosphere diameter (nm)

Figure 2.12 The variation in the peak absorption wavelength of the electron
plasma as a function of the gold nanosphere diameter. A suspension of 10 nm
nanospheres in water, or in glassware, looks ‘reddish’ whilst 100 nm spheres
appear more towards the blue since now the peak absorption is in the red.
 2.5 Problems 23

Much of the story here is concerned with the ability of structures to produce
what are, in effect, wavelength filters. Examples of such phenomena occur
extensively in nature – butterfly wings are amongst the most common and
striking. This comes initially as something of a surprise: let’s make something
a different shape and it will appear as something different. However, a few
thoughts on electronic circuits shows how we take this for granted. As far as a
radio frequency wave is concerned, a piece of straight wire appears as an
object very different from a coil of wire yet both have dimensions significantly
less than a wavelength. What has become known as ‘nanophotonics’ has been
described as a manifestation of this familiar concept – so here light becomes an
electromagnetic wave but in the sense of an electric current.

2.4 Summary
Is light a wave, a particle or a ray or an electric current? The answer is yes to all
four possibilities, but it all depends on the context. The preceding discussion
has given a glimpse of all these aspects and exemplified some of the circum-
stances in which light is more readily perceived using one or the other model.
There is some interchangeability but usually an appropriate combination of
points of view is the simplest way to arrive at a usable understanding. It is quite
a fight, say, to model the refractive index through quantum mechanics and
particles, but in general bringing one model into the domain of another, as Nils
Bohr did with his hydrogen atom model, can give useful, workable and, with
care, accurate results. Of course, those of a philosophical nature should consult
examples of the many extensive texts on the nature of light.

2.5 Problems
These problems could involve collaborative searching on the internet.
1. What features determine the colour of the objects around you? Consider, for
example, butterflies, plants, the bloom on a camera lens or some spectacle
lenses, the iris of your eye.
2. What are the light sources around you – for lighting in your room, for your
DVD player, in your fluorescent watch, in a ‘black lighting’ tee shirt, in the
sun, in the TV screen and in those old cathode ray tube TV sets? How do
these sources vary in structure, in their physiological effect on the observer,
in beam quality etc.?
24 The Nature of Light

3. What determines the colours you perceive when looking at a particular

object? Why do colours apparently differ when viewed at different times of
day, or when one is using different light sources or when one goes abruptly
from the dark into a brightly lit room? How, for example, could the texture
and colour of a wall make a huge difference to the brightness of a room as
sunlight comes into it – what might be the mechanisms at work? You
should take into consideration the way in which the eye responds to
different colours and levels of light.
4. Think about a planar interface between two materials with refractive
indices, n1 and n2, with n1 < n2.
(a) Demonstrate that Snell’s law relating the angles of incidence and
refraction from material 1 to material 2 can be understood by consider-
ing the projection of the components of the optical wavelengths of the
incident and refracted rays along the optical interface.
(b) What is described in part (a) is an example of ‘phase matching’, also
sometimes called ‘resonant coupling’. Look around and discuss where
this general effect may be taking place, both in everyday events and in
the technological world in which we live.
(c) We mentioned the Brewster angle in the text. At this angle, when light
traverses a planar interface from a low-index region into a high-index
region, the reflection coefficient for the electric field vibrations parallel to
the plane of incidence is zero, giving rise to partially polarised light in the
reflected beam. Investigate the generalisation of this observation – look up
Fresnel reflection on the web – and note not only the changes in reflection
amplitude but also the changes in phase.
5. Would a suspension of gold nanospheres (Figure 2.12) appear the same colour
in water, air and high-density glassware? If so, why, and if not, why not?
6. (a) Look up typical values for the free electron density in the ionosphere, in
gold and in n-type silicon and use equation 2.3 to arrive at the plasma
resonance frequencies.
(b) From this, comment on when and how these materials become trans-
parent to electromagnetic waves and evaluate the significance of your
calculations in terms of possible applications.
7. We have indicated in this chapter that there are practical circumstances
when the refractive index can be less than 1. This implies a velocity of light
which exceeds the velocity of light in vacuum. However, it is widely
accepted that the speed of light in a vacuum is the fastest that light can
travel. How can we rationalise these two observations? The key lies in the
subtleties of phase and group velocities, hinted at in problem 4 above and
discussed a little more in Appendix 4.
 3
Light Interacting with Materials

Here we shall explore the principal phenomena which occur when light passes
through, and therefore interacts with, a homogenous material.

3.1 Linear Optical Materials

The word ‘linear’ in this context indicates that light emerging from a material
is at exactly the same frequency as the light which entered the material. In
other words, all the ‘springs’ in the mass–spring–damper model (the Clausius–
Mossotti equation) mentioned in Chapter 2 are operating within their linear,
Hooke’s law, region. We also mentioned the concept of the real and imaginary
parts of the refractive index, respectively describing the propagation delay and
the rate of energy loss by absorption. These are, as with all linear oscillatory
systems, intimately connected to each other through the Kramers–Kronig
relation1 for the complex refractive index n + jk, relating the real part n and
imaginary part k:
ð
2 ∞ k ðΩ Þ
n0 ðω Þ ¼ 1 þ P dΩ (3.1)
π 0 Ω2  ω2
where P is the Cauchy principal value operator. Entire textbooks have been
written with this equation as their sole topic – the principal message is that the
real and imaginary parts of the refractive index are intimately related and that
by knowing k over the integration range as a function of the optical angular
frequency Ω, in principle it is possible to calculate n at the desired frequency ω.

1
In terms of the dielectric constant ɛ = n2, the Kramers–Kronig relation is
Ð ∞ ðεÞ
ReðεðωÞÞ = 1 þ π2 P 0 ΩΩIm
2
ω2
dΩ.

25
26 Light Interacting with Materials

(a) (b) (c) (d)

(e) (f) (g)

Figure 3.1 Some of the potential resonant modes in a triatomic molecule, indicat-
ing the range of rotational, vibrational, orbital and nuclear potential resonances, all
of which contribute to the wavelength dependence of light energy, which pro-
duces the characteristic colours of materials, their unique spectroscopic signatures.
(a) A molecule, e.g. water or carbon dioxide, (b) rotational modes, the entire
molecule and bonds rotate, (c) ‘longitudinal stretch’ vibrational modes, (d) ‘lateral
stretch’ vibrations, (e) vibrations of ‘bond orbits’, (f) vibrations of atomic orbits,
(g) nuclear vibrations in individual nuclei.

There are, as already noted, many resonant modes in a molecular material

(Figure 3.1 illustrates some of these). The principal modes of interest here are:
– the various stretching and rotational modes that are possible in a molecule,
typically found in the terahertz and microwave regions;
– electronic orbital resonances corresponding to excitation between particular
electronic energy levels, typically found in the infrared through to the
ultraviolet region.
There are also numerous hybrid oscillating modes where some lower-frequency
resonances mix into higher-frequency resonances and produce sidebands, thereby
giving a so-called ‘fine structure’ to the absorption characteristics. The whole
story can also extend up to the far ultraviolet and X-ray regions and beyond, where
portions of the spectrum embrace resonances within atomic nuclei.
This very brief exploration of the possibilities within the resonant spectrum
indicates that these spectra are inherently very complex – a mixed blessing,
since on the one hand this complexity implies individual identity for a particu-
lar molecular structure, even down to the isotope level, but on the other hand
intricate frequency signatures require very careful analysis.
Thus far we have only considered one molecule in isolation. When we bring
together groups of molecules, as gases, liquids or solids, then the scope for
interactions is considerably enhanced. In this context, recall that if any two
molecular structures are sufficiently close to interact with each other, then the
corresponding energy levels therein begin to split, as illustrated in Figure 3.2.
Going back to the classical oscillator model, similar ‘splitting’ in resonances
can also occur in mechanically coupled systems.
 3.1 Linear Optical Materials 27

Simplified energy level model of a single atom, An interacting group of atoms –

as in a gas at low pressure diffused multiple levels –
as in gases, liquids and solids
Energy

Excited
states Levels progressively split
as molecules get closer,

often overlapping in
Energy liquids and solids

Stable ground state Stable ground states

Figure 3.2 A schematic of the splitting of energy levels as atoms get closer to
each other. For most solids and liquids these levels begin to overlap into energy
bands. Molecules also have similar energy level diagrams – but with significantly
more detail therein.

In groups of molecules there are also other forces at work, most noticeably
in gases, associated with changes in pressure and temperature. Increases in
pressure obviously involve increases in molecular density and consequently
also increases in the value of the refractive index. Additionally, bringing mol-
ecules closer together increases the level splitting, and thereby broadens any
particular atomic or molecular energy level. Bringing molecules together also
increases collision probabilities, thereby broadening the perceived linewidth.
Increasing the temperature speeds up the molecular motion, thereby increas-
ing the mean velocity as the square root of the absolute temperature. Hence,
because of the Doppler shift, the perceived range of energy levels participating
in a particular transition also increases – that is, the absorption line broadens
again. However, with gases there is another factor at work. The preceding
argument applies to a constant-volume rise in temperature. If the volume of the
gas is unconstrained, the answer is somewhat different. The number density of
molecules decreases, with the opposite effect.
These pressure- and temperature-related phenomena are present in the
spectra of liquid and solid dielectric materials but are far less evident than
for gases, except during phase transitions from gas to liquid to solid or, in some
cases, when solids or liquids move from one particular molecular state of
organisation to another.
The linear propagation of electromagnetic waves in conductors is also
important, even in the range of the spectrum where photon interactions become
significant. We have previously mentioned the extrapolation of the Claussius–
Mossotti equation into the free electron case appropriate to conductors. Here
the restoring force is generated by the separation of fixed and mobile charges,
giving rise to the concept of collective oscillations at a plasma frequency,
28 Light Interacting with Materials

above which the metallic conductor becomes electromagnetically transparent.

Conceptually, an incoming electromagnetic wave above the plasma resonance
can be thought of as reversing the directions of free electrons before they have
had time to collide. The collision process is the principal source of conductive
losses, so these decrease dramatically above the plasma resonance. Below this
frequency, the conductor will typically reflect lightwaves incident upon it, with
a wavelength dependence determined by the molecular resonances within the
conductor itself. These are observed through the corresponding absorption
spectrum but with some losses due to surface-induced currents within the
conductor. These induced currents within the conductor penetrate to a depth
δ determined by the classical skin effect:
sffiffiffiffiffiffi
2ρ
δ= (3.2)
ωμ

where ω is the optical angular frequency, ρ is the resistivity and μ is the

permeability of the conductor. Strictly speaking, this approximation is applicable
for ω << 1/(ρɛ), where ɛ = n2 is the dielectric constant. At frequenciespwell
ffiffiffiffiffiffiffi in
excess of this value, the skin depth δ approaches an asymptotic value, ρ ε=μ.
Provided the optical frequency is sufficiently below the nominal plasma
resonance the low-frequency values apply (see Appendix 3 for more detail).
The skin depth as a function of wavelength for gold and aluminium is shown
for reference in Figure 3.3. This is effectively another way of representing an
equivalent concept to the imaginary part of the refractive index. If, however,
our metallic layer is thinner than the skin depth then some fraction of an
incident wave will propagate through it. In conventional terms, the metal
conducts. Note, however, that there are significant losses. To give some scale
to these losses, the broken and double-dotted line shows the resistance in ohms
of a wire one skin depth in diameter and 0.1λ in length.
In the optical region, skin depths are typically in the tens of nanometres;
layer thicknesses are easily controlled through straightforward vacuum depos-
ition processes. The currents in a metal layer much thinner than the skin depth
are often referred to as surface plasmons. Through similar reasoning, if the
dimensions of the metallic sample become small compared with the skin depth
then we end up with, in effect, a conductor – a piece of wire. We are now in the
domain of photonics referred to as ‘plasmonics’, one part of ‘nanophotonics’.
We have thus encountered the principal features of the optical properties of
homogeneous materials. Measuring these optical properties gives us a valuable
insight into material structures, and the colour of everything we see is deter-
mined through the absorption spectra of the materials around us – what is not
 3.2 Non-Linear Optical Materials 29

Skin depth 1000

(nm)

100
Region of
plasma resonances

10

Resistance of 0.1l length of gold wire

1
0.1 1 10 100 1000
Wavelength (microns)
Aluminium Gold Silver Pt

Figure 3.3 Skin depth in silver, gold, aluminium and platinum as a function of
wavelength. The last of these has significantly higher resistivity, so the skin
depths are substantially higher. The line with negative gradient shown as
gives the resistance in ohms of a gold wire 0.1 wavelength long
and one skin depth in diameter.

absorbed is either transmitted or reflected and it is these reflected remnants of

the visible spectrum which characterise the world we see. Finally, a word of
caution. The optical properties of the things we see are also, as we shall
explore in Chapter 4, influenced by their structure: witness the butterfly wing
or indeed the sight of oil on water.

3.2 Non-Linear Optical Materials

Non-linear effects occur when the optical frequency emerging from a material
contains components which were not present in the optical input. We could
interpret this as occurring when our ‘spring’, in mass–spring–damper model
(the Clausius–Mossotti equation), goes beyond its elastic limit. However, it is
usually more convenient to interpret non-linear optical phenomena through
photons and energy levels.
The optical Kerr effect is the exception. It applies to all materials and is
present for all levels of optical power density input. The Kerr effect is the
phenomenon in which the refractive index of the material through which an
electromagnetic wave is transmitted undergoes a slight change that is depend-
ent on the electric field Eopt and is given by

Δn = λKE 2opt = n2 I (3.3)

30 Light Interacting with Materials

Molecular energy levels

Virtual or metastable levels – anywhere other than excited state

levels. Here the molecule temporarily ‘captures’ the photon
Excited
states

Incoming
photon
Energy Outgoing
photon
leaving after
relaxation time

Stable ground state

Figure 3.4 The concept of ‘virtual’ energy levels, where an incoming photon can
temporarily excite an atom or molecule to a level not within the energy level
spectrum. The photon will escape a short time later. This is a quantum mechanical
equivalent to the Clausius–Mossotti oscillator.

where K is known as the Kerr coefficient, n2 is the Kerr-induced non-linearity,

and I is the optical intensity, typically measured in W/cm2. At optical frequen-
cies this effect can often be ignored but there are sometimes surprises, e.g. in
optical communications and precision optical measurement systems, where
intensity-dependent phase changes can have a profound impact on system
performance. (For silica, used extensively in optical fibres, n2 ~ 310–16
cm2/W.)
The more interesting non-linear phenomena in photonics are best viewed
through the energy level approach, however. We have seen how, in the mass–
spring–damper arrangement, energy comes in at the original optical frequency,
is stored in the ‘spring’ for a short time and then released. The delay manifests
itself as the refractive index of the material. Viewing this in terms of energy
levels and photons, our incoming photon temporarily excites the molecule,
putting it into an energy level state which does not correspond to any of the
molecular resonances and which is termed a virtual or metastable energy state
(Figure 3.4). We have already seen that the stable state of the molecule – the
ground state – comprises a wide range of closely spaced energy levels charac-
terised by the molecular structure and the influences on the molecule within its
immediate proximity. Consequently, there is the possibility that when the
molecule drops down from a virtual energy state to its ground state, it may
select a different energy level to which to drop. The existence of these many
levels means that photons are emitted at different wavelengths, referred to as
Stokes and anti-Stokes, corresponding to wavelengths that are longer or
shorter, respectively, than that of the incoming photon. There are two sets of
such effects giving rise to the emission of photons with a wavelength slightly
 3.2 Non-Linear Optical Materials 31

Metastable level

Stokes

Anti-Stokes
Incoming photon

Ground states (of sample material)

Relative power
levels Illuminating
frequency

Optical frequency (scale much larger for Raman)

Anti-Stokes
Stokes

Figure 3.5 (a) The scattering process: energy level diagrams for Raman and
Brillouin spectroscopy, where the virtual state releases a photon into a range of
different ground states. These ground states are characteristic of the constituent
material. (b) A resulting spectrum. The optical frequency scale is much larger for a
Raman spectrum, with its greater energy differentials. Because of this, Raman
spectra are more easily filtered from the incoming excitation spectrum.

different from the incoming photon, namely Brillouin and Raman scattering.
Brillouin scattering is initiated at a lower power density threshold, with
corresponding smaller wavelength shifts; for Raman scattering the wavelength
shifts are much larger.
This explanation indicates one important feature of these processes: the
range of different ground states which can be occupied gives a spectrum of
energy differences between the outgoing photons and the incoming photons.
This range of differences is uniquely determined by the material through which
the light is passing; the effect is present over a broad range of input optical
wavelengths and, in particular, at wavelengths well away from the absorption
spectrum of the sample. Here, then, we have another optical means for
characterising materials, namely examining the Brillouin or Raman spectra
and, thanks to the greater photon energy separation between input and output,
Raman spectra are more useful (Figure 3.5).
This gives rise to an important question – what happens to the energy
differential between the incoming and outgoing photons? The answer is that
it released as, or sourced from, phonons, that is, thermal vibrations. In the
case of Brillouin scattering these phonons create an acoustic spectrum
typically in the tens of GHz region. These phonons – very high frequency
32 Light Interacting with Materials

Metastable
excited state
Decay to material
excited-state level

Incoming
optical
Emitted photon
photon
at lower energy
than input photon
Energy

Stable ground state

Figure 3.6 Fluorescence spectroscopy. Here the metastable excited state is well
defined but has a short lifetime, in contrast with the Raman and Brillouin cases.
Decay from it to the material excited state produces heat. Decay from the material
excited state produces luminescence.

acoustic waves – are eventually dissipated as heat. For Raman scattering these
phonons are in the terahertz region and again finally manifest themselves as
heat energy.
Another other important and surprisingly commonplace non-linear optical
phenomenon is fluorescence, as illustrated diagrammatically in Figure 3.6. Here
a high-energy photon excites an electron to a well-defined transient ‘metastable’
state and from this state the molecule relaxes into an intermediate level corres-
ponding to an actual excited energy level within the atomic structure. Thereafter,
the molecule collapses to its ground state, emitting a photon with a characteristic
wavelength. This usually occurs in liquids and solids (only rarely in gases), so
there will be a range of energy levels available, giving the fluorescent radiation a
characteristic output spectrum. Fluorescence is similar to the Raman effect but
had established its identity long before the Raman effect was discovered, in 1928.
Also, there is one important difference. In fluorescence the incoming photon
excites the molecule into a short-lived but well-defined energy level. This photon
relaxes via the two-stage process mentioned above into states related to the
complex refractive index of the fluorescent material. For the Raman effect the
initial transient state can be at any level and is not related to the complex
refractive index of the material concerned. Additionally, Raman scattering
occurs for all materials, but fluorescence for relatively few.
 3.3 The Diversity of Optical Absorption 33

The overall results are very similar, though fluorescence is the more famil-
iar, from black lighting, where ultraviolet lights trigger visible fluorescence in
clothing and décor in nightclubs, to glow worms, which excite their radiating
molecules through biological rather than a directly optical interaction, in a
process usually referred to as bioluminescence. The basic process is the same –
an input stimulus, in the glow-worm context, biochemically excites a molecule
to an intermediate state prior to its dropping a little energy before returning to
its ground state and emitting a corresponding photon. There are many other
variations on the basic theme. Electro-luminescence is perhaps the most
familiar, where an electric current excites atoms which thereafter relax and
emit photons; this happens in, for example, a light emitting diode. All these
variations, however, have one common feature in that the light emitted is
characteristic of the material from which it originates and consequently can,
among other possibilities, be used as an optical signature in materials analysis.

3.3 The Diversity of Optical Absorption

From straightforward experimental observations, we can conclude that if light –
from the sun perhaps – shines on a material, the energy of the light which
bounces off that material almost always adds up to a lower total than that of the
light which arrived. Consequently, by the conservation of energy, the remain-
der must be converted into something different, and the question is, what?
The most obvious answer is heat. A black sheet of card left out in the
sunshine soon feels warmer than a white one placed beside it. Both sheets will,
however, take time to stabilise to their eventual steady temperature – a
temperature at which the light absorbed is exactly balanced by the heat lost.
This in turn is a function of the thermal conduction and convection routes from
the sample and its thermal capacitance. Since the samples are relatively large,
the time taken for this equilibrium to become established is measured in
seconds or minutes. This in turn leads to the question, what happens if we
switch the light off again before this equilibrium is established? The answer is
that everything cools. Thermal expansion is happening too, so in this process
the actual physical size of our sample increases while exposed to the light and
decreases when the light is removed.
There are some interesting implications if we change the context slightly to
that of a tightly focussed beam, usually from a laser, incident on a large metallic
sample and switch this beam on and off very rapidly (Figure 3.7). We know,
thanks to the perceived colour of the metal, that invariably some incident light
will be absorbed. We also know that this absorption will take place within one
34 Light Interacting with Materials

Incident Reflected
light light

Absorbed
Local section
light in ~ 100 nm
through plate
sample

Low frequency High frequency

Thermal diffusion lengths

Figure 3.7 Photoacoustic conversion. The light is absorbed in the optical skin
depth, producing a thermal wave whose characteristic skin depth is dependent
upon the frequency at which the light is modulated.

optical skin depth, which is usually (Figure 3.3) less than 100 nm. All the heat is
initially generated in this tiny volume but rapidly diffuses away into the bulk
material. The thermal conduction processes are exactly analogous to the elec-
trical conduction process so that, again, if the dynamic heat source (our incident
light beam) is switched on and off at a particular angular frequency ω, at this
frequency the thermal diffusion depth will be given by

δth = (2σth / (sth Δ)1/2)/ω1//2 (3.4)

where σth is the thermal conductivity, Δ is the density and Sth is the specific
heat. Values of the thermal diffusion depth for brass, which is typical of many
metallic materials, are shown in Figure 3.8 for a range of frequencies ω up to
around 100 MHz. Acoustic wavelengths for compressional waves in a typical
metallic material are also plotted for comparison. We can then see that if the
light beam is switched on and off sequentially then the local temperature will
rise to a specific depth, which, in this range of acoustic frequencies, is much
less than the acoustic wavelength. This will induce local thermal expansion,
resulting in pressure differentials and hence ultrasonic waves. Such laser-
generated ultrasound is an extensive topic in its own right, with important
applications in acoustic imaging and non-destructive testing, and numerous
full texts describe its principles and applications.
Thus optically induced localised heating, typically from a short-pulse laser
beam, produces a highly localised temperature rise, initially within the optical
skin depth and rapidly thereafter spreading further into the target material.
This leads to another observation, that the smaller the laser spot and the greater
 3.3 The Diversity of Optical Absorption 35

Structural
Thermal and acoustic waves in brass dimensions
of interest
1m

Acoustic
Diffusion depth / wavelength

1 mm

Thermal

1 mm

10 kHz 1 MHz 100 MHz

Figure 3.8 Relative dimensions for thermal diffusion depths and acoustic wave-
lengths, here given for brass.

the energy density applied within a given time period (i.e. the higher the peak
power density), the higher the localised temperature rise within the optical skin
depth. This highly localised short-term temperature rise can be high enough to
melt or even evaporate the metal.
An important application is laser machining, now extensively used in many
production processes. For some materials, notably polymers, this machining
process may not necessarily involve melting the target but, rather, introducing
light at a high enough photon energy to break the chemical bonds which hold
the polymeric molecules together.
Closely related to this is laser-induced breakdown, typically but not exclu-
sively in gases, which can give a highly intense source of light stimulated by
tightly focusing a laser beam in the sample material (Figure 3.9). In this case
breakdown is induced by the large electric field within the light beam at the
focus, similar to the arcing sometimes seen around high-voltage power lines.
This laser-induced breakdown produces a spectrum which is characteristic of
the material in which the breakdown occurs. Laser-induced breakdown spec-
troscopy (LIBS) is a spectroscopic tool with an immense range of application
spanning atmospheric analysis to the characterising of biological cell structures.
Turning light into heat on a less dramatic scale also has its applications.
Indeed, thermally driven solar power can provide hot water in warm countries.
Large-scale arrays of mirrors focussing the sun onto a black boiler structure, a
36 Light Interacting with Materials

High-peak-power laser beam Convex focussing lens

Sample

Breakdown region

Figure 3.9 Laser induced breakdown spectroscopy, where a tightly focussed laser
beam produces very high, very localised electric fields which ionise the sample
into a gas, producing a characteristic, usually line, spectrum.

solar furnace, can even provide steam for electrical power generation. In a very
different context, imagine that you could fabricate the perfect ‘black body’,
which would absorb all photonic wavelengths. The temperature rise in such a
body would then be proportional to the optical power incident upon it. If this
thermal change were coupled into an electrical thermistor or a thermocouple
then measuring the temperature change would give a direct reading propor-
tional to the optical power; this is realised in practice in a device called a
bolometer, which has applications across the electromagnetic spectrum.
In other forms of photodetection, incident light to creates an electric signal,
usually a current, proportional to the optical power absorbed. The origins of
this direct conversion lie in the photoelectric effect, which led to the concept
of the photon via the idea of a characteristic ‘work function’ necessary to
eject an electron from a cathode material into a vacuum.
The photomultiplier (Figure 3.10) was thereafter for many years the princi-
pal means through which optical signals could be turned into proportional
electric currents. However, it is important to note that in general one photon
creates only a single electron, so the resulting current is proportional to the
photon arrival rate rather than the optical power. It is also important to note
that the idea of the photomultiplier is that each photon-generated electron
produces many electrons via the electrode cascade. This process is statistical
and introduces noise into the detected current. There have been many vari-
ations on the photomultiplier theme. For example, early television cameras
were based upon the vidicon, a device which operated through a directed
electron beam and a photocathode on which an image was projected.
 3.3 The Diversity of Optical Absorption 37

Cathode – input electrons

Intermediate electrodes

Input light
Anode contact

Figure 3.10 The principle of the photomultiplier tube. The input light causes
electrons to escape from the cathode and these are multiplied through successive
intermediate electrode stages.

In 1947 the semiconductor transistor was invented; this, it is safe to say,

changed everything. The semiconductor photodiode now detects optical power
in CD and DVD players, optical fibre communication systems, responsive
illumination systems, camera systems and a host of other areas.
The basic principle of the semiconductor photodiode requires that the
incident photon has an energy exceeding that of the bandgap in the semicon-
ductor (which is comparable with the work function of a photocathode). The
photodetector is usually reverse biased (Figure 3.11) and this reverse bias
sweeps away the free electrons generated by incident photons in the conduc-
tion band and the free holes in the valence band, to produce an electrical
current proportional to the incident optical power (remember, though, we are
really looking at the incident photon arrival rate, as for the photomultiplier).
This observation leads immediately to one of the principal features of
semiconductor photodiodes (and photomultipliers) – there is an optimum
wavelength, corresponding to photon energies somewhat above the threshold,
where the sensitivity of the photodiode, measured in amperes per watt, reaches
an optimum. This can be understood by recalling that each photon can only
create one electron–hole pair and each pair will need available energy levels in
the valence and conduction bands respectively. If the photon energy is too close
to the band edge, the number of available states will be limited so that the
absorption process will be relatively inefficient. If the photon energy is too high
then, yes, there will still be many states available but each of these higher energy
photons will still only produce one electron–hole pair. Consequently, the current
produced for a given power level will be reduced, since, for a given total power,
the number of photons per second arriving at higher photon energies will be less.
This results in a responsitivity plot which depends upon the semiconductor
material; it is shown for some common examples in Figure 3.12.
There are other factors which determine the optical detection efficiency
(the percentage of incident photons converted to electron–hole pairs) of the
38 Light Interacting with Materials

(a) Photodector

+
-

n+ substrate Intrinsic Transparent

p region top contact
on thin p+
region
(b) Avalanche photodiode

+
-

n+ substrate Intrinsic Transparent

p region top contact
on thin p+
Precisely doped p+ region
avalanche region

Figure 3.11 (a) Reversed-bias p–n junction used as a photodetector. In (b) we

have the avalanche photodiode, which is very similar except that there is a high
electric field region near the n+ region. ‘Intrinsic’ here means very low p-doping
levels.

semiconductor p–n junction photodiode. The most important of these is the

depletion length, the length of the region swept clear of carriers in the absence
of any photon flux. Any photons absorbed within the p+ type or n+ type
material on either side of this region will create additional carriers. However,
these carriers will find themselves in a very low electric field and so will not
separate significantly before recombining, so that these photons have been
‘lost’. Nothing comes without compromise and so, in the case of the photo-
diode, the wider its region, the longer it takes for carriers created within this
region to arrive at the ends of the region. This delay affects the available
operational bandwidth of the photodiode, which is restricted by the speed with
which carriers cross this depletion zone.
The carrier recombination process mentioned above also has some import-
ant features. The band structure of a semiconductor is a little more complicated
than that hinted at earlier. There is structure not only in particle energy but also
 3.3 The Diversity of Optical Absorption 39

1.0 90%
InGaAs
Quantum
Responsivity (A/W) efficiency
50%

0.6
Si
Ge

0.2 10%

0.7 1.1 1.3

Wavelength (mm)

Figure 3.12 Responsivity in amperes per watt as a function of wavelength for

three typical photodetector materials: silicon, germanium and indium gallium
arsenide. The quantum efficiency is the ratio of free electronic carriers produced
to the number of incident photons.

in particle momentum, as indicated in Figure 3.13. In order to generate an

electron–hole pair the incident photon needs to match both the energy and
momentum states available within the semiconductor band structure. This is
particularly evident in silicon – initially the most common photodetector
material – where the narrowest bandgap in energy corresponds to a shift in
momentum also. The absorption process then has two stages. The photon lifts
the electron into ‘no man’s land’, as indicated in Figure 3.13; then the thermal
energy within the semiconductor lattice ‘kicks the electron sideways’ into a
vacant spot in the valence band, in order to match the energy and momentum
released by the photon. In contrast, gallium arsenide (and the many derivatives
thereof; InGaAs, a much used photodetector material, is one) has a ‘direct’
energy gap, so that the absorbed energy from the photon dislodges an electron
from the valence band which goes directly into the conduction band. The
recombination processes follow similar paths, with profound effects on the
output from this recombination process, which we shall explore later.
There are also two major groups of photodiode detectors that depend on the
operation of the intrinsic region. Thus far we have implicitly been discussing
the p–i–n photodiode detector, version, with heavily doped p and n regions
sandwiched between an intrinsic layer, as illustrated in Figure 3.11(a). The
intrinsic region is easily swept free of carriers, and so it is the major optical
absorption zone in the photodiode. The electric field in this region can be
40 Light Interacting with Materials

(a)
INDIRECT band gap
e.g. silicon

Conduction
bands
Energy
Band Absorbed photon to electron–
gap hole pair in two steps

Valence
band

Momentum
Zero

(b)
DIRECT band gap
e.g. gallium arsenide

Conduction
bands
Energy
Band Absorbed photon to electron-
gap hole pair in single step

Valence
band

Momentum
Zero

Figure 3.13 Band structure diagrams for (a) indirect-gap and (b) direct-gap
materials.

raised to a relatively uniform and high enough level to ensure the fastest
possible removal of any free carriers which have been generated. However,
there is only one electron–hole pair for each absorbed photon.
Also shown in Figure 3.11 is the principle of the avalanche photodiode.
Here we again have a short p+ region through which, as with the p–i–n,
photodiode the input photons arrive, followed by a relatively low-field intrinsic
region. However, immediately before the n+ substrate contact region there is a
short, carefully controlled, p+ type region. An appropriate voltage applied to
this structure creates a high electric field zone in the region just before the free
carriers (here electrons) arrive at the n+ (positive) contact. The design of these
diodes is quite tricky, since the idea is that in the high-field region the electric
field is just sufficient to cause controlled multiplication (but avoiding total
 3.3 The Diversity of Optical Absorption 41

(a)
Hole diffusion
leaves negative charge
p+ thin top layer
Lightly doped n region

Electron
n+ substrate diffusion
leaves
positive
charge

(b)

Current flow to external load

Photon-
induced
hole flow
Inject photons

Energy > Eg Photon-

induced
electron
flow

Current flow from load

Figure 3.14 (a) An unbiased p–n junction, showing the origins of the built-in
voltage. (b) This is changed by providing an external load and injecting photons of
energy exceeding the bandgap Eg to produce a flow of carriers for a solar cell. In
both diagrams the lightly shaded arrows indicate natural diffusion into the
depletion layer.

breakdown) of the input electrons arriving from the intrinsic region, therefore
increasing the detected current by a controlled factor. There is inevitably some
statistical variation within this process, as with a photomultiplier, so the
avalanche diode is subject to additional current fluctuations. These are appar-
ent as additional noise. However, the extra sensitivity garnered from the use of
this process often far outweighs the noise penalty. The light sources them-
selves also carry their own inherent and inevitable noise component, as will be
discussed in more detail in Chapter 5.
Thus far, we have briefly explored the action of the p–n junction as a
photodetector when operating under reverse bias. However (Figure 3.14), even
when unbiased from external sources, a p–n junction sets up its own depletion
layer and its own in-built, albeit small, reverse bias: the free electrons in the
n-type material diffuse into the p-type material and the holes move vice versa.
42 Light Interacting with Materials

This zero-bias voltage is typically in the region of a very significant fraction of

one volt. Suppose that now we have light shining on the depletion region.
Again there will be some electron–hole pairs generated within the depletion
region and a new equilibrium state will be established. This equilibrium state
becomes particularly interesting and, indeed, highly relevant if, as indicated in
the figure, we connect a resistive load around the outside of the illuminated
semiconductor device. The new equilibrium state will then establish itself as an
electric current proportional to the photon flux absorbed within the junction.
The p–n junction has now become a solar cell, an increasingly important
device for the generation of clean electrical energy.
This absorption of photons to create free electron and/or hole carriers in
semiconductor materials, has another manifestation even when no external bias
is applied and there is no p–n junction. The electron–hole pairs produced by
photon absorption generate an electric field in the semiconductor material and
that field will persist until the carriers recombine. Thereafter what was origin-
ally photon energy will be released as heat. However, whilst the electric field is
there, both the field itself and the presence of the additional carriers in the
material introduce tangible mechanical forces (nothing whatsoever to do with
the thermal forces discussed earlier). This leads to a phenomenon known as
photostriction, in which a mechanical strain is introduced that is dependent
upon the incident photon flux and also on the absorption depth within the
material sample, resulting in a greater force on the illuminated side. In the case
of silicon this photostrictive strain is negative; the material shrinks very
slightly when illuminated. In germanium the effect goes in the reverse direc-
tion. The photostrictive effect has also been observed in many other materials
not commonly thought of as photoelectric. Of these the most important class is
the perovskites, which are more commonly associated with piezoelectricity,
the generation of electricity by pressure on a crystal (and also, more recently,
solar cells). However, light does introduce free charge carriers in these struc-
tures, and these free charge carriers modify electric fields and thereby chemical
bond strengths in ways which are varied and quite complex. In all cases the
photostrictive effect is relatively small, so sensitive mechanical structures need
to be devised to observe its impact.
Thus, the absorption of light in materials has diverse and practically important
results. The process can generate heat, with implications ranging from solar
furnaces to laser generated ultrasound. It can also generate free electrical
carriers – electrons and holes – resulting in electric currents that can be used
both for measuring light and for generating electricity. The free carriers can also
create mechanical changes simply by their presence, through electrostatic forces
interacting with the overall material structure and also with the chemical bonds
 3.4 Light Emission 43

within the structure. Furthermore, the wavelength dependence of absorption in

solids has enormous application in the spectroscopic analysis of materials and
material structures. Additionally, there are some very interesting consequences
when the absorption process is viewed in reverse, namely optical emission
phenomena, which we shall briefly cover in the section which follows.

3.4 Light Emission

Light emission from sources other than the sun, invariably associated with
thermal emission, has been an important contributor to human evolution for
half a million years or thereabouts. From a spectroscopic perspective such light
has properties determined by the ‘blackness’ of the burning body, its tempera-
ture and the spectroscopic photoemission properties of the material which is
burning. For example, the classic ‘table salt in the gas flame’ demonstration
graphically illustrates the yellow sodium lines.
A perfect ‘black body’ is an idealisation of a structure which perfectly
absorbs all electromagnetic radiation at all frequencies. Since it reflects
nothing, it is black. When heated to an absolute temperature T kelvins, this
black body produces a spectral density measured as the power per unit area
radiated per unit solid angle, per unit wavelength, Bλ (λ, T). This is given by
Planck’s law:

2hc2  
Bλ ðλ; T Þ ¼ in W= sr m3 (3.5)
λ ½ exp ðhc=λkB T Þ  1
5

with spectral characteristics as shown in Figure 3.15. The visible spectrum is

also indicated. The position of the peak in the black body spectrum is deter-
mined by the Wien displacement relationship,
2900
λmax ðT Þ ’ (3.6)
T
for λ in microns and T in kelvins.
This peak in the radiation spectrum has given rise to the concept of the ‘colour
temperature’ of a source of light. The sun, for example, has an effective colour
temperature of 5778 K, corresponding to a peak radiation at about 500 nm – in
the green. If in contrast, we look at the earth’s surface, at typically 290 K, then
the peak radiation at this temperature occurs at about 10 µm. By coincidence,
carbon dioxide gas has a strong absorption line at around this wavelength.
Consequently, radiation emitted from the ‘black body’ earth is absorbed by
the carbon dioxide gas and this absorption manifests itself as heat. The more
44 Light Interacting with Materials

14
Visible spectrum
12
Spectral density (kW / (sr m2 nm))
10

8
5000 K
6

4000 K
2
3000 K
0
0 0.5 1 1.5 2 2.5 3
Wavelength (microns)

Figure 3.15 Planck’s formula for black body radiation from sources at various
temperatures with the visible spectrum superimposed.

carbon dioxide, the more the absorption; hence global warming. Furthermore,
carbon dioxide is not the only gas which absorbs within this spectral range –
atmospheric methane is another contributor, with a far higher (by a factor >20)
absorption cross section, but with a somewhat shorter stable lifetime and lower
concentration than that of the CO2 in the atmosphere.
As electricity became more available and understood, the concept of passing
electric current through a filament, thereby heating it to produce light, grad-
ually emerged, with early demonstrations from Alessandro Volta and Hum-
phrey Davy consolidating into Edison’s carbon filament lamp, enclosed in a
weak vacuum to prevent burning of the filament. This well-known invention
emerged into everyday use for over a century. It was the best readily available
light source despite its inevitably inefficient (Figure 3.15) conversion of heat
into visible light: most of the radiated energy appeared in the infrared, heating
the immediate surroundings. The colour temperature of an incandescent elec-
tric light bulb is typically around 2000 K, which is less than half that of natural
daylight, and corresponds to a peak emission wavelength of about 1.5 microns.
Technically, in spite of their huge contribution to human well being over
many years, the vast majority of incandescent (i.e. hot-filament) light bulbs do
not directly exploit the spectroscopic properties of the materials involved in
their fabrication. However, the fluorescent light bulb is deeply rooted in
spectroscopic understanding. Figure 3.16 illustrates the principle of operation
 3.4 Light Emission 45

AC supply
Circuit maintaining breakdown

Tube filled with neon, Starter coil – induces high

lined with phosphor voltage to break down gas

UV lines Radiation from phosphor

from
neon arc

Visible
range

300 500 700

Wavelength (nm)

Figure 3.16 The basics of the fluorescent lamp, where a UV discharge introduced
by electrical breakdown of the gas (neon) in the tube excites the phosphor to
produce visible radiation.

of a fluorescent lamp, indicating the origins of the relatively dull warm-up time
before the lamp becomes fully operational. The effective colour temperature,
which is determined by the properties of the phosphor, is typically around
4000 K, but the spectrum also includes some sharply defined lines. These lines
depend on the combined properties of the gas discharge lamp used to excite the
phosphor, the phosphor itself and the local temperature distributions. A typical
spectrum is also shown in Figure 3.16. This combination of a source spectrum
and a phosphor spectrum, originally conceived over a century ago, continues
to find application in today’s light emitting diode illuminators.
The light emitting diode, based upon a semiconductor p–n junction, has
evolved over the past half century or so into the preferred source of artificial
light, thanks primarily to the greater electrical-power-to-optical-power conver-
sion efficiency and the significantly enhanced overall reliability. The principle
is relatively straightforward (Figure 3.17). First and foremost, the semicon-
ductor must be a direct bandgap system, to minimise ‘non-productive’
electron–hole recombination resulting in phonons and hence heat. The direct-
bandgap system significantly improves the probability of electron–hole recom-
bination resulting in photon emission. The second principal consideration is that
the recombination region must be situated as close as possible to the surface
46 Light Interacting with Materials

Recombination zone

p+ thin top layer

Hole flow from p+ contact Light emitted on recombination

+
Lightly doped
n region - Electron flow from n+ contact

n+ substrate

Figure 3.17 The basics of a light emitting diode, in which free carriers combine in
the lightly doped n region to produce light.

because, by definition, photons generated through recombination could also be

readily absorbed into another electron–hole pair. The structure shown in
Figure 3.17 comprises a highly doped n+ type substrate material with a shal-
lower, much less doped, n-type layer and a thin p+ top electrode.
When this structure is forward biased (that is, the n+ type region is negative
with respect to the p+ type region), the electrons emerging from the n+ region
enter the lightly doped n region and there recombine with the holes injected
from the p+ contact. The recombination process produces photons at a wave-
length determined by the bandgap of the semiconductor concerned and over a
wavelength range determined by the available energy level distributions in the
conduction and valence bands of the semiconductor. In most semiconductors
this wavelength range is about 5% of the bandgap. Early light emitting diodes
were based upon gallium arsenide and consequently, since gallium arsenide
has a bandgap of about 1.4 electron volts, were at the extreme end of the red
part of the spectrum. Gradual progress in the synthesis of this class of
semiconductors, designed to increase the bandgap, eventually produced a blue
LED in 1995, recognised by the award of the Nobel prize in 2014.
In principle, the current 20% global electrical consumption in incandescent
light bulbs could be reduced by a factor of five through transfer to this
technology. Like the fluorescent bulb, the LED lamp uses short-wave-length
light (blue and the near ultraviolet) together with a phosphor to produce an
apparent white light, similarly to the principle of the fluorescent tube. In
common with the fluorescent bulb, much of the visual acceptance of white
LEDs depends upon the detailed design of the phosphor and the way in which
it is mounted. The phosphor will allow through some of the original blue
 3.4 Light Emission 47

One solution:
Intermediate-
Refractive index ~ 3.5 index
Critical angle ~ 17 degrees coating/lens
Radiated light

Lost light

Figure 3.18 (left) Light lost through total internal reflection and (right) a typical
solution comprising a surface mounted lens.

whilst generating the longer-wavelength components of white light to produce

acceptable illumination.
The light which is generated within the semiconductor material in an
LED can radiate randomly in all directions. This indicates one of the basic
properties of simple LED structures, namely that, thanks to the principle of
total internal reflection, any light incident on the interface between the semi-
conductor and air (Figure 3.18) at an angle exceeding about 17 (the refractive
index of LED materials is typically 3.5 to 4) is reflected back into the structure,
where it will eventually be lost due to absorption processes within the semi-
conductor material. This in turn has led to a variety of ingenious mechanisms
designed to circumvent this significant loss; these mechanisms are based
upon combinations of lenses, coatings and gratings and result in reasonable
efficiencies for LEDs, now typically ~ 100 lumens per watt, and continuing to
improve.
This measure of efficiency is, in itself, somewhat strange. The lumen is a
physiological unit based on perceived visual intensity. The watt is the input
electrical power. Strictly then, any such efficiency values should take into
account this somewhat subjective physiological average, which incorporates
the visual sensitivity of the eye. This in turn (Figure 3.19) is very different both
between individuals and in bright daylight hours (cone vision) compared with
the gathering darkness (rod vision). Daylight vision peaks in the green, and
there is now an internationally agreed equivalence between lumens and watts,
defined for a specific wavelength of 555 nm, for which 683 lumens is taken as
equivalent to 1 watt of light at the same wavelength. A value of 100 lumens per
48 Light Interacting with Materials

Need about three times the optical flux density

in the dark to give same perceived light level

1
Relative Dark- Daylight-
sensitivity adapted adapted

0
400 700
Wavelength (nm)

Figure 3.19 The visual sensitivity of the eye showing the shift to the blue for
dark-adapted rod vision. Also, the rods and cones have different spatial distribu-
tions on the retina, giving different resolutions in bright sunshine compared with
those in the gathering dusk.

watt then corresponds to a conversion efficiency of about 15% , far higher than
the conversion efficiency of an incandescent bulb.
There is much more to understanding human vision than these comments
can provide. However, the different roles of rods for low light vision and cones
for higher light levels, and their different spectral sensitivities, provide some
clues to the roles of real and artificial illumination in human lifestyles. In
particular, the relatively low-level ‘warm glow’ of incandescent lamps is
arguably conducive to a much more restful frame of mind than the much
brighter blue-emphasised light from LED illumination. Indeed, as LED ‘white
lights’ becomes more widespread, there is an increasing awareness that such
illumination not only tends to disturb sleep patterns but also produces visual
confusion when one is moving between dark areas and bright areas illuminated
by LED streetlamps. There is much to learn on optimising the balance between
spectral distribution, brightness and human physiological and psychological
reactions in bringing this application into broader acceptance.
The LED has, however, found wide acceptance after a modest start in
calculator displays of just a few digits. Not only is it evolving into a preferred
form of illumination, it has also found its place in display technology, from the
huge, on the football field, to the tiny, in hand-held devices. The recent
emergence of the OLED (organic LED) has added flexible screen technology
to the list of possibilities.
There is an important variation on the LED. Thus far we have looked at
surface emitters and acknowledged the impact of total internal reflection. If we
 3.4 Light Emission 49

can confine the radiation within a waveguiding region and arrange for it to
come from just one face then, in principle, a much more efficient device is
realised. Additionally, if we make the system sufficiently small in cross
section, it becomes compatible with optical fibres; the edge-emitting LED is
an important light source in fibre communications.
Suppose that we make a relatively minor modification to the basic LED of
Figure 3.18 and coat both ends with a wavelength-selective mirror with a very
high reflection coefficient (say 99%). This will send nearly all light which is
endeavouring to escape from that end surface back into the cavity where the
light is generated. We have to ensure that the mirrors are carefully aligned with
each other – preferably they should be perfectly parallel. In this case any light
which bounces back into the cavity will not escape (Figure 3.20). In doing so
some light will be reabsorbed in the semiconductor between the reflective
surfaces to create yet more electron–hole pairs. Whilst some elections and
holes in such pairs may go their separate ways and be reabsorbed in the
contacts, the majority will recombine to form another photon. This creates
yet more light, in addition to the light created through the ongoing current
injection. But this mechanism only works effectively if the wavelength of
this new photon corresponds to the reflection wavelength of the mirrors.
Within a very short time the light bouncing to and fro will select the specific
resonant frequency dictated by the mirror spacings, and the outcome is the
semiconductor laser. The semiconductor laser takes two basic forms, the
edge emitter and the surface emitter, and is now essential in optical fibre
communication systems, in CD and DVD players and in the ubiquitous laser
pointer.
Laser light has special properties, namely spatial and temporal coherence,
which express the correlation of the optical phase factor across the beam
emitted by the laser and in time. Temporal coherence is often expressed as
a coherence length, along the direction of propagation of the beam. The
coherence time refers to the phase correlation along the beam. The coherence
time is very short for an LED and much longer for the laser. It is equivalent to
the frequency linewidth of the laser oscillator, which quantifies the output
frequency jitter over a long observation period compared with the coherence
time. It is a reasonable assumption that the coherence time is the inverse of the
linewidth in frequency units, though there will be slight variations on this
depending on the detailed definitions applied.
Spatial coherence is very low in a light emitting diode, in which individual
photons are generated from randomly selected energy level differentials in a
emitter and emerge in random directions. However, in a laser the light emitted
from the cavity has been selected in both frequency and direction by the
 Emitted
laser
beam
Emitted laser beam
Highly
reflective
surfaces Spatial
Temporal Coherence Coherence
Is the optical phase Is the optical
distribution along phase distribution
the beam predictable? across the beam
predictable?

Figure 3.20 (left) Turning an LED into a laser using reflective surfaces and (middle and right) the concepts of temporal and spatial coherence.
 3.5 Materials with Controllable Optical Properties 51

Electrical or optical ‘pump’ to excite active medium

High reflectance mirrors tuned to laser

wavelength

Lasing output

Optically active gain medium

Figure 3.21 The generic schematic of a laser system using external excitation
into an optically active gain medium, with feedback from high-reflectivity
wavelength-selective mirrors.

characteristics of the resonator itself and the optical-gain medium therein.

Consequently, the light emerging is very well correlated in phase across the
emitting face; hence the concept of a coherence area. In most laser systems the
coherence area is identical to that of the emerging beam. Figure 3.20 illustrates
the concepts of spatial and temporal coherence.
The same mechanism – surrounding the optically active medium by a
suitable resonant cavity – can be applied to any light source which depends
upon energy level transitions rather than on thermal radiation, to create a lasing
system (Figure 3.21). Electric arcs can excite radiation from gases, and a very
bright white light can be absorbed by some materials to produce a specific
fluorescent output. The act of enclosing such systems in a suitable mirror
structure (selected to the fluorescent wavelength) can give the conditions for
lasing. This is exactly analogous to providing a tuned capacitance–inductance
electronic circuit around an amplifier in order to inject some of the output into
the input and produce an electronic oscillator. We shall encounter lasers many
times during the remainder of this book.

3.5 Materials with Controllable Optical Properties

The optical properties of materials depend on two basic sets of parameters. The
first set comprises the properties of the individual molecules within that
material, which can be understood either through the Clausius–Mossotti rela-
tionship or through quantum mechanics and energy levels. The second set of
parameters comprises the density of the molecules and their orientations with
respect to each other. All these optical properties are essentially encapsulated
52 Light Interacting with Materials

Mechanical force

Figure 3.22 How a mechanical force can modify the refractive index of a solid –
here shown with a regular molecular structure – by forcing atoms closer together
or further apart. Generally, any anisotropic force distribution will, as shown,
produce birefringence. External electric fields can produce similar effects in some
types of crystalline solid.

in the refractive index, and as we have seen, the greater the molecular density,
the greater the refractive index at a particular wavelength. There is, then, one
obvious means for changing the index of a material and that is to simply
compress it. We have already seen this stress effect in the behaviour of gases
under pressure. In solids it is sometimes called photoelasticity and character-
ised through a stress-optic coefficient.
The stress effect in solids depends upon the way in which the forces are
applied. For example, an isotropic pressure brings the molecular structure
closer together in all directions whilst a force applied from a single direction
(Figure 3.22) compresses the structure in the direction of the force but, as
expressed in Poisson’s ratio, causes expansion in the other directions. Conse-
quently, light polarised with its electric vector in the direction of the compress-
ing force will see an increase in the index, whilst light polarised orthogonally
will see a decrease. Thus birefringence is induced.
This effect could, for example, form the basis of an acousto-optic phase
modulator. Here, an optical beam passing perpendicularly to the direction of travel
of an acoustic wave in a transparent liquid or solid, and with dimensions in all
directions significantly shorter than the acoustic wavelength, will see the index
oscillate as the acoustic wave passes through the optical beam. Consequently, the
optical beam will experience a variable delay at the acoustic frequency.
By the same token, if we pass an acoustic wave through a relatively large
transparent material extending over many acoustic wavelengths we have a
travelling periodic variation in refractive index. If we then launch a wide
optical beam (compared to the acoustic wavelength) across this structure then,
via the acoustic wave, we have the basis of an electronically tunable diffraction
grating system, of which more in the next chapter.
 3.5 Materials with Controllable Optical Properties 53

Applying an electric field to a molecule or group of molecules can also have

an impact on the refractive index of that group of molecules, but this does
require a ‘cooperative’ crystalline structure and cooperative constituent mol-
ecules. However, the electro-optic effect, as this is known, is a much more
versatile phenomenon than the stress-optic effect. In electro-optic materials
either or both of the real (delay) and imaginary (absorption) parts of the
refractive index of the material can be modified by the application of a varying
electrical signal.
One electro-optic effect which applies to all transparent materials is the Kerr
effect, which has already been mentioned. There are, though, many other
circumstances in which an applied electric field at low frequencies can affect
the optical propagation constants. Of these the best known is the Pockels
effect, according to which the refractive index of a material varies linearly in
response to an electric field. The Pockels effect does, however, require very
specific crystalline structures within the material of interest and these crystal-
line structures are, of necessity, birefringent; that is, they have different indices
for in-plane and out-of-plane polarisation states. Related to this is the possibil-
ity of electro-gyration, which applies to crystalline circularly birefringent
structures (see Appendix 1). Here the circular birefringence can be tuned in a
way that is linearly dependent upon an externally applied electric field. Some
sugars, in solution or when purified as solids, are circularly birefringent;
dextrose is a well-known example.
These phenomena can all be viewed as electrically induced modifications to
the coefficients in the Clausius–Mossotti relationships, which in turn result in
changes in optical properties. These phenomena all depend upon the optical
wavelength and in particular are highly sensitive around material resonances.
These resonances often coincide with peaks in optical absorption, and
consequently there are many routes through which the absorption, of a material
around these resonances can be modified. For instance, changes in absorption
in semiconductors are feasible and, for some materials, it is possible to
electrically modify the perceived colour.
The electro-optic effect is versatile but immensely complex. In contrast with
stress-optical effect, which is evident in all materials, the electro-optic effect is
only ubiquitous in the case of a relatively low-level Kerr effect. The many
other possibilities for changing both the real and imaginary parts of the index
under the influence of an electric field do require ‘cooperative’ material
structures and often imply operation in the vicinity of a material resonance
and/or within a particular structural crystalline configuration. With careful
material selection and structural implementation, this effect can be utilised to
make optical phase or frequency modulators, optical intensity modulators,
54 Light Interacting with Materials

spatial light modulators (as in a digital projector), tunable filters and a host of
other useful devices. There are indeed many possibilities!

3.6 Summary
The most obvious optical property of any material is its refractive index. The
real part indicates the speed with which an optical phase front propagates (at
the phase velocity) through the material and the imaginary part indicates the
inherent absorption losses within the material. The real and imaginary parts
are, in common with any other linear oscillatory structure, linked through
the Kramers–Kronig relationships. These optical phenomena can also be
viewed in terms of quantum mechanical energy levels and photon inter-
actions. Both approaches give useful insights, and the choice of approach
is, in the final reckoning, a matter of suiting the concept to a particular
situation.
These material properties all vary significantly with optical wavelength, as
they also do at wavelengths well outside the optical region. This variation
provides a unique signature and is an indispensable tool in the characterisation
of materials through spectroscopy. There are also non-linear effects, notably,
in Raman and Brillouin spectroscopy, which provide useful signatures that are
independent of the probing wavelength. However, these do require much
higher optical power densities.
The absorption of light in materials also has numerous manifestations:
absorbed light becoming heat via a phonon excitation process; re-emergence
of the light at a lower photon energy in fluorescence and phosphorescence;
electron–hole pair generation and photo-detection; laser driven photo-
acoustic effects. These processes all operate, to some extent, in reverse and
the optical properties of materials can themselves be modified in response to
numerous external stimuli, We have mentioned mechanical forces and elec-
tric fields but anything which can modify the local molecular structure and/or
the binding forces between nucleus and electrons within that structure will
also change the apparent refractive index; consequently, intense heat and
nuclear radiation are two other possible candidates for changing optical
material properties.
Much of photonics is concerned with understanding the optical properties of
materials and applying this understanding to a particular situation. Here we
have only mentioned briefly the underlying theoretical analysis necessary for
detailed design. However, a basic conceptual understanding is essential in
identifying and utilising appropriate optical materials.
 3.7 Problems 55

In practice, this understanding of materials needs also to be complemented

by an understanding of the interaction of light with particular structures which
is the topic of the next chapter.

3.7 Problems

1. (a) The sunlight that reaches the outer atmosphere of the earth has a spectrum
very similar to that of a black body at about 5775 K (see Figure 3.15).
Discuss how the spectrum might appear when it reaches the earth’s
surface and explain any differences, both in spectral structure and in
power density per unit wavelength.
(b) Bright sunshine reaching the earth has a power density of ~1 kW per
square metre. Consider the sunny-day ‘magnifying glass onto paper’
experiment. What would be a reasonable estimate for the power
density in the focussed spot?
2. (a) The value of the Kerr coefficient for silica can be obtained from data
given just after equation 3.3. Using this value, estimate the Kerr-
induced phase change in a total optical path of 50 km of optical fibre
carrying on average 1 mW of power in a circular cross sectional area
8 microns in diameter. The input power is attenuated in transmission –
hence we use the average power transmitted. If we were to model this
more accurately and take the variation around the average into
account, would the total perceived phase delay increase, decrease or
stay the same – and why? (You may need to look up an appropriate
value for the fibre attenuation.) Additionally, the power density will not
be uniform across the fibre cross sectional area; the typical fibre mode
shape is roughly Gaussian across the diameter. Using the same logic, will
this increase, decrease or not significantly change the impact of the Kerr
effect from the original ‘uniform in all directions’ estimate?
(b) Suppose two optical signals at different wavelengths are introduced
into a fibre at power levels high enough to give significant Kerr
effect phase changes. What changes might you expect to find in the
spectrum of the output signals, and why? State your criteria for a
significant change.
3. (a) There are two basic approaches to detecting light – as photons and when
absorbed as heat. Light can also be regarded as an electromagnetic field.
For radio waves we use this field to excite movement in electrons in a
wire – why don’t we do this with light? Or could we also do it with
light?
56 Light Interacting with Materials

(b) Compare the basic properties of (i) the spectral response of thermal
detectors (bolometers) and bandgap detectors and (ii) discuss the
factors influencing the intrinsic sensitivity of the two detection pro-
cesses, endeavouring to derive some comparative performance esti-
mates (the reason why photodetection is dominated by photon
detection processes might be a starting point).
(c) Suppose you have a bandgap-based detector material with a bandgap of
1 eV and a quantum efficiency equal to zero above the threshold
wavelength and unity immediately below the threshold wavelength.
Calculate the responsivity in amperes per watt for such a detector over
the wavelength range 0.1 to 2 microns. Explain why this curve looks
nothing like the practically observed values in Figure 3.12 and why
there is a peak in the responsivity in the curve which you have derived.
(d) You are given the task of optimising the sensitivity of a detector system
to visible light (say to 300–800 nm wavelengths, which goes a little
beyond the visible at both ends) that is focussed down to a spot. You
have no limitations on technological design for the detector. How
would you approach this?
4. (a) Estimate the percentage of generated light that actually escapes from a
simple planar-surface light emitting diode which produces light through
the full 4π steradians solid angle.
(b) There are lensed versions of light emitting diodes mentioned in the
chapter – design what you feel would be an optimum lens using ray
optics. (Assume that glasses of refractive index up to 2 are available.).
Using the formulae for reflections at interfaces and similar effects given
by the Fresnel equation (see equation 4.3 in Chapter 4), estimate the
efficiency of your modified design. How might you improve on this
design? Does your design include a consideration of the glasses that are
available glasses for the lens? If not, what would the implications be of
replacing your ‘optimum’ glass with the best available?
(c) We have also mentioned lasers in the context of light generation from
semiconductors and gases. The operating mechanism was described in
the text. It involves a combination of a photon bouncing back into the
active region and being absorbed and the consequent excited state in the
material producing another, coherent, photon before it hits the electrodes
in the generation region to become ‘electric current’. Think through this
process carefully and see if you can (perhaps collectively) arrive at a
relationship between regeneration times, lifetimes and feedback percent-
ages in the process: the photon lifetime needs to be matched by the
regeneration time for the steady state to be reached. Look up on the
 3.7 Problems 57

internet the laser ‘rate equations’ to rationalise your thoughts with the
algebra! It can all be found in the description in the chapter, but careful
thinking through the process is essential!.
5. (a) Polarising sunglasses are readly available. Try twisting the frame of a pair
of polarising sunglasses through 90 on a sunny day and explain the
differences you see whilst changing the angle. Then take two pairs and
cross the polarisers so that no light comes through. (You may also have
access to polarising sheets – readily available from many internet sup-
pliers.) Place a transparent plastic sheet between the two crossed polarisers
and stretch it in various directions. Explain the patterns and also the colours
you see.
(b) The above approach is sometimes used as a means for examining
stresses in plastic samples. Usually the results are taken qualita-
tively – how would you make them quantitative?
6. (a) You have been given the task of designing a silicon photodetector. In
silicon the saturation velocity for electrons (roughly speaking, this is the
‘terminal’ velocity at which any increase in field causes no further
increase in velocity) can be taken as 107 cm/s and saturation is reached
at a field level of 20 kV/cm. The ionisation field can be taken as
~300 kV/cm. Assume a 10-micron-long slightly p-doped depletion
region and 1 micron of p+ (see Figure 3.11(b)) In your estimates, use
Gauss’s law to arrive at the doping levels needed to achieve the field
levels for saturated velocity operation from the carriers, and from this
estimate obtain the bias voltage.
(b) What would be the potential maximum detection bandwidth for
modulation applied to the optical signal incident on the structure?
(c) Look up the optical absorption coefficients in silicon and comment, with
reasons, on the wavelength range over which this 10-micron / 1-micron
design is practical.
 4
Light Interacting with Structures

In the previous chapters we explored the essential features of the interactions

between light and homogeneous materials. In fact, we may have cheated a little
by already looking in Chapter 2 at the optical characteristics of a planar
interface between dissimilar homogeneous materials and in Chapter 3 by
discussing the device implications of structures such as photodetectors. The
present chapter is dedicated to expanding this discussion into, initially, non-
planar interfaces, and thereafter examining the behaviour of material structures
and how this varies as the structural dimensions, compared with the optical
wavelength, are progressively reduced.

4.1 Structures with Dimensions Much Greater than

the Optical Wavelength
The simplest possible material structure is a straightforward planar interface
between two different but transparent optical materials, as discussed in
Chapter 2. In Chapter 2 we also mentioned the polarisation dependences of
reflection and refraction at such interfaces. Polarisation is in general an import-
ant basic feature of electromagnetic waves – most noticeable in the optical
domain but present throughout the spectrum. The concepts are explored in
more detail in Appendix 1. The essence of that exploration is that polarisation
states can be manipulated by birefringent components (i.e. components having
polarisation-dependent optical paths) and that any given polarisation state can
be split into components aligned with the principal axes (i.e. each of these
optical paths) of the birefringent medium. These principal axes can relate to
linear states – by far the most usual situation – or to circular states in circularly
birefringent (chiral) materials.

58
 4.1 Dimensions Much Greater than the Wavelength 59

Configurations based on birefringent transmission paths – often involving a

separate component known as a phase plate – have an enormous range of
application in the tuning of polarisation states in order to make specific
measurements on materials, for example. In reverse, they are used in the
analysis of reflected and/or transmitted polarisation states. There are many
possibilities since, for example, one polarisation state may see an entirely
different index behaviour (the index may go through an absorption resonance)
and so can yield important information about the material through which it has
passed.

Prisms and Lenses

Optical devices based upon large material structures are extremely common in
everyday life, from videos screens to the focussing optics in CD players, to
celestial telescopes, to raindrops forming rainbows. The last of these exempli-
fies one very important feature of most interfaces between two different
materials, namely that the refractive indices on both sides of the interface are
a function of wavelength. For most solid materials (glasses for example) in the
visible the blue gets the biggest ‘kick in the pants’(i.e. the largest deviation);
this is often referred to as normal dispersion. The triangular prism exemplifies
this (Figure 4.1).
Usually the prism is seen as a means of dispersing optical wavelengths, as
indicated in the figure. Prisms, however, have other interesting properties,
particularly those with a right-angled isosceles triangle cross section. A light
beam incident perpendicularly on the hypotenuse of such a triangle will go
straight ahead (though some will be reflected), will hit the far face at an angle
greater than the critical angle and therefore will be totally internally reflected
and return in the same direction but displaced by an amount determined by the
prism geometry and the position of the incident beam. Reflecting prisms of this
nature have extensive uses and have also emerged in slightly modified forms to

Red
Parallel white
light input
Blue

Figure 4.1 Dispersion from a triangular prism with a parallel white beam incident
at an angle on one face showing the effect of the different refractive indices for red
and blue.
60 Light Interacting with Structures

Input light ray

Retro-reflecting
mirrors

Reflected ray parallel to original direction

but displaced vertically

Figure 4.2 Retro-reflecting mirrors and a spherical-lens retro-reflector.

Convex lens,
focal length f
1 1 1
Object, + =
Back focal u v f
distance u from lens
point

Front focal
point Image,
distance v from lens

Figure 4.3 The paraxial approximation image relationships for a thin convex lens
illuminated by a parallel beam.

make, for example, compact large-area reflectors and even to form the basis of
the ‘pentaprism’ viewfinder systems in some types of single-lens reflex cameras.
These features of the right-angled prism give this direct reflection effect even
for slight changes in the angle of incidence of the input beam. This back
reflection effect also occurs for two mirrors with their surfaces at right angles
to each other, as indicated in Figure 4.2 – a configuration often referred to as a
retro-reflector. This retro-reflector is somewhat limited, because a more careful
examination shows that it only works for an input beam incident at right angles
to the out-of-plane surface of the mirror. However, this can be made into a much
more versatile system: a corner cube retro-reflector, with reflecting surfaces on
three mutually adjacent surfaces of an opened cube. Taken a step further, a
spherical retro-reflector can effectively cope with a wide range of input angles,
as also illustrated in the figure. These retro-reflectors and several variants
thereon feature prolifically in everyday life, in the ubiquitous cat’s eyes on
roads, in reflective jackets, traffic signs, LED optics and a host of other places.
The lens – typically spherical and convex, as sketched in Figure 4.3 – is
probably the most familiar imaging structure. The figure also indicates the
basic principle of operation through refraction at the input and output faces.
 4.1 Dimensions Much Greater than the Wavelength 61

For the so-called thin-lens paraxial approximation, where all the angles of
incidence and refraction obey the sinθ ~ θ approximation, the lens formula
applies:
1 1 1
þ ¼ (4.1)
u v f
where the object distance u is related to the image distance v through the focal
length f. This shows that, for example, if the object is at infinity (that is, a
parallel beam arrives) then the focal point is also the image point.
For a practical lens to be used in the visible range, an important caution must
be noted. We have already mentioned that the refractive index for the blue part
of the spectrum is typically higher in glass than that for the red part of the
spectrum, so the focal length for blue light will be shorter than that for red
light – giving rise to chromatic aberration. There is also another important
observation – the lens is an aperture and so not only will there be ray
propagation through the lens but there will be diffraction (discussed later) at
the edge points; this diffraction will in turn create a ‘blur’ on the idealised focal
point. The angular spread of this blur will be of the order of the wavelength
divided by the lens aperture D and consequently the blur reduces as the
aperture gets larger. This also leads to criteria for lens resolution, since the
minimum feature size that this lens will be able to discriminate when imaging
will also be of the order of the spot size. Again, this is wavelength dependent,
with better resolution in the blue than the red.
Increasing the resolution – an incessant demand whether for cameras,
telescopes or microscopes – automatically implies increasing the aperture of
the lens, in other words, the ratio of the diameter to the focal length. This
takes us away from the paraxial approximation, with implications shown in
Figure 4.4. For a simple spherical surface, the focal length becomes a function
of the distance away from the lens axis in the manner indicated. However, it is

Back focal Back focal

points: points:
Off-paraxial ray off-paraxial rays paraxial rays
blue
Paraxial ray red Opcal
axis
Nominal front
focal point
red
blue

Figure 4.4 The off-axis impact for a thick convex lens, indicating changes in the
focal length with wavelength and the range of foci for rays incident at varying
distances from the lens axis.
62 Light Interacting with Structures

Parabolic reflector

Input rays parallel to opcal axis

Opcal
All rays arrive at same focal point – regardless of colour and distance from axis axis

Figure 4.5 A parabolic mirror reflector, for which all rays arrive at the same focal
point, regardless of their distance from the mirror’s principal axis. The image
receptor is at the focal point.

relatively straightforward to calculate the ‘aspherical’ shape necessary to

ensure a constant focal point. Aspherical lenses are now commonplace but
they function precisely for only one wavelength and acceptably so for a limited
band of wavelengths around that particular one. All currently available glasses
are chromatically dispersive. There are ingenious corrections for this, typically
involving combinations of glasses with different dispersion characteristics and
also with concave and convex surfaces. The details are involved and complex
and require an intimate knowledge of the range of available glasses and their
optical properties. There are some commercial software packages which ease
the burden, but high-aperture lens design remains a specialised area.

Mirrors
The whole process becomes much simpler using mirrors (Figure 4.5). Mirrors
have been particularly useful in astronomical telescopes since then almost any
object of interest is, to a very good approximation, an infinite distance from the
mirror so all images will cluster around the focal point. The mirror can be made
large, so that a useful imaging area is available without too much masking of
the input light (also, a large mirror both maximises the light collected from
distant stars and galaxies and enhances stellar-image resolution). The accept-
ance angle of such a system is also limited to the order of Di/f, where Di is the
dimension of the image receptor at the focal point, typically in centimetres, and
f is the focal length of the mirror, typically in metres.
If the mirror is to provide a perfect focus, first of all its front surface must be
coated with the reflecting medium, rather than its rear surface. Also, the surface
must be very accurately machined and kept totally clean thereafter. Second,
its shape has to be exactly parabolic to an ‘optically flat’ degree, implying
better-than-wavelength precision! The details of this shape are given
 4.1 Dimensions Much Greater than the Wavelength 63

Incident plane wave

wavelength l
P

A
q
O
B

Figure 4.6 Diffraction from a single out-of-plane slit AB, of width a in a screen.
The first minimum occurs when OP – AP ≏ BP – OP ≏ λ/2.

immediately by the focal length required. Since, as mentioned above, in

astronomical telescopes the mirrors are typically metres in diameter, this gives
an idea of the machining tolerances needed in making such a mirror. Precision
grinding to the necessary parabolic shape is an essential part. Additionally, the
need for extremely precise assembly throughout and very stable operating
conditions (thermal expansion in the mirror materials can cause severe distor-
tions, for example) becomes apparent.
Atmospheric turbulence also needs to be avoided. We have all seen the
effect hot air rising from a black road surface. Such effects can severely limit
the imaging capacity of these telescopes and hence the desirability for locating
them on a high mountain or, even better, extraterrestrially. Furthermore, very
little light is received from distant stars, so the image collection system needs
time to acquire a workable image and the telescope must be stable with respect
to the star (not the earth) as the image is collected. Hence, precision positioning
systems are also mandatory, complemented by precision manoeuvring, to
ensure that the stellar image remains fixed on the image receptor for the entire
observation time.
Material structures which are large compared with a wavelength obviously
have many manifestations in both daily and scientific life, from camera lenses
to bicycle pedal reflectors through precision imaging, whether at the micro-
scopic or galactic scale. There will also no doubt be some readers with
corrective spectacle lenses!
The prism has been a central element in spectroscopic measurement
systems, but recently prisms have been largely superseded by diffraction
64 Light Interacting with Structures

gratings – which we cover in the next section. Diffraction grating spectro-

meters need lenses, however. Overall, our knowledge of both the intergalactic
and microscopic worlds relies upon the understanding and exploitation of the
concepts which have been outlined here.

4.2 Structures with Dimensions of the Order

of the Optical Wavelength
We are now concerned with structures whose dimensions are of the order of
0.1 to 10 times the wavelength of the incident light, which is just a few
micrometres. Arguably, this is the most important dimensional range in the
study of photonics. Even the relatively large structures which we examined
briefly in the previous section inherently rely upon precision interfaces defined
to this tolerance. We mentioned the precision grinding of astronomical tele-
scope mirrors, for example.
It is interesting that domestic window panes are, thanks to the float glass
process invented in the twentieth century, surprisingly flat. On the other hand,
look through the surviving panes of glass from before the twentieth century
and the ripples are immediately obvious. Also – in all cases – the tiniest scratch
on a glass window becomes immediately apparent despite its having dimen-
sions within the micrometer range.
So why is this tiny scratch so apparent? In simple terms the scratch blocks
out the transmitted daylight, so that its presence is evident. This can of course
be described just by the ray model of light. However, if in dark conditions a
laser is used to illuminate the scratch, a pattern of spots due to the scratch will
be seen: the laser light diffracts around the corners of the scratch and an
interference pattern is formed. An interference pattern would also be formed
in daylight, but it would be hard to see it.
The inverse situation to the scratch in window glass is that of a single narrow
slit of width a in an opaque screen illuminated by a monochromatic light source
(Figure 4.6). According to Huygens’ principle, semicircular waves will diffract
from each point on the slit. The waves will interfere with each other. If we travel
far enough away from the slit, say a distance Df, constructive interference will
occur at certain directions. The diffracted beam centred around the first outer
peaks will separate from the straight-through beam. The distance Df is given by:
Df ~ a2/λ (4.2)
(which assumes a small angle of diffraction (for which sin ϕ ~ ϕ), correspond-
ing to a slit more than ~ 10 wavelengths wide). This particular distance Df is
significant. It is known as the beginning of the far-field region for the aperture,
 4.2 Structures with Dimensions of Order of the Wavelength 65

a term used interchangeably in optics, microwaves, and acoustics or anywhere

waves may propagate. This region is also sometimes referred to as the Fraun-
hofer region. It defines a distance from the slit or aperture beyond which the
perceived diffraction pattern simply spreads out in space as if from a point
source (as seen in two dimensions, in Figure 4.6).
To find the zeroes in the single-slit diffraction pattern, imagine that the slit is
divided into half; then, at a certain angle to the axis, the wave travelling from the
top point of the slit will be exactly out of phase with the wave travelling from the
point halfway down the slit. If the angle at which this occurs is θ, then the path
difference between these two waves will be (a/2) sin θ, and this will equal λ/2
since the waves are out of phase. For small values of θ we thus have θ = λ/a.
Since the same reasoning applies to each pair of waves coming from corres-
ponding points a/2 apart on the slit, there is complete cancellation of all the light
in the direction θ, and so zero light intensity is observed on the screen at that
angle. (Dividing the slit mentally into four, six etc., the same arguments apply at
angles 2λ/a, 3λ/a etc., to give subsequent zeroes on the screen). The theory of
single aperture diffraction is important for the lenses used in optical systems.

Diffraction Gratings
We now consider what happens when there is more than one slit. The case of
two slits was mentioned in Section 2.3, see Figure 2.11, and the expression for
the maxima was given by sin θ = nλ/d. Now let us think about a device which
has not two but many evenly spaced slits. Two simple diffraction gratings are
shown schematically in Figure 4.7. Here we see, in an opaque screen, rulings,
servings as slits, with very sharp edges which scatter the light in all directions.
As for two slits, complete constructive interference occurs every time the path
difference between light diffracted from adjacent slits is an integral number of
wavelengths. For wavelength λ and slit separation d, this again occurs at angles
θn given by
nλ
sin θn ¼ (4.3)
d
Since we have assumed that the diffracted light is equal in intensity in all
directions, then all these image lines, at angles θn, would be of equal intensity.
In the region where sin θ ~ θ the angle θ is related linearly to the inverse
spacing, 1/d, of the grating (i.e., the spatial frequency per unit length of the
grating) and the images are all of equal intensity. For clarity, only the first-
order angle of diffraction for each structure is shown in the diagram. The
central observation here is that the angular dispersion of a grating near the axis,
66 Light Interacting with Structures

First order beams diffracted by the object – the through beam is omied for clarity

Object

Illuminaon beam

Figure 4.7 The finer the detail in an object the wider the angle of diffraction and
therefore the wider the aperture of the lens required to collect that detail.

where sin θ ~ θ, is in fact a Fourier transform of the spatial grating pattern;

here it is a good approximation to a row of periodic delta functions. This is not
a coincidence, since it is relatively straightforward to demonstrate that in the
paraxial region this Fourier transform relationship between any structure such
as a grating and its far field is universally applicable. (This is discussed further
in Appendix 2.) This Fourier transform relationship is an extremely powerful
tool, used in many situations in photonics.
Thus far, we have assumed that the grating has infinite extent (aperture), but
of course it is not actually infinite, so what about the edges? First of all,
consider an aperture similar to that indicated in Figure 4.6 – a wide slit aligned
vertically to the plane of the diagram. Illuminating the aperture corresponds to
a single simple pulse and its Fourier transform is a sinc function. Then consider
the spacing of the zeroes in the sinc function. Again, for small diffraction
angles, as we saw easlier the zeroes are given by θ = mλ/a, where m = 1, 2, ...
and a is the aperture (width) of the slit.
Suppose this aperture is now used in front of an infinite square wave grating.
In effect we are multiplying the image spaces together, so consequently we
will get a convolution in Fourier transform space, as indicated in Figure 4.8.
This relationship has many implications, of which perhaps the most obvious
lies in the use of a microscope as an imaging system. The general principle
here is that the object is placed in the focal plane of the objective lens to
produce eventually an image at infinity in the eyepiece. In order to resolve the
fine detail in the object the imaging system has to collect as much of the light
diffracted from the object as possible. Consequently, the larger the ratio of the
diameter of the input objective lens to its focal length, the more detail it can
collect and the higher the resolution of the microscope. This is another facet of
the ‘blurring’ due to finite aperture size mentioned earlier.
 4.2 Structures with Dimensions of Order of the Wavelength 67

Line diffraction
pattern for very
Square wave wide aperture
diffraction grating Sinc2
diffraction
pattern

Line pattern
convolved with sinc2
diffraction pattern

Figure 4.8 The diffraction pattern from a square wave diffraction grating and its
modification after passing through a finite but wide aperture (see text). Also
indicated, on the right, is a spatial filter, a narrow aperture allowing through only
the central three diffraction orders – what would be the image following this
spatial filter?

These observations also point toward techniques for image processing and
image enhancement, much of which take place by manipulation of the spatial
frequency content of the image using optical elements (or even software tools
on digital images) in the so-called Fourier transform plane (Figure 4.9). This is
in the focal plane of the imaging objective; here the diffracted beams from the
object are focussed to produce an image of the spatial frequency spectrum
(strictly speaking, a squared image, since what we see is intensity rather than
amplitude). A basic example was given in Figure 4.8. Here we started with a
square wave grating and producing the Fourier transform distribution as indi-
cated. Each of the image lines has an amplitude sinc function around it arising
from the finite aperture of the lens and/or grating. Suppose we then insert a
spatial filter (narrow aperture), shown in Figure 4.8, which allows through only
three beams, corresponding to the zeroth-order and first-order diffracted beams.
The image no longer fully represents the original. And what would the resulting
image be if this spatial filter is placed in the front focal plane of a second lens?
It can be shown that if on the other hand we attenuate the central (low spatial
68 Light Interacting with Structures

Diffracted beams at many angles depending on structure

Illuminating laser beam

Fourier transform in back

focal plane, definition limited
by convex lens aperture

Object in front focal plane

Figure 4.9 Collecting the diffraction pattern from an object in the front focal
plane of a lens produces a focussed representation of the periodic components as
the image in the back focal plane. Also, note the impact of the lens aperture, which
cuts off the higher angles of the diffraction pattern.

frequency) beam with respect to the higher-order beams then the result is edge
enhancement – a sharpening of the image definition. Spatial filtering, by means
of aperture, attenuating screens, or even spatially varying phase delays (see
below), is a frequently used tool in image enhancement.
The Fourier transform relationship between an object and its far-field diffrac-
tion pattern (or between the input plane and the transform plane in a simple
convex lens system) also applies to transparent objects, many of which have
structure within them which is invisible in a conventional image. However, this
structure will correspond to variations in optical thickness – that is, optical
phase and phase functions also have Fourier transform properties. For example,
a sine wave variation in the optical thickness of an object (which corresponds to
a spatially varying phase delay) gives rise to a set of sideband frequencies
which recombine to produce a zero-contrast image, with no intensity variations.
In the simplest case we can remove the central background component – the
average transmission through the object – to obtain the frequency-doubled
intensity distribution indicated in Figure 4.10(a). Whilst this is distorted, it
can be interpreted by an experienced observer. This technique is known as
dark-field imaging. You may like to consider why in Figure 4.10(a) the image is
spatial-frequency doubled, as indicated, with a zero every half-period of the
original sinusoidal phase grating.
There is also a related technique, known as the phase contrast method,
through which the phase of the zero-frequency component is shifted by 90
(this will only apply exactly for monochromatic illumination. Useful results
 4.2 Structures with Dimensions of Order of the Wavelength 69

(a) (b)
Intensity variaon:
spaal frequency
Intensity variaon
doubled in image plane
in image plane
Transparent phase
object

Front focus
On-axis stop
at rear focal point
Quarter wave delay
on-axis at rear
focal point

Figure 4.10 Making phase changes in a transparent object visible: (a) using an
on-axis stop to produce a frequency-doubled intensity image or (b) using an on-
axis quarter-wave delay plate to produce an intensity image of the original phase
distribution.

can, however, be obtained over a relatively wide range of wavelengths – the

reader might like to estimate what this range might be). This technique can
also, for small phase deviations, produce an amplitude image which is directly
related to the original ‘phase object’.

Three-Dimensional Imaging
The above brief discussion contains much conceptual food for thought. It also
points towards the principles of holography – in effect three-dimensional
imaging without lenses. Here light scattered from the object (Figure 4.11) again
forms a diffraction pattern but in this case the diffraction pattern is combined
with a larger sinusoidal phase-reference beam extracted from the same illumin-
ating laser. The interference between the diffraction pattern and the laser beam
produces an intensity distribution containing sinusoidal components exactly
related to the diffraction angles, and therefore containing all the three-
dimensional information about the illuminated areas of the original object.
Consequently, recording this pattern to make a permanent hologram and
subsequently shining a laser onto this modified diffraction pattern retraces the
original paths and the viewer sees a three-dimensional image. Need this laser be
the same wavelength, and why do we need the high-intensity reference?
Imaging is among the most important manifestations of photonics. This
section has highlighted most of the essential underlying principles of imaging
systems. Some demonstration experiments with everyday objects – a magni-
fying glass, a laser pointer and possibly some simple samples and apertures
70 Light Interacting with Structures

Photographic plate records

interferogram containing
amplitude and phase of
the scattered light:
the hologram Rays scattered from object

Reference beam

Parallel
input
laser
beam
Illumination for object

Figure 4.11 Making a hologram on a photographic plate, shown at top left of the
diagram.

Reflected wave –
Incident wave at constructive interference
correct angle necessary from all
contributory waves

Thick phase grating

(e.g. ultrasonic wave in water)

Figure 4.12 The essential concepts of Bragg diffraction. A three-dimensional

grating is formed by a set of crystal planes on by a high-frequency acoustic wave,
which forms a periodic variation in refractive index.

which can illustrate these principles – are outlined in the problems at the end of
the chapter.
In preceding discussion we implicitly assumed that all the diffracting objects
were thin compared with the wavelength of the light which is being diffracted.
Thick gratings – phase objects many optical wavelengths in thickness – also
play their part; this is exemplified in Bragg diffraction (Figure 4.12). In optics
the Bragg grating is typically a phase object structure which can be created by,
for example, the pressure variations induced by a travelling ultrasonic wave.
 4.2 Structures with Dimensions of Order of the Wavelength 71

White
input
light

Oil film
Incident wave
Water

Figure 4.13 (left) The oil-on-water effect embodying multiple reflections from the
surface and the oil–water interface and (right) the incident and reflected beams for
normal incidence.

For an incident beam at the appropriate angle and an appropriate total phase
delay within an ultrasonic-wave grating, the entirety of the incident optical
beam can be diffracted as indicated in the diagram. Since the ‘grating’ is
moving, the diffracted beam is Doppler shifted in frequency by an amount
identical to the input frequency of the ultrasonic driving signal, a feature which
is often used in optical frequency shifters.
Finally, another observation on diffraction gratings. Suppose a parallel beam
of white light, rather than monochromatic light, shines upon a diffraction grating
of the type shown in Figure 4.10. In this case the blue component will be
diffracted rather less (by a factor of 2 or so) than the red component; therefore,
by appropriate spatial filtering, the apparent colour of this grating can be made to
change – the first hint that perceived colours depend not only on spectroscopic
properties, as discussed in Chapter 3, but also on the structural properties of both
a material itself and the system through which it is observed.

Thin Films
A commonly observed example of this is the coloured structure of a film of oil
on water. Figure 4.13 illustrates how this might happen. Oil and water have
different refractive indices and therefore both the interface between the oil and
the water and the interface between the oil and the air will introduce partial
reflection, with amplitude reflection coefficient R. At each interface, at normal
incidence R is given by the Fresnel equation, namely
n1  n2
R¼ (4.4)
n1 þ n2
72 Light Interacting with Structures

where n1 is the refractive index of the medium from which the light is incident
on the interface and n2 is the refractive index of the medium into which it passes
(this equation is a special case of the Fresnel reflection relationships mentioned
in Appendix 6.2). Note that this equation 4.3 implies a phase change of 180 for
light incident from a low-index material to a higher-index material. The value of
the reflection coefficient R (remember that it is an amplitude), for light travelling
from air to glass, where the index of the latter is 1.5, is around 0.2, correspond-
ing to 4% power reflection. The situation for the case of oil and water is slightly
different, in that a typical oil has an index of approximately 1.4 whereas water
sits at 1.33. Consequently, there is no phase inversion at the oil-to-water
interface and, typically, the amplitude reflection coefficient for normal incidence
on the oil-to-air interface is about 0.14 compared with –0.1 (the negative sign
indicates 180 phase reversal) at the water-to-oil interface. However, reflections
from the two interfaces will interfere, with a range of possible results varying
from constructive interference, producing a total net reflected amplitude of 0.24,
to destructive interference, producing 0.04. This is a 30 to 1 ratio in perceived
optical intensity (we have ignored multiple reflections since these are relatively
small though they will have some impact). The visible spectrum covers around a
factor of 2 in wavelengths so the above discussion indicates that, depending
upon the oil thickness and the angle of observation, the reflected colour will vary
throughout the spectrum – another example of where the spectroscopically
defined colour of a material can be radically changed by its structure.
Thin films and the design and realisation thereof constitute a vast topic.
Antireflection coatings are a common application. If we arrange for the two
reflected beams to be out of phase (this happens if the film thickness is one
quarter of a wavelength and the reflection amplitudes are identical for each
interface) then the resulting destructive interference at the surface implies zero
reflectance. This concept finds widespread applications. Camera lenses, spec-
tacle lenses and a host of other optical surfaces typically exhibit a bluish tinge
when viewed from an appropriate angle. This is a manifestation of anti-reflection
coatings. These are incorporated to minimise the impact of reflections on image
quality typified by the glowing, often yellowish, circles seen in photographs
taken when the camera is pointed towards, but not directly at, the sun.

Waveguides
Figure 4.14 shows the simplest possible implementation of an optical wave-
guide, a step-index fibre. Here, provided the light is incident on the interface
between the core and cladding at an angle exceeding the critical angle, the light
will be guided through the higher-index core region. Typically, light would
be launched from some external source – often air. There will be a maximum
 4.2 Structures with Dimensions of Order of the Wavelength 73

Higher-index
core
(a) (b)
Lower-index
cladding

Figure 4.14 A step-index optical waveguide showing, in the main part of the
figure, ray paths for two different modes of propagation and, on the right, the light
amplitude distributions for (a) the lowest-order mode and (b) a slightly higher
mode, corresponding to interference between the beams at steeper angles of
incidence. See Appendix 4, Figure A4.2.

angle that the incident launched light beam can make with the axis of the
waveguide, beyond which guiding cannot take place. This angle corresponds
to the critical angle for light travelling from the core to the cladding. The
acceptance angle θna is given by

sin θna = (ncore2 – nclad2)1/2 ~ (2nΔn)1/2 (4.5)

where n is the average refractive index of the core and cladding. The quantity
sin θna is often referred to as the numerical aperture.
The approximation in equation 4.4 is often applicable since the index
difference Δn is typically small compared with the indices themselves. (It is
straightforward to verify this from Snell’s law relating the on angle of inci-
dence and the angle of refraction.) This basic structure forms the basis of the
optical fibres used to interconnect domestic electronic appliances or to provide
short-distance low-capacity communication links.
One limitation of these optical waveguides becomes apparent when one
observes that there are two extreme paths which the light rays can take and a
whole host of other paths in between: the path difference between a central-
axis ray (the lowest-order mode) and a ray reflected at the critical angle can be
estimated as lΔn, where l is the waveguide length. For index differences of 1%
and time delays of 5 μs per km, which are typical, this path difference
corresponds to ~50 ns of dispersion over a 1 km length of waveguide. For
monochromatic light this dispersion is purely an artefact of the structure of the
waveguide. Thus pulses of 20 ns duration (about 20 Mbits/s – slower than
many domestic broadband speeds) will be smeared out completely within less
than a kilometre and therefore lose all value. However, these step-index fibres
are very simple to produce and can be made from plastics as well as glass, and
so for short distances they find extensive application.
Perhaps one way to solve the problem of dispersion is to design a structure
which has only one propagation path, leading to the concept of the single-mode
74 Light Interacting with Structures

30
4
20

3 10 Material
0 dispersion
Attenuation (ps/km)
2
(dB/km) –10

1 –20

–30
0
0.5 1 1.5 2
Wavelength (microns)

Figure 4.15 Losses (solid line) and material dispersion (broken line) in silica. The
zero dispersion point at 1.3 µm is a small, but significant distance away from the
lowest attenuation at 1.5 µm wavelength. See Section 5.7.

optical fibre. These fibres have been extensively used for over three decades
in long-distance optical fibre communications and among other things are essen-
tial in fuelling the even-increasing demand for domestic and industrial band-
width. The internet could not function without optical fibres, as one of many vital
contributing technologies. Even the single-mode fibre, though, has its inherent
limitations. There is dispersion of different wavelengths in the glass which forms
the fibre as well as the dispersion introduced by the optical structure; longer
wavelengths penetrate further into the cladding via the evanescent wave (see
below). Furthermore, there are spectroscopic wavelength-dependent losses
within the materials themselves. The losses and the dispersion characteristics of
silica (the principal raw material for optical fibres) are shown in Figure 4.15.
The basic operating mechanism of the single-mode fibre points towards a
wave rather than a ray interpretation of transmission along a high-index region
surrounded by lower-index regions. In this wave interpretation, only a limited
number of ray directions are permitted even in the large-scale case of
Figure 4.14. The reason is that across the interface between the high-index
region and the lower-index outer region there must be continuity in the electric
field. In the case of a perfect metallic interface this would imply that the electric
field is zero at the interface but at a dielectric interface the continuity is a little
more complex and requires some limited penetration of the electric field into the
low-index region; this is often referred to as the evanescent tail.The net result is
a field distribution of the type shown in Figure 4.14(b). For the two-dimensional
case (in other words, a film enclosed between two other films), this can be
considered as interference between two beams at a specific angle (Figure 4.14
(a)). In principle there are other options which would fulfil the same basic
 4.2 Structures with Dimensions of Order of the Wavelength 75

criteria, namely that at the boundary between the dielectric interfaces the
appropriate field relationships hold. An example of such a higher-order mode
is also indicated in Figure 4.14(b). This mode has more than the minimum half
wavelength of an interference pattern across the core region. However, in order
to achieve this second interference pattern, the imaginary beams which are
interfering need to be incident on the interface at a higher angle. For a single-
mode fibre, this angle would be chosen to exceed the critical angle and therefore
the higher-order mode would not be guided. Hence a combination of index
difference and core dimensions gives rise to the modal structure of the fibre. The
circular cross sectional structure characterising an actual fibre involves slightly
more complex beam structures but the underlying concepts are identical. The
basic idea of viewing guided waves as interfering beams for which the interfer-
ence pattern satisfies the necessary boundary conditions along the wave-guiding
interface gives a very useful insight. Essential features such as the ideas of
spatial waveguide modes, the number of modes possible in a given structure and
important propagation features like intramode and inter-mode dispersion can all
be approached in this manner.
The waveguide structure has other interesting implications. There is obvi-
ously a range of wavelengths for which the basic single-mode boundary
conditions can be fulfilled and indeed, in principle, this range extends well
beyond the optical region to the microwave and radio spectrum, though the
strength of the waveguided beam becomes vanishingly small at longer wave-
lengths (less and less of the energy is held in the core region and eventually, at
wavelengths significantly longer than the structural cross sectional dimensions,
most spills out into the surrounding air). At shorter wavelengths there comes a
point where at least two angular options can give rise to an appropriate
interference pattern and therefore satisfy the dielectric boundary conditions,
a point at which the waveguide becomes over-moded.
In the single-mode-transmission case there are dispersion mechanisms at
work similar to those on which we commented for multimode guides – namely
that the variation in reflecting angle for the imaginary interfering beams will
inherently give a variation in propagation time, with shorter wavelengths
experiencing longer delays due to geometric dispersion than the longer wave-
lengths – a similar trend to that experienced due to material dispersion in the
‘normal’ dispersion region, of a dielectric.
These dispersion effects are significant and have stimulated much ingenuity
in refining the basic concept in order to cram more and more terabytes along a
single fibre. There are other factors which limit fibre transmission distances, of
which the most important are the inherent losses in the silica which forms the
basis of the core of the fibre. These become very low (<1 dB/km) in the 1.5 μm
76 Light Interacting with Structures

Higher index:
solid material
Lowest index:
(or less ‘holey’)
gas
Lower index:
‘holey’ material
Diffracng layers:
Bragg guiding
‘holey’ material

Figure 4.16 Two examples of simple ‘photonic crystal’ structures, showing how
structures can provide innovative waveguide geometries. The structure dominates
the optical transmission properties of the material from which it is composed.

region (Figure 4.15). The material itself has another interesting property: its
inherent dispersion goes through zero in the 1.3 μm region – a case in which
nature conspires to be helpful to the technologist. However, the dispersion in
the silica at 1.5 microns, the lowest-loss region, remains significant, even at
~20 ps/km, for very-high-capacity communication systems.
Consequently, much has been achieved to minimise dispersion in this 1.5
micron wavelength region through the ingenious use of structural and material
properties, so that the transmission of hundreds of gigabits per second, and
more, over hundreds of kilometres through a single channel is now available.
The losses in silica exemplify different basic phenomena. The loss at long
wavelengths in the infrared is fundamental to the absorption bands within
silica itself and consequently there are many practical difficulties in extending
optical fibre transmission much beyond the 2 μm wavelength barrier. The
attenuation glitch at about 1.3 μm is due to water absorption but, thanks to a
great deal of processing effort, this can be virtually eliminated. At shorter
wavelengths the impact is seen of tiny inhomogeneities within the structure
itself and those that inevitably arise in the processing and preparation of the
fibre owing to the slightly random process of cooling down. These tiny
inhomogeneities cause Rayleigh scattering which technically applies to struc-
tures much less than the wavelength; we shall return to this in the next section.
The increasing precision of fabrication technologies has continually
expanded the level of control available for these photonic materials.
A conceptual example is shown in Figure 4.16. Here – on the left – a number
of voids (holes) have been introduced into the structure of an optical wave-
guide and these voids will in turn reduce the local effective refractive index, in
the planar case with different changes for vertically and horizontally polarised
inputs. The key point here is that the holes are a small fraction of a wavelength
in diameter, so that, as seen by the propagating wave they are effectively
 4.3 Dimensions Much Smaller than the Wavelength 77

averaged out. This gives a birefringent guide, from which, conceivably, a

structure could be found which will guide one polarisation and not the other.
Indeed, as we shall see such, structures based on metallic films can be designed
to do exactly that.
There are many variations on this basic theme of which perhaps the most
extreme is the waveguide with no core (or strictly an air core), indicated on the
right Figure 4.16. The waveguiding properties here are determined by diffraction
from the structure which surrounds the core and, for a given wavelength to be
transmitted both the Bragg diffraction conditions which we discussed earlier and
the waveguiding material boundary conditions need to be simultaneously met.
Consequently, the propagation characteristics of, for example, a circularly sym-
metric version of this structure – “hollow-core photonic crystal fibre” – exhibit
quite complex structure including some forbidden bands, for which the Bragg
condition cannot be met. Viewing this as normal index guiding is not an option!
The flexibility in design options for waveguides based on photonic crystals
is enormous, featuring variations in waveguide dispersion and propagation
spectra which are extremely diverse. The hollow-core fibres also have the
interesting property that gases and liquids can be introduced into the void,
thereby offering a new tool for spectroscopic investigation; this comes with the
proviso that the core is typically less than 10 μm in diameter, so that it takes
some time for fluid to enter the void and to leave thereafter. However, similar
functional flexibility is available for synthesising material properties on sur-
faces, particularly in the planar waveguide context, thereby enabling a surface-
based interaction with an optical transmission channel.
The interfaces between voids and solid materials in these photonic crystal
structures are never totally smooth, so that Rayleigh scattering at irregularities
(see below) in these interfaces becomes a persistent issue. Consequently, the
losses within photonic crystal fibres are inevitably much higher than in the in
all-solid-state forerunners, which are fabricated with very smooth glass-to-
glass interface structures.

4.3 Structures with Dimensions Much Smaller than

the Optical Wavelength
Subwavelength structures feature strongly in our daily lives. Thanks to the
scattering from such structures, sunlight diffuses throughout a room and the
sky in daylight hours is blue. Such phenomena have been reasonably well
understood since around 1870. Essentially, for scattering particles with dimen-
sions much less than the wavelength of the light, the strength of the scattered
78 Light Interacting with Structures

signal is inversely proportional to the fourth power of the wavelength. The

scattering sites can be tiny bubbles in a glass or even minute variations in local
refractive index, minute particles of ice in the atmosphere or tiny undulations in
the surface of a solid. The inverse fourth power relationship is at its most evident
in the blue sky. Clearly, blue light will scatter about 16 times as much as red
light on its way from the upper atmosphere to the earth’s surface and conse-
quently the sky appears blue. The red sky at sunrise and sunset sounds paradox-
ical but a brief investigation into its origins soon provides the explanation.
Scattering is also the cause of the rapid onset with shortening wavelength of
attenuation in optical fibres, mentioned briefly in the previous section.
In nature, colour originating in structural artefacts rather than in material
properties is extremely common, familiar examples being butterfly wings and
peacock feathers. The range of wavelength-filtering structures is immense,
embracing diffraction gratings, tiny coloured mirrors, photonic crystals and
nanoscale particles.
This understanding of nature’s very tiny structures, and the many optical
properties which they can have, has stimulated technologically based endeav-
ours to emulate and exploit these structures. The technology, of course, needs
to be compatible with structural fabrication to nanometre tolerances and in turn
has become known as nanophotonics. This builds upon the accumulated
experience of the past century or more in making controlled-response, signifi-
cantly subwavelength, electrically functional material structures, better known
as electrical circuits. As with any extension of the frequency range for elec-
trical circuits, two basic features now need to come together. The first is a
thorough understanding of the characteristics and impact of what, in electrical
circuit terms, are referred to as stray components (capacitance and inductance
in wires, for example); this understanding become increasingly complex as the
frequency goes up. The second is the necessary technology to realise circuits
with acceptable spatial and material composition tolerances. Indeed, nanopho-
tonics has been described as simply realising electrical circuits at extremely
frequencies.
Perhaps the most intriguing aspect of these synthetic ‘metamaterials’ is their
theoretical (but to some extent demonstrated) optical properties, of which can
be totally controlled by the structure itself and the materials therein. Possibly
the invisibility cloak gains the most headlines! This is suggested as a material
coating that redirects input light so that the light emerges apparently
undeflected: the object in question is no longer visible. The detailed design
of such cloaks is complex but the essential concept lies in the realisation of
negative-index metamaterials (or, to be more precise, material structures), for
which the refracted ray lies on the same side of the normal as the incident ray,
 4.3 Dimensions Much Smaller than the Wavelength 79

Metamaterial,
index = -1

Perfect
Object image

Figure 4.17 A conceptual diagram of the operation of a perfect superlens with

n = 1.

giving rise to a negative sign in the refractive index. The behaviour of such a
metamaterial structure is indicated in Figure 4.17. In principle such a structure
can produce a totally undistorted image of an object. It can beat the diffraction
limit, the ‘blur’ from the edges of the lens or mirror, which characterises
traditional imaging.
The detailed principles are quite complex but some insight into the basic
ideas can be derived from the observation that negative-index metamaterials
can be synthesised using periodically spaced arrays of resonant circuits.
A feature of such a circuit is that at low frequencies, well below resonance,
the current (i.e. the magnetic field) and voltage work in phase whereas at
higher frequencies, well above resonance, they are in antiphase. Within an
electromagnetic wave this will attempt to pull the electric and magnetic fields
out of phase with each other, a feature which can be compensated by the
wave’s moving to the opposite side of the normal. Metamaterials, in particular
those with a negative index, have found application in acoustics (in inverting
the phase of the pressure and velocity components of the acoustic wave) and in
electromagnetics. Early demonstrations were at relatively low (microwave)
frequencies but there have been ventures into the optical, constrained to extent
by the properties of conductors at optical frequencies.
One of the principal prospects for nanostructured materials lies in the use of
metallic conductors at frequencies well below the plasma resonance and at
thicknesses sufficiently low for optical signals to cope with the inherent losses.
One manifestation of such configurations is that the presence of a metal on an
interface between two dielectric materials can significantly enhance the mag-
nitude of the electric field on the face of the dielectric interface opposite to the
direction of incidence (Figure 4.18). There is another important aspect to this.
The losses are low if the electric field is perpendicular to the plane of the
80 Light Interacting with Structures

(a) (b)
Launch a TM (P) wave
along here. Thin metal film

Dielectric guide for phase matching Analyte

noverlay

nsubstrate Substrate

Surface plasmon metal layer – dashed to

indicate a possible grating coupling system

Figure 4.18 Surface plasmon waves giving high field-penetration into the overlay
but requiring careful phase matching whether in multiple layers (a) or waveguide
format (b). The latter arrangement finds application in, for example, liquid
characterisation, hence the reference to ‘analyte’ (a substance under chemical
analysis).

interface. However, if the electric field is in-plane, then it sees a long length of
metallic material – the in-plane polarisation is very lossy. So here we have a
polariser, much used over the past few decades! This basic phenomenon,
which is an essential aspect of plasmonics, raises prospects for a range of
photonic devices emulating their electromagnetic forerunners, from wire-based
antenna arrays capable of refocussing a diverging beam to optically excited
electro-chemical sensing.

Quantum Dots
Nanostructures other than butterfly wings have also been with us since ancient
times. Colloidal gold nanoparticles have given permanent colour to glassware
and pottery for centuries. These tiny gold spheres are perceived from red to
blue over the entire visible spectrum – completely at odds with the visual
appearance of bulk gold. These particles, which are typically a few tens of
nanometres in diameter, have now extended their presence beyond ancient
pottery into medicinal potions and (bio)chemical assay. How does this struc-
ture so efficiently modify the bulk reflected colour? There is a clue in that the
larger particles appear more to be blue than the smaller particles, which veer
towards the red. It would seem, then, that the large particles absorb the red end
of the spectrum and vice versa for the smaller particles. Clearly, then the key
lies in the dimension of the spheres but also in the observations on surface
currents discussed above. A large sphere, which is typically about 100 nm in
diameter, resonates in the red and therefore absorbs the red whereas a small
sphere, around 30 nm in diameter, resonates in the blue.
 4.4 Summary 81

Venturing towards even smaller dimensions – controlling them to nm or

greater precision – can tailor the apparent colour of most solid materials.
A large ‘quantum dot’ has dimensions of around 5 nm. At this ‘large’ size
the dots tend to absorb, and equally can fluoresce, in the red. This potential for
fluorescence distinguishes the behaviour of quantum dots from that of gold
nanoparticles and their surface plasmon resonances. The smaller quantum dots,
2 nm or thereabouts, by the same token tend to absorb and fluoresce in the
blue. In this case it is not surface plasmon waves and the electromagnetic
absorption therein which is the source of the colour change for these tiniest of
particles: the physical size of the structure is now so small that the quantum
mechanical energy levels within the structure are involved. Consequently, the
absorption band composition and hence the colour is affected. Quantum dots
typically find application in the context of lasers, solar cells and light emitting
diodes but also in some forms of medical imaging and in quantum computing.
There are many possible applications, but most will take some time into the
future to mature.

4.4 Summary
Perhaps the principal message from this chapter is that structures are an
extremely flexible, even fundamental, tool for the manipulation of light,
whether regarded as rays, waves or photons. The operational functionality of
this tool changes in a very versatile manner with the dimensions of the
structure in comparison with the wavelength of light.
For a very large structure – essentially a fraction of a millimetre dimension
and above – we have seen how the shape of interfaces and of optical compon-
ents, typically glasses, modifies optical functions, prisms and lenses being the
dominant systems. We have also seen that using a mixture of glasses and shape
functions can enhance functionality; this is most apparent in the design of
compound lens systems. The basic phenomenon of refraction, together with
the associated polarisation-dependent reflection effects at interfaces, is at the
heart of the familiar optical components within this dimensional range.
Even such large structures are, however, affected by the impact of phenom-
ena most conveniently thought of as relevant to the medium-dimensional
range; that is, around the optical wavelength. The resolving power of lenses
is the most obvious feature, as it can be explained in terms of the essential
medium-scale concepts of diffraction and interference. These medium-scale
features underlie holograms, fibre optics, the appearance of oil on water and
many other familiar optical phenomena.
82 Light Interacting with Structures

Advancing to the minute scale, we have structures with features small

compared with an optical wavelength. Here we enter the domain of nanopho-
tonics, where carefully considered extrapolations from the concepts of elec-
trical circuits come into play. The ideas underlying plasmonics are possibly the
closest conceptually to circuits. Metamaterials, another essential concept from
the tiny minute scale, involve the fabrication of carefully organised material
mixtures with features well below the optical wavelength, again conceptually
familiar from, for example, a TV antenna.
Going to even more minute structures, of the order now of 1% of the optical
wavelength on a few nm, we reach the quantum dot regime where the
dimensions of the structure begin to impact upon the quantum mechanical
properties of the material. Very tiny structures will, for example, shift the
energy level distribution. This entirely modifies the actual optical properties of
the original material. This is a relatively new domain, but it has already
demonstrated for example, the potential for very flexible laser design and
innovative features in solar cells.
Optical materials and material structures combine to offer great flexibility in
the functionality of optical artefacts. In fact, almost every application of optical
systems relies on carefully and critically combining the optical material prop-
erties as briefly considered in Chapter 3 with versatile application of structural
features outlined in the present chapter.

4.5 Problems

1. (a) Think about the basic spherical biconvex lens of the type indicated in
Figure 4.4. Now use your knowledge of the slopes of the spherical
surfaces (just consider two dimensions) to gain some insight into how
the focal point might vary as a function of lens aperture (compared with
the focal length) at a single wavelength. At what aperture would this
variation in the position of the focal point begin to compromise the
resolution of the lens for illumination at 500 nm wavelength (think of
the implications of Figure 4.6)?
(b) It is possible to achieve a perfect (monochromatic) focus with an
aspherical surface. How would you approach determining the shape
of this surface?
2. This problem concerns retro-reflectors.
(a) Consider a prism with a right-angled isosceles-riangle cross section. It is
simple to demonstrate that a beam normally incident on the hypotenuse
 4.5 Problems 83

will be retro-reflected. Over what angular range (for angles lying within
the plane along the hypotenuse and perpendicular to the input beam)
will a parallel incident beam be directly reflected (take the refractive
index as 1.5)?
(b) How will the reflected beam angle change with incident angle? You
will probably have instinctively approached the problem in two dimen-
sions – varying the angle about an axis which is perpendicular to the
hypotenuse and in the plane of the paper on which you’ve sketched the
system. How might this change if the beam angle were varied around an
axis a plane perpendicular to the hypotenuse?
3. Red sky at sunrise and sunset, blue skies in day time. Why are clouds are
essential for an impressive sunset? On the same theme – why are clouds
white (except at sunrise and sunset) when relatively thin, but go through
shades of grey into a threatening black?
4. (a) Convince yourselves that a parabolic mirror gives a perfect focus at all
light input wavelengths (Figure 4.5).
(b) Assume that this applies to a 5 m diameter reflecting telescope. Give a
reasonable estimate for the angular resolution and practical angular
coverage of such an instrument. Atmospheric turbulence can distort
images for even a perfect reflector. How would you approach setting
up a system to overcome these disturbances?
5. (a) How would you measure the number of threads per centimetre in a piece
of cloth with just a laser pointer and a piece of white paper? Do the
experiment, comment on how the answers change when the cloth is
stretched in one direction and try shining the pointer on a piece of
scratched or smeared glass and looking at the spots that are transmitted.
(b) What might be some practical implication of these observations in
optical-system design and in the processing of digital images?
6. (a) Create an expression to describe a simple phase object with a one-
dimensional sinusoidal distribution. From this derive the features of the
diffraction pattern from such an an object, and demonstrate how the
diffraction may be modified (i.e. spatially filtered) to create an intensity
object that is related to the original phase object, and sketch a suitable
system.
(b) There are at least two options on the approach to the above procedure –
both mentioned in the text. Compare and contrast the features of these
two approaches and also find examples of phase objects, some of which
can be perceived with the naked eye. How is that possible?
7. (a) Convince yourself that the Bragg conditions (angle of incidence equals
angle of reflection, and path difference between adjacent reflected rays
84 Light Interacting with Structures

equals one optical wavelength; see Figure 4.12) can give 100% con-
structive interference for an input parallel beam meeting a suitable phase
object. What are the conditions for the object to be ‘suitable’?
(b) A Bragg diffraction process uses ultrasonic waves in transparent media.
The reflections involved are now from a moving object, so that the
output beam is shifted in frequency according to the Doppler effect.
Explain how this frequency shift is related to the ultrasonic frequency.
8. (a) Figure 4.14 shows the cross section of a dielectric waveguide compris-
ing two low-refractive-index regions surrounding a slightly higher-
index core region. Typical refractive indexes are in the region of 1.5
and typical index differences are around 1% of this. By considering the
interference of intersecting beams at symmetrical angles with respect to
the core axis in the waveguide, estimate the cross section of the guiding
region at which the guide will become overmoded for a wavelength of
1 micron.
(b) The intersecting-beams approach works well for metallic interfaces in
waveguides at much longer (for example, microwave) wavelengths but
for dielectric guides, it can give only a good insight especially at these
much higher frequencies. What are the differences between dielectrics
and metals and how might these differences modify the original esti-
mate for the depth of the waveguide?
 5
Photonic Tools

In this chapter we look briefly at some of the many applications which are
enabled through photonics, ranging from photo-therapies to intercontinental
communications. The majority of these applications rely upon extracting an
optical signal from various forms of background noise, which is where we start.

5.1 Optical Systems

When discussing an optical system the initial question is ‘can we identify the
parameter – the signal – in which we’re interested’? Assuming it can be found,
the next question is, ‘how well can it be identified’? The answer to the first
question is encapsulated in another question, ‘does this parameter’s presence
exceed that of all the other phenomena which are happening at the same time’?
Does the signal exceed the noise? The answer to the second question can be
found in ‘by how much does this signal exceed the noise’?
Noise in optical systems arises through a variety of mechanisms, depending
upon the particular context. In an optical communication system, for example,
an electronic receiver sees nothing but the modulated signal of interest. In
contrast, imaging systems see the background illumination and spurious reflec-
tions which often compromise image quality. There is also the possibility that
random noise originating within the light source will affect the signal of
interest, and finally there are noise sources within the detection mechanisms
themselves.
The simplest situation occurs when the light source of interest is incident
upon a detector – typically a photodiode – in an otherwise darkened environ-
ment (Figure 5.1). Here there are inevitable sources of noise. First is the
thermal noise within the load resistance on the photodetector. This rms noise
power, Pth, is related to the absolute temperature, T, and bandwidth, B, through

85
86 Photonic Tools

(a)
RL
Input light

1mW
at 1mm

Shot and thermal

noise Load resistance RL Ohms
contribuons 1E-17
1 10 100 1000 10000 100000 1000000 10000000
1E-18
Noise Power was (Hz) -1

1E-19

1E-20 Thermal Noise in RL

1E-21
at 300K

1E-22

1E-23

1E-24

1E-25

Phase noise
(b)
Intensity
Carrier vector noise

Figure 5.1 (a) A basic photodetector circuit demonstrating the trade-off between
load resistor thermal noise and shot noise on the incoming optical signal. The load
resistance and detector capacitance define the detection bandwidth. (b) A vector
representation of noise components with equal distribution in phase and amplitude
components.

the Boltzmann relationship Pth = 4kTB (k is Boltzmann’s constant). It gives a

noise current dependent on the value of the load resistor.
The second noise source reflects the fact that the detector recognises the
presence of the light as photons. These photons do not arrive at exactly regular
intervals; they follow Poisson statistics, so that an average rate of arrival of Np
photons in a particular time interval will statistically fluctuate by Np1/2 when
observed over successive time intervals. This is known as shot noise. Add-
itionally, only a fraction of the arriving photons actually create electrons in the
detector – a fraction quantified through a parameter known as the quantum
efficiency. There will also be additional noise sources within the detection
amplifier circuitry, not to mention inherent noise sources in the source of light
itself, due to, for example, minute variations in the current driving the light
 5.1 Optical Systems 87

source. However, if we assume good design these effects can, in many

practical cases, be considered negligible. Under such optimal circumstances
and assuming 100% quantum efficiency, the ratio of the total current to the
noise current within the photodetector is Np1/2.
In the ideal case – the shot noise limit – the noise due to the fluctuating
photon arrival rate will dominate, so that, fundamentally, the noise perform-
ance of any optical system is determined by the average photon arrival rate
over a time interval given through the reciprocal of the system’s operational
bandwidth.
This gives a straightforward and reliable indication of achievable signal to
noise ratios. As an example, Figure 5.1 also includes a plot indicating the
operational regime in which the shot noise on a 1 μW optical source at 1 μm
wavelength will dominate system performance at the shot noise limit: this
occurs when the induced shot noise current exceeds the thermal noise current
in the specified load resistor. In the example, this requires a load resistor in
excess of ~20,000 ohms.
This process can be applied to any photonic system which relies on elec-
tronic detection, but even then the impact of stray light can be a source of
interference. Unscrambling the desired wanted signal from the unwanted
signal in this more general case can be significantly more involved. There
are also many occasions where the characteristics of the human visual system
come into play, particularly in the context of displays and illumination. An
emerging application sector in photonics showing immense promise lies in
biomedicine.
Even in the shot-noise-limited case some subtleties need to be considered.
The variability in photon arrival rate previously alluded to presents only a
simplified perspective. Here, a complementary wave motion perspective,
describing the same total noise power as a variation in the phase or amplitude
of the mean wave vector (Figure 5.1(c)) is helpful. Thus the shot-noise
behaviour can be viewed in the wave vector domain as an equally probable
perturbation in the phase or the amplitude of the received optical signal, with
equal power distributions in the phase and amplitude noise. Many optical
systems rely on detecting changes in either the phase or the amplitude of the
received optical vector; it can be helpful to retain the overall power signal to
noise ratio dictated by the shot-noise limit but move it more towards the
intensity sector for systems utilising phase modulation, or vice versa. An
example follows.
The ‘squeezing’ of light uses elegant optics to ‘push’ the phase-noise
component in Figure 5.1(c) into amplitude noise and hence to improve phase
detection, or vice versa; this has been demonstrated using numerous
88 Photonic Tools

approaches. This squeezing can in principle squeeze the phase, or the ampli-
tude, noise away completely to zero and leave only the complementary
domain. In practice, this has yet to happen completely. However, at present
there is one outstanding example; it is being used in earnest on just one
instrument, the gravitational-wave telescope, LIGO. Here, without optical
squeezing giving a 10 dB or more additional signal to noise ratio, the detection
of gravitational waves would have remained elusive. This gain is achieved
through an impressively complex system, but to date there are no simpler
approaches. Squeezed light exemplifies the many recent conceptual develop-
ments in photonics that combine the wave and photon approaches.

5.2 Spectrometers
Spectroscopy is a crucially important tool in photonics for characterising both
materials and material structures. In practice, most spectrometers follow the
format shown in Figure 5.2, where parallel white light is passed through the
sample of interest to interact with the dispersive element (here a prism) after
which the emerging light is focussed onto a detector array. The basic procedure
is that the white light source is calibrated in the absence of the sample. Hence,
any differences between the calibration readings on the array and those after
the source interacts with the sample relate uniquely to the spectral properties of
the sample. The dispersive element can in fact be either a prism or a diffraction
grating; the former deflects the blue more than the red whilst the grating does

White
Collection
light
lens
input Red
deflection

Blue
Sample for analysis
deflection

Figure 5.2 The basic dispersion diagram for a prism spectrometer. Grating spec-
trometers operate similarly but give the greater deflection to the red rather than
the blue.
 5.2 Spectrometers 89

the reverse. A grating also offers the benefit that the angular dispersion is
directly related to the grating period, whilst for a glass-based prism the
relationships are more involved.
The intrinsic wavelength resolution of these instruments is dictated by
several factors. The number of photodiodes in the detector array obviously
limits the ultimate achievable resolution, (though with too many detectors, the
achievable noise limit on each detector, which is shot noise limited, also
contributes) Additionally, the angle of separation between one colour com-
ponent and another comes into play. In the case of a grating this is determined
by the grating periodicity whilst for a prism the dispersive power of the glass
of the prism is the determining factor. Finally, the beam width of the interro-
gating optics (comprising a combination of the beam width, the depth of the
sample container, and the effective incident width) provides an ultimate limit,
dictated by diffraction from the edges of its optical aperture. In an optimal
design all three of these limitations would come into play together.
There is also an important variation on the basic theme of the photodetector
array. The same functionality can be achieved if a single photodiode, with a slit
in front of it, is scanned across the resulting spectrum either by moving the
photodiode or by rotating the spectrum. The overall result is that, for a typical
prism or diffraction grating spectrometer, a spectral resolution of the order of
0.1 nm over a dynamic range of 1 μm is achievable.
There are occasions when better resolution is desirable and also occasions
where the photodetection processes become more involved; typically, both
factors apply in the near and mid infrared, where an alternative approach based
on Fourier transform spectroscopy (Figure 5.3) is frequently used. In this
system a single detector records the total output as the path difference between
the two arms of the interferometer formed by the two mirrors is scanned from
around zero to a maximum value. At each setting the interferometer acts as a
filter with co-sinusoidal response, as indicated in the figure. Consequently, in
effect each optical frequency in the input spectrum is multiplied by this co-
sinusoidal function in the frequency domain for each setting of the interfer-
ometer path difference – in other words, the system is performing a Fourier
transform at each setting with spectral components determined by the path
difference. Assuming that the path difference increments by equal amounts
between a minimum (which sets the range of frequencies that can be resolved
by the interferometer) to a maximum (which determines the resolution of the
resulting spectrum), then a direct Fourier transform of the output gives the
spectral response of the sample. The detector needs to be calibrated appropri-
ately in the range over which it is to used, and some precautions are needed to
ensure that the Fourier transforms are appropriately performed, taking into
90 Photonic Tools

Fixed mirror
Path
difference:

Input Zero
light Moving mirror
0

Detected output
Beam splier
Sample L

2L
Calibrated
photodetector
0
Input optical
frequency

Figure 5.3 Fourier transform spectroscopy, showing how varying the position of
the moving mirror by L multiplies the input spectrum by a defined co-sinusoidal
function. The beam splitter divides the input light equally between the two paths
through the interferometer. The top line on the graph indicates that at zero path
difference the output power is fixed for all wavelengths.

account any spectral variations i.e., variations in the light output from the
interrogating optics as a function of wavelength. Overall then, the system is
functionally identical to the dispersive element in a prism or grating
spectrometer.

5.3 Display Technologies

Display technologies manifest themselves in two basic formats. The first is the
familiar luminescent screen, to be seen on TVs, phones, tablet, computers and
the like. The second uses an external light source to illuminate a variable,
usually transmitting, object; such spatial light modulators are most familiar in
the operation of digital projectors. The details of these technologies are many
and varied, though essentially luminescent displays rely upon light emitting
diodes, whilst spatial light modulators rely on liquid crystals and usually
incorporate appropriate colour filters.
Naturally, display technology needs to be designed with the human visual
system in mind, usually targeting the daylight-adapted eye, which resolves
colour through cone receptors. These receptors are sensitive over a wide range
but individually peak in the red, the green and the blue, giving rise to the
concept of primary colours from which all other perceived colour combin-
ations can be synthesised. As a group, the receptors have the overall spectral
 5.5 Three-Dimensional Imaging 91

response typified in Figure 3.19. This synthesis process from the three primary
colours was quantified nearly a century ago in the CIE (International Commis-
sion on Illumination) chromaticity chart, still a very useful tool in the detailed
design of visual systems.
These factors are mentioned here as among the most important consider-
ations in the design and implementation of both display technologies and also
electronic image acquisition technologies, such as the image capture array in a
digital camera. Hopefully this will give the reader a little insight into the
considerable efforts required in balancing colour-filtering systems, relative
brightness, spatial resolution and a host of other factors in the design of these
very commonplace but extremely subtle devices.

5.4 Images and Imaging

All image formation, whether by a telescope, a microscope or even a magnifying
glass, is based upon the same principle. A lens collects light from the object and
thereafter projects an image which is subsequently processed by the viewer and
received via the retina before being processed by the brain. There are often many
steps in the process – camera to data file, to image on a screen, to the viewer, for
example. Images, as we have already discussed qualitatively, can be decom-
posed into Fourier components, a representation of which is found in the back
focal plane of the collection lens. This Fourier transform plane is widely
exploited as a means for image processing using both direct-optical and digital
techniques. As examples, we have already mentioned the concepts of contrast
enhancement and phase-object-to-intensity-image manipulation.

5.5 Three-Dimensional Imaging

The basic principles of holography were outlined in Chapter 4. It emerged in
the mid twentieth century as a concept which was facilitated through the
evolution of the laser. The basic observation which underlines holography is
that if we can somehow record both the phase and the amplitude of a
propagating optical wave, we can determine whence it came. Photography,
whether digital or on film, like our eyesight perceives only the intensity of the
arriving wave as a function of position, producing the familiar two-
dimensional images. The key point, then, is to change phase information into
some form of intensity information which can be recorded, and thereafter to
find a playback mechanism.
92 Photonic Tools

Holographic systems have evolved from their original novelty as a means of

producing strange and intriguing three-dimensional images which just hang in
space (numerous science museums continue to excite young minds through
this very artefact) into useful tools, derived from variations on the basic
principle, for surface inspection, vibration detection and similar uses. The
principle of this use of holography is as follows: record a hologram of an
object which is to form the reference and place the sample object exactly in the
position where the reference object is reconstructed. Any differences between
the reference object and the sample object will then appear in the form of
interference fringes, the detectable differences being of the order of a fraction
of a wavelength.
Holograms make their daily appearance as security checks on credit cards.
They are being evaluated for extremely high-density data storage – especially
in the more versatile volume-hologram variant, where the propagating wave
front phase distribution is recorded not in a two-dimensional plane but through
the thickness of an emulsion extending over many wavelengths of light.
In common with many photonic concepts, holography has evolved from an
interesting scientific principle into a widely applied tool, ranging from entertain-
ment, to high-precision surface inspection, to security and identification checks.
Optical-coherence tomography (OCT) presents an entirely different
approach to deriving a three-dimensional image. The principle is shown in
Figure 5.4. The object is illuminated using a very low coherence length
source – typically a super-luminescent diode. A reference mirror can be moved
along the optical axis as indicated and the consequent interference pattern seen
by the receiver will produce peaks and troughs only when the optical path
difference between the imaging arm and the reference arm is around zero,
thanks to the very short coherence length of the super-luminescent diode.
Consequently, recording the image as a function of this path difference gives
the reflectivity of two-dimensional slices of the object as a function of depth.
Stacking together these slices gives a three-dimensional image of the object.
The depth for which this system can operate is effectively dictated by the range
of path differences which the moving mirror can access and is limited by
attenuation and scattering within the object under investigation.
The system of optical-coherence tomography has found extensive applica-
tion in producing diagnostic images of the eye, and the apparatus is found in
many high street opticians. It has also been used intravenously for monitoring
cardiovascular degeneration and, in an entirely different domain, as a tool for
imaging and assessing silicon and silica device-fabrication processes. Thus
OCT has grown to be an important and versatile three-dimensional imaging
tool since the original concepts were identified in the early 1980s.
 5.6 Precision Measurement 93

Low-temporal-coherence source
Single-mode
e.g. super-radiant LED
optical fibre – for
spatial coherence CCD camera
Collimating lens objective

CCD camera
detection
array

Moving
reference To computer storage
Multi-layered and image presentation
mirror
sample software.

Figure 5.4 The basic principles of optical-coherence tomography. The moving

reference mirror selects a plane within the object which will produce constructive
interference and give an image on the camera array corresponding to the structure
within that plane.

Predating these techniques were the – still used – three-dimensional imaging

systems exploiting the fact that we can see in three dimensions owing to subtle
processing of the differences between the images seen by each of our two eyes.
If we take two simultaneous photographs using cameras spaced by the same
amount as a person’s two eyes then, using a projection system which sends
one image to each eye, the two images will be perceived as a three-
dimensional image. The early three-dimensional cinema operator projected
orthogonally polarised images for each eye and equipped the audience with
left and right eye selective glasses tuned to the two polarisation states. Three-
dimensional virtual reality systems are based on the same dual-image photog-
raphy concept.

5.6 Precision Measurement

Optical interferometry, through which movements of the order of a small
fraction of an optical wavelength can be reliably detected, is a natural tool
for precision measurement. The application sectors have gradually been
expanding for over half a century.
As we have seen, interferometers measure changes in the optical path
difference between two points. In the case of a two-beam structure, this path
difference is between a fixed reference object and some other moving reflective
structure. For the simplest two-beam interferometer, in which the coherence
94 Photonic Tools

Reference
mirror with
Beam splitter
quadrature
offset

Reflective target
for measurement
Detector 1

Detector 2

Fringe pattern
Detector 1 Fringe pattern
Detector 2

Fringe
Intensity

n-1 n n+1 n+2

Change in interferometer path difference in optical wavelengths

Figure 5.5 Using a quadrature offset system (i.e. the beams are 90 out of phase)
in a Michelson interferometer to extract directional information and therefore
measure absolute changes in position.

length of the source greatly exceeds any possible value of the path difference
between the two arms, one cannot distinguish whether the path difference is
increasing or decreasing as it monotonically changes. However, a change in
the direction of movement can be detected from a single out put provided that
this change does not occur at a peak or trough of the interference pattern (see
Figure 5.5). Consequently, interferometric precision-measurement systems
need to incorporate some method of determining the direction of changes in
path difference. One approach, shown in the figure, is to utilise a quadrature
offset between, in effect, two reference beams. This arrangement enables a
combination of fringe counting and directional assessment to produce highly
accurate measurements. This approach, based upon quadrature detection, can
routinely yield measurement precision better than 1% of an optical wave-
length; measurement accuracies can even be of the order of several parts per
million, depending upon the stability of the source illumination and the
variations in local environmental conditions.
Interferometry has found enormous application in the precision measurement
of mechanical objects, especially in systems based upon coordinate-measuring
machines. In these machines, a very precise x–y–z translation system is coupled
to interferometric displacement-measurement systems. The translation system
 5.6 Precision Measurement 95

Object under test

Drive system for
interferometer

Optical flat
Quadrature interferometer
Ultrasensitive
tactile probe
controls drive

Object to be measured
Interference between flat and
test object

Figure 5.6 (left) Measuring a fixed external object using a sensitive probe and
quadrature interferometer. (right) Using interferometry to measure objects with
respect to an optically flat surface (here interference produces Newton’s rings,
which are observed from a spherical partially reflecting surface as indicated). The
same basic approach is used in holographic inspection systems.

carries a very sensitive tactile probe which responds precisely as it touches the
object under study. There are several such probes, at least three in a particular
instrument, to facilitate three-dimensional measurements. The displacement of
these tactile probes can be interferometrically monitored and from this the
surface profile of the object can be precisely determined. Submicrometre reso-
lution can be routinely obtained.
Figure 5.6 illustrates this basic tactile system and also shows how a preci-
sion optically flat glass can be used to characterise the shape of a lens. Now
referred to as Newton’s rings, this patteon was first observed over 300 years
ago. When viewed with white light the rings are coloured, but when viewed
with monochromatic light, for example from a gas discharge lamp or even
through the rudimentary colour filtering of sunlight, the characteristic interfer-
ence pattern shown in the figure provides an accurate measure of the lens
curvature, and variations thereof, with respect to the reference optical flat. The
spacing of the fringes depends upon the local tangential angle between the lens
surface and the optical flat.
Yet another variation on the interferometric measurement theme is shown in
Figure 5.7. The Sagnac interferometer sends light in counterpropagating direc-
tions around a single-spatial-mode loop, which can be a fibre or the active part
of a laser. As the loop rotates, the light travelling in the same direction as the
rotation needs to travel a small distance further than the light travelling against
the rotation in order to return to the entry point. Consequently, if this system
96 Photonic Tools

modulator
Polariser

DC1 DC2
rotation

Fibre loop

Architecture of the optical fibre gyroscope:

DC1 and DC2 are directional couplers

Figure 5.7 The basic principles of the Sagnac interferometer, as used as an optical
fibre gyroscope, and an example of its precision realisation for space navigation.

forms the cavity of a laser, there is a small difference in oscillation frequency

for each direction of travel. In the fibre-optic case shown, there is a corres-
ponding small phase difference. With subtle and elegant signal processing, this
phase difference can be measured with a precision of a few nanoradians, which
translates to a rotation rate sensitivity of less than 10–4 degrees per hour. This is
sufficient to guide space exploration and two orders of magnitude better than
that required to navigate an aircraft across the Pacific Ocean. In the laser case,
there are issues in separating the clockwise and counterclockwise lasing
systems when the resonant frequency differences become very small (the
two separate oscillators become coupled to each other and lock onto a single
frequency), so that this variant on the Sagnac gyroscope tends to have a limit in
the region of 10–2 degrees per hour.
These brief examples merely scrape the surface of the uses of interferometry
in measurement but are intended to give some indication of the precision with
which interferometric measurements can be made.

5.7 Optical Fibres and Communication Systems

The basic concepts of optical fibres were first mooted more than half a century
ago. The initial attraction was that losses below 20 dB/km appeared feasible
provided that the glass, strictly silica, was extremely pure and the interface
between the core and the cladding was sufficiently smooth to prevent scatter.1
Copper cables for electronic systems in communications were then reaching

1
The term ‘scatter’, abbreviating ‘the scattering of electromagnetic radiation’ is often used when
we are considering a phenomenon rather a designed process.
 5.7 Optical Fibres and Communication Systems 97

their limit at around this attenuation level for carrier rates in the 250 Mbits/s
region. However, the skin effect losses (see Figure 3.3) in copper increase with
frequency, yet even higher frequencies were needed to carry more and more
data. As we saw in Chapter 4, numerous paths in the simplest fibre will
continue to guide until cut off at the critical angle between the core and the
cladding. This range of paths implies a travel time difference which is typically
of the order of a few tens of nanoseconds per kilometre. This dispersion in turn
implies the need for maximum modulation frequencies into the fibre in the
region of many tens of megahertz on a one-kilometre length. Thus, initially
multimode fibres offered little obvious benefit for long-distance communica-
tion when compared with copper.
However, there is a modification which changes the picture somewhat.
Suppose the refractive index profile were to change more gradually than the
abrupt high-in-the-core and low-in-the-cladding version profile found into a
step-index fiber. If the profile were instead parabolic (along the same principle
as the parabolic mirror in an astronomical telescope) then, at least for one
specific wavelength, the dispersion effect could be removed. There are prac-
tical issues regarding how closely can we make the refractive index profile into
a parabola, and there is also material dispersion in the core itself, but the
graded-index multimode fibre remains a useful tool. You may wish to explore
further how this might work and why it could be useful.
Single-mode fibres offer improvements in dispersion over even a graded-
index fibre. However, even here the minimum-attenuation wavelength (1.5
microns) and the zero-material-dispersion point at 1.3 microns do not coincide.
Making the zero-dispersion point coincide with the very low attenuation region
then promises enormous benefits!
The 1.5-micron lowest-attenuation region has emerged as the optimum
wavelength for optical communications, and so minimising dispersion is crit-
ical. Considerable ingenuity has been directed at this problem, based on the
observation that the evanescent tail (see Section 4.2) of the optical field distri-
bution in the fibre cladding extends further into the lower-index cladding as the
wavelength increases. Consequently, that mode will have a faster velocity as it
penetrates further into the lower-index region. A carefully designed ‘W’ refract-
ive index profile (Figure 5.8) controls this wavelength-dependent evanescent
field by means of a high-index core followed by a lower-index cladding
surrounded by a slightly higher-index cladding. This design can in fact compen-
sate for the material dispersion; it controls the relative amounts of optical power
in the core and depressed cladding regions and the process is optimised to realise
this objective. The reader can consult Figure 4.16 to verify the logic behind this
approach. The result is extremely low-loss transmission (a small fraction of a
98 Photonic Tools

Guided
optical
field
Depressed cladding: lowest index

Core: highest index

Fibre cross section Refractive index plot

Figure 5.8 The basic concept of dispersion compensation, combining the prospect
of low attenuation at 1.5 μm whilst correcting for finite material dispersion (see
also Figures 4.14 and 4.15). In the central part of the figure are shown a shorter-
wavelength pulse and a longer wavelength pulse, which penetrates further into the
depressed cladding region.

decibel per kilometre) and the residual dispersion is measured in picoseconds

per kilometre. This performance is critical in enabling the unprecedented cap-
abilities offered through the current communication systems and the enormous
internet bandwidth capability now taken for granted throughout the developed
(and increasingly in the developing) world.

5.8 Lasers in Machining and Surgery

The inherent spatial coherence of a laser beam has many benefits, of which
possibly the most evident, and definitely among the most remarkable, is that
straightforward focussing optics can produce spot sizes with dimensions in the
microns. Huge power densities are then easily attained; a one watt laser beam
focused to a 10 µm diameter produces in excess of 1 MW/cm2 power density.
Lasers with power in the hundreds or even thousands of watts are now
available, upping the available focussed power densities to GW/cm2.
In almost any material this will produce huge, highly localised temperature
rises, of the order of thousands of kelvins, with melting or even evaporation in
the very small focussed-beam area. Even at much lower power densities,
significant changes in highly localised structural (bio)chemistry can take place.
The applications of these small focused spots are diverse and Figure 5.9
indicates some of the prospects. The most obvious is laser machining, both
cutting and welding. Such laser cutting enables consistent precision; there is no
 5.9 Environmental Analysis 99

Focussing lens • Metals – precision machining by

Target high-temperature cutting,
object excellent surface finish
Laser beam
input
• Plastics – laser ablation of
polymers

• Surgery, tissue cutting – cataract

removal, precision internal access,
oral surgery etc.

• Surgery often aided by fibre delivery

Figure 5.9 The essentials of laser machining and laser surgery, utilising a tightly
focussed spot from a high-peak-power pulsed laser, very similar to laser-induced-
breakdown spectroscopy (LIBS) but with a different aim.

cutter wear and minimal machine vibration. The process also offers smooth
edges and the ability to cut in otherwise inaccessible places, particularly with
the flexibility of fibre-coupled cutting systems.
Laser surgery has had a particular impact in the treatment of ophthalmic
problems such as macular degeneration and cataracts where highly localised
cutting and preparing without the use of the surgeon’s knife is particularly
beneficial. The laser scalpel has also found its place, especially in soft-tissue
surgery where its precision and the minimal consequential bleeding that it
produces, owing to self-cauterisation, offer great benefits. Also, a laser beam
clearly does not require careful sterilisation between procedures, though there is
the need to pay due attention to the delivery system, which could even be a
disposable fibre. Laser surgery has been incorporated into endoscopically
observed surgery and a host of other medical procedures.
Lasers have also found their way into dentistry. There is a clear optical
distinction between decay in a tooth and the intact regions around it, with the
former much more optically absorptive. There is then the potential for thermal
removal rather removal by drilling. Lasers are also useful in oral surgery, again
especially when fibre-coupled to facilitate the precision location of a cutting
tool, and this, together with the above-mentioned automatic self-cauterisation,
are attractive features. Indeed, possibly the only disadvantage of lasers is that
the capital outlay compared with a scalpel or a mechanical drill remains high.

5.9 Environmental Analysis

This embraces one of the most important application sectors for spectroscopy
in its many forms. Measurements using absorption spectroscopy, Raman or
100 Photonic Tools

D2
Scaer

White light input

Scaered light Angle q Switchable white

Wh
light inputs:

ite
D1
spectral nephelometry

li
gh
Input at l Wh

t in
Sample ite
ligh

pu
Transmied t in

t
light pu
Nephelometry t
Spectrometer
q Sample
White light input

q t
spectroscopy for a white light input

Figure 5.10 Spectral nephelometry characterises samples (usually liquids) for

both absorption and scatter as a function of angle. The detector diodes are
indicated by D1 and D2.

laser-induced-breakdown spectroscopy (LIBS) reveal much about the air that

we breathe. Particulates are present in air, and light scatter reveals much about
this form of atmospheric pollution. Sometimes it is obvious in the smog which
affects many cities worldwide. The atmosphere is, however, the simplest
puzzle to solve optically. Absorption spectroscopy can uniquely identify
nitrous and sulphur oxides, for example, and can accurately measure the levels
of carbon dioxide. Particulate scatter is also straightforward to measure and
readily characterised.
Solids and liquids present different issues in that – as we have already
discovered – the spectra of individual elements within a particular compound,
and those of the compositional components in a mixture, overlap, so that direct
spectroscopic measurements are much more difficult to interpret. Figure 5.10
shows one of the many ways around this difficulty, which is based upon
measuring a combination of scattering signature and absorption spectrum
coupled into elegant signal processing – a technique known as spectral
nephelometry. This has enabled differentiation among olive oil samples, beer,
whisky and wine and is becoming accepted as an evaluation tool (Figure 5.11).
Water analysis presents other challenges, many associated with its almost
total transparency in the visible. There are, however, important features in both
the near infrared and the ultraviolet. Additionally, scattering can give an
indication of particulate contamination, though bacteriological characterisation
requires, thus far, the biochemical analysis of samples. Nevertheless the future
may well hold other prospects, involving for example induced fluorescence,
optically active test bacteria seeking complementary species or advanced
imaging.
Hyperspectral imaging is another important technique, particularly for land
and atmosphere characterisation. The technique entails taking the spectral
 5.9 Environmental Analysis 101

(a) Transmission spectra at 0 degrees

0.4

Normalised Typical
transmittance extra
virgin

0.2 Typical
non-
extra-
virgin

0.0
500 550 600
Wavelength (nm)

(b) Transmission spectra at 30 degrees

0.4

Normalised
transmittance
0.2
Typical non-
extra-virgin

Typical extra virgin

0
500 550 600
Wavelength (nm)

(c)
EXTRA VIRGIN

0
PC3
-5
-10
-15 NON-EXTRA-VIRGIN
20 –10
10 0
PC1 0 10 PC2
–10 20

Figure 5.11 Extracting information from nephelometry data using principal com-
ponents analysis to separate extra virgin and non-extra-virgin olive oil. There are
many applications for the technique, and numerous signature identification
algorithms.

signature, in the limit, pixel by pixel, of a photographic image. As an example,

recall that many gases have significant absorption spectra in the near infrared.
Methane absorbs strongly at about 3.3 μm wavelength – a wavelength at
which there is considerable solar illumination. Suppose, then, we take a picture
along a gas pipe with a special camera (FLIR, forward looking infrared) which
102 Photonic Tools

is filtered to look at this specific wavelength only. If there is an actual gas

escape then this will absorb the background radiation at the position of the leak
and appear as a black cloud. This idea is used in practice, and not only for
methane but for many other species. The principle can also be generalised to
examine, for example, crop distribution through satellite imaging by chromat-
ically characterising a reflected spectrum to give a plot of land fertility, arid
and moist regions, or the distribution of crops. Environmental assessment
currently features among the principal applications of photonic technologies.

5.10 Light as a Therapy

Biomedical photonics has become an increasingly important discipline and
continues to expand as the many biological and physiological influences which
light can have become more accurately characterised. Many diseases of the
skin respond constructively to ultraviolet light, typically in the 250–400 nm
band. There is also evidence that appropriate photonic treatment can enhance
the healing of cuts and bruises and assist in skin cancer therapies. More
recently neurological possibilities have been explored, where very precisely
positioned optical probes, typically fed through fibres and coupled into the
appropriate source wavelength, can implement beneficial neurological changes
in the brain.
The influence of light and spectral distribution on sleep patterns, not least as
affected by jet lag, and on sleeping disorders, is well documented, the red and
orange part of the spectrum (with the lowest frequencies) being much more
restful than the higher-frequency blues and violets. (Perhaps the reds and
oranges are reminiscent of a soothing sunset?) Ultraviolet light has been used
as a sterilising agent both in water treatment and in bacteriological control in
humans. This just gives an indication of what is currently happening in
phototherapy, which promises to become an immensely productive area for
future research and application.

5.11 Astronomy and Very Large Telescopes

Stars have been a source of fascination since ancient times, driving a desire for
ever improving tools to observe the distant universe. The essential issue
remains how to get the maximum light from these distant worlds to observe
details beyond the capability of the naked eye whilst simultaneously minimis-
ing the ‘twinkle’ which results from even very minor atmospheric
 5.11 Astronomy and Very Large Telescopes 103

perturbations. Gathering the maximum light implies greatly expanding the

collection area, whilst maintaining highly precise aberration-free optics, which
in turn inevitably increases a telescope’s sensitivity to atmospherically induced
twinkles. The requirement for large aberration-free optics leads to the use of
mirrors rather than lenses, which both simplifies the surface preparation and
minimises the overall system weight and thereby improves a system’s rigidity.
Minimising the twinkle points towards taking apparatus up mountains or
indeed into space.
Mount Palomar in California was the first of these high-attitude locations, at
about 1700 m altitude and with a 5 m telescope. Both the height and diameter
of such telescopes have increased since then, and Mauna Kea in Hawaii at
4200 m hosts a range of instruments with mirror diameters to 10 m. The
principal secret of all these telescopes, and indeed of the astronomical equiva-
lents such as the Hubble and the Webb, lies in achieving the necessary
precision over a huge area in order to produce a parabolic surface for large-
aperture high-resolution imaging; the mirror should not deviate by more than a
fraction of a wavelength from an ideal parabola over its entire surface.
Simultaneously, the necessary supporting framework, compatible with trans-
portation along a difficult route, whether up a mountain or in a rocket, must be
realised. The mechanical design must both withstand rough handling and also
retain its shape through all the operational conditions to which the telescope
will be subjected. Whilst the basic optics shown in Figure 4.6 illustrates the
principles, it belies the huge feats of mechanics and precision assembly
necessary for a successful implementation.
There have been enormous improvements in the basic design (Figure 4.6) of
such telescopes over the past few decades, of which perhaps the most import-
ant aspect concerns the adaptive optics (Figure 5.12), which facilitates very
precise mirror shape control to correct for thermally induced strains and other
mechanical stresses, as might occur during the launch of systems designed to
be extraterrestrial.
These adaptive optical systems also offer another important benefit in
addition to fine tuning the initial mirror machining: they are used to correct
for atmospheric perturbations and to ensure that the telescope remains
focussed on a particular star over long periods, to give the longest possible
exposure in any photographic record. The essential concept behind both these
functions is the ‘guide star’: the telescope motion system and the adaptive
optics are both guided by a particular celestial object. This guide star needs to
be bright enough to be detected rapidly and hence to provide input to the
feedback system, and it also needs to be close to the area of interest on the sky.
In the absence of a bright guide star, a terrestrially located laser beam provides
104 Photonic Tools

Adaptive optics adjustment – feedback from guide star

5 metres
Front surface parabolic
or more
reflector
Rays from distant galaxies
Guide star

Image capture
e.g. by camera

Rigid
mirror
mount Focal point

Figure 5.12 Adaptive optics precisely adjusting the parabolic reflector in, for
example, a very large space-borne optical telescope.

similar functionality, especially in correcting for atmospheric aberrations.

Tuning the pointing direction of the telescope then becomes a little more
involved, but combinations of celestial patterns, a less bright guide star and
precision navigational instruments to lock onto the reference-laser guiding
beam can ensure that everything functions well.
Nevertheless, the astronomical community continually yearns for more light
gathering, that is, for even larger telescope areas with the consequent enhance-
ment in image resolution and the corresponding need for even more stability in
image acquisition. Venturing into the extraterrestrial is an inevitable conse-
quence, not only in the search for ever more detail in the astronomical map but
also in endeavouring to answer questions about extraterrestrial life. The James
Webb Telescope coordinated through NASA is an example of such a project in
progress. The Webb will have a 6.5 m primary mirror and will orbit a million
miles from earth, with all that implies – especially from the presence of solar
heat and cosmic rays. There are also terrestrial aspirations – the European
Extremely Large Telescope, with 39 m primary mirror, is finding a home high
in the mountains in Chile.

5.12 Optical Tweezers

Optical tweezers have evolved since they were initially demonstrated towards the
end of the last century into a much used tool in biomedical sciences and similar
applications involving the manipulation and spatial stabilisation of tiny particles
such as biological cells. The essential idea involves the carefully controlled
 5.12 Optical Tweezers 105

Indicative power distribution in beam

Input laser beam

Objective lens
Forces on particle

Radial resultant,
opposing incident-
laser pressure

Trapped particle Radial forces

Longitudinal
incident-laser
radiation pressure

Figure 5.13 The basic principles of optical tweezers. Just below the waist of the
beam the longitudinal radiation pressure on a particle is balanced by the longitu-
dinal opposing component of the radial forces due to refraction through the
particle; the momentum transfer is against the photon flow and can balance it.

utilisation of radiation pressure and the gradients thereof in a focussed laser beam,
as indicated in Figure 5.13. Radiation pressure occurs when a photon, of energy hν
and momentum hν/c, is reflected from a surface of a particle; then the particle feels
the rebound as a force 2hν/c, corresponding to a pressure 2I/c where I is the
incident light intensity (the optical power per beam area).
The forces on the particle from the radiation pressure will increase towards
the focal point of the lens and there will be a cross sectional gradient tending to
keep the particle in the centre of the beam. As the light strikes the particle,
some will be refracted through it and will emerge at a different angle,
imparting a momentum change to the particle equal and opposite to the
momentum change of the light. When the particle is just below the focus (or
waist) of the beam, its momentum change has a component towards the laser
source and under appropriate conditions the radiation pressure and refraction
forces balance to suspend the particle in the beam (for further information see
the Wikipedia article on optical tweezers).
The forces involved are very small, in the 100 nano-newton range, and the
focussed spot is typically produced by a high-aperture microscope objective
lens with input laser power in the mW range. Particles with sizes from a few
microns down to 5 nm can be trapped, facilitating the examination of objects
ranging from biological cells to DNA molecules.
Optical tweezers have evolved into many formats since their inception over
30 years ago. They are used in microfluidic experimentation to trap and
106 Photonic Tools

examine cellular structures in biological assays, to furnish microfluidic

pumping systems, for trapping microscopic liquid droplets, for trapping and
examining particles in Brownian motion and for a host of other manipulative
functions on the microscopic, or even nanoscopic, scale. Radiation pressure is
a very versatile tool.

5.13 Summary
This applications chapter has indicated the diversity and versatility of photon-
ics as a tool, as an enabler, as a corrective treatment. Of necessity, much has
been omitted and all the discussions had to be brief. Nevertheless, hints have
been given of the roots of the large expansion in photonics since the mid
twentieth century, The discipline is merging into everyday life, with DVD
players, fibre optics to the home, powerful photographic tools and displays, the
laser pointer, laser surgery and machining, photonic therapies and photonic
systems in the local opticians.
There is also is a great deal more emerging for the future. The functionalities
offered by photonics will expand enormously. The fabrication tool set, espe-
cially for extremely high-precision machining at the nanometre level, and also
the conceptual framework for future prospects, will broaden significantly. We
shall briefly explore some of these emerging possibilities in the next chapter.

5.14 Problems

1. The bandwidth of an optical receiver is basically determined by the product

of the photodetector’s capacitance C and its load resistance R. A bandwidth
of 1 GHz is well below what is achieved in many optical systems.
As an example, take beam with received power, at 1.5 micron wave-
length, of 1 microwatt with 100% quantum efficiency. Now, arrive at some
reasonable values for R and C for this bandwidth remembering that the
thermal noise current in the resistor needs to be below the shot noise current
for optimum performance. Look up achievable values for the junction
capacitance and comment on (a) whether that would be all the relevant
and unavoidable capacitance at 1 GHz, and (b) any trade-off between the
necessary load resistances, the desired bandwidths and the received shot-
noise-limited optical power, after arriving at a reasonable operational signal
to noise ratio for the detection system.
 5.14 Problems 107

2. Grating spectrometers are regularly used for many types of spectral analy-
sis. You are seeking a spectrometer with an operating wavelength range
from 1 micron down to 200 nm with a resolution of 0.5 nm.
(a) What would be needed in terms of the detection array, i.e. the number
of sensors in the charged-couple device (CCD) detector, the aperture of
the input lens and the power spectral density of the illuminating white
light source? What would be an acceptable signal to noise ratio for
each detection point? Remember that you will need to quantify the
attenuation in an absorption band with an accuracy of, say, 1% of the
input source in order to achieve this performance level.
(b) How might the signal to noise ratio be enhanced without changing any
system elements (except in the detection process – assuming that time
is not too pressing), and also what might be the implications of taking
the desired resolution down to 1 pm?
3. Fourier transform infrared spectroscopy (Figure 5.3) is noted for its poten-
tially high resolution. Consider a spectrometer that is required to operate in
the range 0.5 to 1.5 micrometres with a resolution of 10 pm over that range.
(a) What would be the necessary travel distance necessary for the moving
mirror? From your answer make some estimates – with reasons – of the
ultimate resolution of such a system.
(b) Could it go to GHz resolution, or better, for example?
4. Optical coherence tomography (OCT) (Figure 5.4) is a widely used tool.
(a) For an optical source operating at 1 micron what would be the minimum
spectral width in nanometers needed to achieve a depth resolution of
5 microns? Would such a source be practically available?
(b) From your observations comment on the practical limits in depth resolution
which may be achievable. Assuming that the necessary precise (how
precise?) mirror drives are available, what could restrict the possible depth
range through which an image may be constructed?
(c) The major applications of OCT lie in medicine, predominantly
ophthalmology, but OCT has been used in, for example, measuring
the thickness of a silicon wafer. Discuss – with reasons – whether
the same system which works for silicon could also be effective for
measurements on the eye.
5. The fibre optic gyroscope (Figure 5.7) relies on the fact that light emerging
from the fibre travelling in the direction of rotation will stay in the fibre for
slightly longer than the light travelling in the other direction. It is based on
the Sagnac effect, originally described over a century ago.
(a) Using this basic observation, arrive at a simple expression for the phase
change in a fibre gyroscope as a function of the source wavelength, the
108 Photonic Tools

fibre length, the coil radius and the rotation rate. Then check to see
whether your answer is appropriate (there are good web sources
for this).
(b) From this expression make some observations on what is needed to
obtain a sensitivity of 0.0005 degrees per hour in a loop the size of
that indicated in the figure (you can work out the rough magnitude
of the dimensions from the cables and connectors).
6. This problem concerns the underlying principles of optical tweezers. Sup-
pose we have a laser beam of total power 1 mW focussed through a
microscope objective to a spot 2 microns in diameter.
(a) Make an estimate of the radiation pressure force exerted on a disc one
micron in diameter placed in the centre of the beam.
(b) From the above estimate, indicate how these forces might distribute
themselves around a spherical object of diameter 1 micron.
These estimates will give an order-of-magnitude insight into tweezer
operation.
 6
The Future

Light, as we have seen, has being intriguing features. We think of it as

generated and detected as particles, namely photons, but these particles are
transmitted as waves (or sometimes can be accurately modelled as rays). The
wavelengths of these waves are typically submicrometre. Most of the current
prospects for photonic systems lie in combining the concepts of electromag-
netic waves and photons with the ever expanding capability to manipulate
materials and material structures to optically subwavelength tolerances. Speci-
fying dimensions to nanometre precision is no longer wishful thinking.
The electromagnetic properties of materials, and the ability to use these
properties in conjunction with structures at subwavelength tolerances, have
already had a profound effect on the evolution of ‘conventional’ electronic
systems. The ever increasing complexity of integrated circuits, the evolution of
the compact mobile telephone and a host of other, now commonplace, devices
have all been enabled by precision structural machining. A piece of copper
wire has varying electromagnetic properties depending upon the frequency of
the current going through it and its geometry; the simplest example of this is
the contrast between a coil and a straight wire. A vital difference in photonics,
however, is that whilst at lower frequencies generation and detection are best
viewed as involving electric currents, for photonic frequencies the generation
and detection of the electromagnetic wave as a photon must also be taken into
account. This opens up a whole new dimension of possibilities. Could it be
possible in a small enough device to generate an optical wave using an
alternating current at 1014 Hz (a wavelength of 3 microns)? However, thus
far the fast evolving precision machining capability has received much more
attention. In this chapter we shall briefly discuss prospects which are emerging
in both these conceptual domains.
Photonics also facilitates a range of new and important applications
influencing not only communication networks but also, environmental

109
110 The Future

Objective
Pinhole lens
aperture Beam splitter
Light Sample
source

High-
Typically,
precision
laser for
sample
spatial
drive
coherence
Pinhole
Detector
aperture
matched
to source

Figure 6.1 The basic elements of confocal microscopy, giving a high-resolution

point-by-point scan of the sample.

characterisation and medical diagnostics and therapy. To complete the explor-

ation of the future potential of photonics we shall also speculate briefly on
where photonics could contribute beyond its current domains.

6.1 Super-Resolution Imaging

The conventional microscope is limited in its angular resolution capability by
the optical wavelength and the numerical aperture of the objective lens: the
limit in angular resolution is of the order of the optical wavelength multiplied
by the numerical aperture (or, alternatively, in photographic terms, the wave-
length is divided by the f number of the lens). In practice, realising this limit is
difficult, owing to the effects of non-ideal illumination and depth of field,
which combine to produce a blurred image. The middle of the twentieth
century saw the arrival of confocal microscopy (Figure 6.1), which produced
substantial improvements by, in effect, placing pin holes in both the illumin-
ation system and the imaging system in order to obtain much higher resolution.
However, Figure 6.1 also demonstrates that the image has to be collected point
by point, with the implicit need for high mechanical precision scanning of the
sample with respect to the optical system. Additionally, even at its best,
fundamentally this cannot break Abbe’s diffraction limit on imaging reso-
lution, which states that the maximum angular resolution is well approximated
as a/λ, where a is the imaging aperture and λ is the wavelength.
The implicit assumption in this imaging limit is that the object, when viewed
through the imaging system, can only be reconstructed as a far-field image; as
 6.1 Super-Resolution Imaging 111

Tapered (single-
mode) optical
fibre end

Metal coating
(Al, Au etc.)

Taper aperture,
typically <100 nm
Sample
diameter

Figure 6.2 The scanning near-field optical microscope, which has a tapered
single-mode optical fibre with metal coating as the illuminating source to ensure
near-field operation. The tip maintains a constant subwavelength separation from
the sample, using a drive similar to that in an atomic-force microscope (see the
main text).

a general observation, the eyepieces in microscopes and telescopes all project

the resultant image to infinity. If, however, the image could be constructed
from the near field, gathering both the phase and amplitude of the illumination
reflected from the object simultaneously point by point, then this restriction no
longer applies. This basic idea of using the near field has been used for decades
in, for example, measurement probes for electronic circuits, which plot an
alternating signal passing through a circuit as a function of position (i.e. point
by point) within it. The same concept has evolved into near-field scanning
microscopy in the optical domain.
The concept is shown in Figure 6.2. The optical-field probe tip confines the
light injected from an optical fibre to a very small, carefully defined, area. It
does this by using a metal coating designed to optimise the trade-off between
the inevitable high lossed incurred as light travells along a metallic path and
the degree of light confinement that can be achieved. This fibre probe typically
both illuminates the object and also collects the locally scattered light field.
The coatings are typically gold, to minimise losses, and the probe's field of
view can be confined to a few tens of nanometres. Such a probe is usually
cylindrical, following the symmetry of the fibre, but some modifications, as
found in, for example, the slit probe can offer specific benefits for particular
applications.
This type of probe needs to make contact with, or be a consistent, very short,
distance away from, the surface of the sample – it is a direct analogue of
the electrical circuit forerunner mentioned above. However, the positioning
112 The Future

systems for the fibre probe need to be very precise mechanically, to either
maintain consistent contact with or, by far the more commonly encountered, to
maintain a stable (to nanometres) distance from the sample.
It is here that another precision microscope technology – the atomic force
microscope (AFM) system – makes its contribution. The probe from the
AFM is modified to hold the optical probe and maintain a highly precise, tens
of nanometres, spacing between the probe and the surface. This facilitates a
local pick up for the optical field and thus generates what has become
referred to as a super-resolution image. The system demands precision
fabrication of the probe itself, especially if it departs from cylindrical sym-
metry. Achieving the necessary precision of the measurement system to
determine the location of the drive network and the precision needed within
the scanning x, y and z stages themselves presents challenges. At present, the
use of these instruments is confined to specialist applications though there are
prospects for applying the basic ideas using simpler and more adaptive
probes. One example at the time of writing is the ezAFM+ from NanoMag-
netics Instruments.
An alternative approach to achieving effective super-resolution imaging is
to treat the sample in such a way that the areas of interest have a highly non-
linear reaction to the illuminating light. Possibly the most reported example of
this concerns fluorescent imaging in biological systems using a cell-selective
label. In this case a somewhat modified version of confocal spectroscopy, in
which a pinhole is used to block out-of-focus light and so improve the image,
can achieve a resolution exceeding the diffraction limit by looking at the
fluorescence induced through the scanning illumination and by knowing, from
the diffraction pattern, the shape of the confocal spot. In principle, then, a
biological cell with dimensions in microns, which is visible using a normal
microscope, can be treated with a fluorescent marker; then, under suitable
illumination and precision scanning, the exact location of this marker can be
determined to accuracies limited by the non-linearity of the transfer function
between the illumination and the consequent fluorescence. Typically, this
corresponds to 20% or thereabouts of an optical wavelength.
This discussion has merely touched on the topic of high-precision imaging
using light. There are numerous other techniques, some requiring well-defined
but complex illumination functions and some requiring advanced deconvolu-
tion mathematics (the received image is a convolution of the structure with the
resolution profile of the system), and there are numerous other labelling
techniques. Currently, all these methods are complex and they all rely upon
precision scanning drives to provide location information. The basic principles
are, however, well rehearsed, so speculation for the future of high-resolution
 6.2 Photonic Integrated Circuits 113

Surface electrodes for modulator Diffused waveguide dimensions ~ microns

Substrate
Light in Light out

Waveguide
End cross section

Figure 6.3 An integrated-optics diffused-waveguide Mach–Zhender interferomet-

ric modulator. The waveguide widths are of the order of a few microns.

imaging is centred around the prospects for making the technology more
readily accessible to a wider audience. There are obvious uses for such
technologies, especially in biomedicine, so there is good reason to be optimis-
tic that they will evolve. For example, three-dimensional printing approaches
to fabricating the probes are already emerging.

6.2 Photonic Integrated Circuits

The concepts for ‘optics on a chip’ emerged in the mid twentieth century, as
silicon integrated circuits became a real prospect. At this stage, the preferred
material, lithium niobate, was expensive and difficult to process reliably.
However, demonstration optical circuits with and without modulators (which
modified the material by the electro-optic effect; see Figure 6.3) showed
intriguing functionality in the visible and in the very near infrared – the region
which, at the time, was of greatest interest. It was in this era that one of the
pioneers observed that ‘integrated optics has been, is now and always will be,
the technology of the future’, because of the idiosyncrasies of lithium niobate
processing, the need to have it in crystalline form and its intrinsic incompati-
bility with the formation of detectors or optical sources on substrate. Addition-
ally, silicon was getting all the attention and all the investment.
The benefits of large-scale integration in silicon are self-evident in everyone’s
laptop computer and mobile phone. Silicon can also potentially act as a very
good optical waveguide material, since most present-day photonics operates at
wavelengths longer than the 1.1 µm cut-off for guided waves in silicon.
Silicon has, however, other benefits too – it can be machined using etching
processes and photolithography into a rich variety of optically interesting
shapes, from simple artefacts such as V-grooves for aligning optical fibres
114 The Future

Optical fibre
sits here

Figure 6.4 Some simple silicon micro-optics; on the left a V-groove alignment
system for fibres and on the right a rotatable mirror approximately 100 µm in
diameter.

with waveguides to much more complex structures such as rotatable mirrors

(Figure 6.4). Silicon has the huge benefit of being transparent in the intrinsic-
ally very important 1.5 μm region, and consequently the late twentieth century
saw increasing interest in silicon as the vehicle for photonic integrated circuits.
The inherent versatility and straightforward availability of the basic silicon
processing platform subsequently triggered a diversity of integration-
compatible optical components. Waveguides can be integrated onto the sub-
strate using a vertical etching technique, resulting in a guide lying upon the
surface. Wavelength filters can be realised using a variety of grating-based
approaches. Splitters and combiners are relatively straightforward extensions
of the basic waveguide structure. Finally, modulators can be implemented
using various forms of free-carrier modulation which influence both the optical
loss and the optical phase of light passing through the modulation region.
Using this principle, modulation rates approaching 100 Gbits/s have been
reported (Figure 6.5).
This inevitably all applies in the near infrared band, where silicon is
transparent; additionally, the question of sources and detectors compatible
with this impressive interconnect facility needs to be addressed. For the
sources, there are two basic approaches. The simplest and best established is
to combine appropriate III–V semiconductor lasers on a silicon substrate in a
suitable geometry in order to couple light into the silicon waveguide. These
‘hybrid silicon lasers’ receive much attention and are proving to be an effective
source.
For detectors the above III–V approach is also valid, but silicon also has a
crystalline structure that is geometrically compatible with germanium, so the
two can be grown together to facilitate a fully integrated detector. These ideas,
illustrated schematically in Figure 6.6, in principle allow a full system
 6.3 Nanophotonics and Plasmonics 115

Rib waveguide cross

section ~100 nm

Figure 6.5 Silicon integrated optics exemplified in (left) a rib waveguide and
(right) the tight bends which a high-index difference permits.

Metal Metal
contact n Si n++ contact Matching optics for
p Si Overlay source waveguide launch
SiO2 Contacts
p++Si
SiO2

Silicon substrate Silicon substrate

Figure 6.6 Some sample silicon circuits including (left) a modulator and (right) an
integrated source.

component set on a chip. Also, this chip can be shared with all the necessary
electronic drives to enable versatile, integrated, optoelectronics platform which
is compatible with the essential elements of CMOS (complementary metal-
oxide semiconductor) processing. At the time of writing this it has yet to
become a mass production item, but the signs are definitely there for photonic
integrated circuits to emerge as a ‘technology of the present’ relatively soon.
Custom-built systems for specialised applications are already emerging.

6.3 Nanophotonics and Plasmonics

Nanophotonics is another example of emerging photonic tools and techniques
which rely upon the fabrication of subwavelength structures and also on an
understanding of the behaviour of light within these structures. Plasmonics is
an integral part of this since it is the study of the behaviour of light within
116 The Future

~ 100 nm ~ 1 micron

(a) (b)

Figure 6.7 Examples of simple nanophotonic circuits including (a) a micro-

resonator, where the gap is roughly equivalent to a capacitor and the ring itself
to an inductor, and (b) a Yagi array antenna for use at optical frequencies.

metallic structures with these dimensions. For metallic structures the concept
of ‘conductivity’ is well understood, and plasmonics is essentially this concept
applied to optical frequencies but recognising that the free electrons in a metal
can change their behaviour as the plasma resonance is approached. Nanopho-
tonics and plasmonics are closely interlinked and have been described, by at
least one of the pioneers in the field, as transferring electrical circuits into the
photonic frequency range, though with due recognition of the differences in
electrical properties at these very much higher frequencies.
There have been many demonstrations of nanophotonic circuits, a couple of
which are shown in Figure 6.7 illustrating a ring gap resonator, which operates
very similarly to an inductance and capacitor, and a Yagi antenna array, which
is very close conceptually to the ultra high-frequency TV receiver antenna and
indeed performs in a similar way.
There are other features of nanophotonic circuits which are broadly analo-
gous to their much lower-frequency counterparts. Perhaps the most useful of
these is the exploitation of metals as electric field enhancers at a metal –
dielectric interface, particularly for pointed metallic structures. An optical
‘lightening conductor’! This concept has been used for many years, particu-
larly in the Kretchmann configuration as a sensitive probe for the dielectric
constant of liquids (Figure 6.8(a)). This basic concept has also been
coupled into photonic integrated circuits illuminated by a fibre waveguide
(Figure 6.8(b)). The generic idea has also been used to enhance the electric
field in solar cells and significantly increase the efficiency of some biophotonic
therapies.
Nanophotonics and plasmonics currently attract widespread and enthusiastic
research endeavours, since even this brief discussion should have indicated
their remarkable potential. In common with all the topics mentioned in this
chapter there are textbooks and journals dedicated to these topics.
 6.4 Metamaterials 117

(a) The guided-wave phase-matching aspects (b) A multichannel fibre-end

of the Kretchmann configuration chemical sensor

Thin
TM (P) wave noverlay
metal
launched. film
nsubstrate

Figure 6.8 A nanophotonic field-enhancement system (a) in the guided-wave

traditional Kretchmann configuration. The launched light must phase-match pre-
cisely into the propagating light, with the field distribution indicated. (b) Metallic
prisms on the end of a fibre, with different segments each targeting a separate
chemical species on a probe just a few microns in diameter. The electric-field-
enhancement properties of the nanophotonic metallic islands are critical to its
operation.

6.4 Metamaterials
A metamaterial (mentioned briefly in Chapter 4) is conceptually a material
ensemble with structural dimensions far below the wavelength of the electro-
magnetic (or indeed acoustic) fields with which it interacts. Through this
interaction, structural variations in the metamaterial can produce a response
that would otherwise be unobtainable. Whilst structures such as gratings
could be argued as falling within this classification, a moment’s reflection
demonstrates that the optical properties of these structures are dictated by
the properties of their component materials and the way in which these
component materials are organised. The component materials behave in such
structures exactly identically to their bulk equivalents. Metamaterials, in con-
trast, are made-to-measure three-dimensional materials with structural vari-
ations in dimensions considerably less than a wavelength, resulting in an
overall material which behaves as a bulk system with entirely different prop-
erties. The incoming optical wavelength is too long to ‘see’ the structural
artefacts!
Typically, metamaterials can comprise arrays of nanophotonic circuits
(Figure 6.7) with resulting properties dictated by this structure (resonance,
for example). Alternatively, arrays of other material structures can be arranged
in such a way that the incident light beam ‘sees’ the average properties of the
structure (i.e. any structural variations must be well controlled and extremely
small compared to a wavelength). Another possibility is arrays of quantum
dots, where the optical properties of the materials themselves are modified by
118 The Future

dimensional features small enough to affect the quantum mechanical energy

levels within the structure. And all these ideas – and others – can in principle
be combined to synthesise almost any conceivable material response!
These brief qualitative comments already point to intriguing prospects in
terms of manipulating light transmitted into a metamaterial, and therefore open
up opportunities to implement previously intractable functionality. Perfect
imaging and cloaking are possibly the most discussed examples.
Perfect imaging thwarts the Abbé resolution criterion by utilising a material
with a refractive index of 1 to produce the imaging function. Assuming that
the material is perfectly homogeneous from the perspective of the light passing
through it and that it has perfectly parallel sides and perfectly flat surfaces, the
ray diagram, as shown previously (Figure 4.17), demonstrates paths through
which a perfect image is formed. It images using ‘rays’ rather than ‘waves’!
The imaging material also has to be thicker than the separation of the material
and the object. Potentially there is the scope for effective magnification
through making ever thicker samples of a material with index 1. From a
‘Clausius–Mossotti’ perspective, a negative index can be achieved through
configuring a synthetic molecule-like structure which reradiates the input as an
antiphase output – remember that is the context of refractive index it is
phase velocities rather than group velocities which dictate what happens.
There has been some practical progress in realising negative-index optical
materials (though not as yet the super-lens envisaged as a lens composed of
material with index 1), but this has been achieved through nanophotonic
resonators evolving from the original electromagnetic demonstrations at
microwave frequencies. There are then practical constraints since the reson-
ators function over only a very narrow spectral range and need to be extremely
precisely matched for full efficacy. The super-lens, however, continues to
fascinate.
A ‘cloak’ renders the object which its surrounds totally invisible
(Figure 6.9). The necessary index distribution to ensure that this actually
happens can, in principle, be readily calculated for a particular candidate object
for cloaking. In practice, such a distribution is tractable only for simple objects,
for example cylinders or spheres. Again the nanophotonic resonator appears as
a key component so cloaking for all colours of the rainbow clearly presents
issues, because resonance only occurs for one wavelength. The cloaking
principle has, however, been successfully demonstrated, most notably at
microwave frequencies where the fabrication tolerances are four or more
orders of magnitude less demanding than at optical frequencies. The invisi-
bility cloak has an instant and easily communicated appeal and continues to
fascinate the photonics community, not to mention the world at large!
 6.5 Photonics and Materials – A Brief Summary 119

Input light
neither deflected
nor reflected

Figure 6.9 How cloaking might work – but broadband cloaking, even for simple
shapes, and full cloaking for complex shapes have yet to be realised.

6.5 Photonics and Materials – A Brief Summary

In the preceding sections we have briefly discussed super-resolution imaging,
photonic integrated circuits, plasmonics and nanophotonics and metamaterials,
all of which are establishing themselves as essential building blocks for future
photonic systems and their applications. At a conceptual level, they all rely
upon three essential building blocks:
• A thorough understanding of optical materials and their properties at photo-
nic frequencies. This also includes the interfaces between different material
components and the variations in interface behaviour with the overall struc-
tural dimensions of a particular material with respect to the wavelength.
• A thorough understanding of the behaviour of ensembles, particularly those
on a regular repeating pattern in two and three dimensions, again with due
attention to the ensemble component sizes with respect to the optical wave-
length. Effectively this is a generalisation of the Clausius–Massotti equation,
which relates what is absorbed by to what emerges from a component after
excitation by an electro magnetic wave.
• A full appreciation of the effects of structural geometry within a sample of
material comprising the synthetic optically active elements mentioned
above. In particular, this applies to three-dimensional regular periodic arrays
with specified quasi-particles on each of the array nodes.
120 The Future

• Interfaces, metal
and dielectric
Material
properes • Dimensional
variaons >>
wavelength

• Periodic arrays e.g.

Ensembles
Designing at ~opcal
grangs, 2 and 3
dimensional
Opcal wave-
lengths • Wavelength filtering
Materials effects

Ensembles • Tiny structures and

well below ‘circuit elements’
opcal
wavelength • Tiny and the
quantum dot

Figure 6.10 The building blocks for designing optical materials and the essential
options dependent upon critical scale.

These three conceptual building blocks (Figure 6.10) are common to any
situation in which waves interact with materials, whether the waves be elec-
tromagnetic, mechanical, acoustic or quantum mechanical. They are at the
heart of much of our scientific and technological evolution.

6.6 Entangled Photons

The concept of ‘entanglement’ consolidated around the turn of the present
century. At a basic level, the concept is relatively straightforward. In this book
we have assumed that the reader is comfortable with the idea that light travels
as a wave and yet is detected as a particle, the photon. Consequently, if we can
build up an image photon by photon in, for example, the Young’s slits
interference experiment, then each photon emitted from the source would
arrive in the detection plane at a position whose statistical probability is
determined by the interference pattern. This would, in turn, reflect the wave-
defined multidirectional path through which a particular photon trajectory
happens to proceed.
Thus, much endeavour in the optical physics community has been
directed towards realising controlled single-photon sources. There has been
some success in this pursuit already. It opens up at least two intriguing
possibilities. First, if a single-photon source can be realised, then by impli-
cation a single-photon detector can also be achieved, and indeed this has
 6.6 Entangled Photons 121

been achived: the single-photon avalanche detector (SPAD) is now a cata-

logue item. This enables the detection, and therefore the characterisation, of
single photons emitted from an illuminated object. There are prospects here
in medical imaging through tissues in order to detect the position of a
catheter probe, by including a fibre source in the catheter and using the
arrival of the first photon to reach the SPAD to indicate the position of the
catheter. This assumes that at least one photon can penetrate through
the tissue without being scattered, an assumption may be plausible but
has yet to be definitely proven. The second possibility concerns
new-generation communications networks, where generating and later
detecting a single photon can contribute towards establishing reliable
connections.
The concept of ‘entanglement’ takes this somewhat further. There are
situations where an atom can generate two photons in a closely linked process
and therefore the properties of these two photons are intimately related, even
permanently bound together. Similar arguments can be extended to other
elementary particles when closely bound together through some single atomic
or nuclear process.
The intriguing features of these two (or indeed more) ‘entangled’ particles
is that after entanglement, even if the individual particles become separated
or take different ‘wave paths’ to their destinations, their overall net properties
will remain the same and they will need to ‘recognise each other’ in order to
be detected again. Consequently, perhaps a sequence of such entangled
photons could be formed to represent a coded message. The components of
these entangled photons would then be separated from each other in some
way and the encoded message could only be read when the two sets of
components come together again. This is the essential thought process
behind quantum communications and unbreakable code sequences, a generic
area of interest which has become increasingly important as the internet
rapidly expands, and security becomes increasingly critical. The recipient
of the code would need to have both components to hand; this would prevent
any interception.
The quantum computer is another intriguing potential long-term manifest-
ation of these ideas, though, as yet, it has only been demonstrated in a very
simple form. In contrast, entangled-photon communication systems were
demonstrated at the beginning of the twenty-first century and have been
gradually improved since then. However, the systems remain many orders of
magnitude slower than their conventional fibre-communication forefathers but
both the science and technology continue to progress leading to an, as yet
undefined, probably currently unknown, final form.
122 The Future

The entangled-photon concept has also asserted itself as potentially the

ultimate computational resource, and simple logic functions have been dem-
onstrated. For example, an entangled communication system can be interpreted
as a form of AND gate, which only gives an output when identical signals and
codes coincide in space and time.
Entanglement will continue to fascinate for a considerable time to come
whether for photons or other elementary particles. Quite how, when and for
what it will emerge into the outside world remain intangible questions. The
concepts do appear in advanced physics textbooks, but to date perhaps the
most accessible version of the story is in Jonathan Dowling’s book, mentioned
in the Further Reading section at the end of the book. This very readable text
tells the story for 400 pages without a single equation though it does contain
conceptually demanding and thought provoking accounts of the basic ideas
and their possible applications.

6.7 Graphene and Other Exotic Materials

The ever expanding capabilities to visualise, fabricate and characterise material
structures with dimensions measured in nanometres – much below the wave-
length of light – has facilitated numerous intriguing innovations some of which
will undoubtedly find their way into engineering applications. Of these gra-
phene (Figure 6.11) is among the most intriguing. This single-layer crystalline

Carbon atoms in the ‘carbon ring’ formation

–
–

–
–
–

Free electrons

Figure 6.11 Illustrating graphene: a single layer of carbon atoms linked together
with free electrons flowing within the layer. The atoms are about 0.14 nm apart.
 6.7 Graphene and Other Exotic Materials 123

100 1000

(W2)
graphene
Transmittance (%)

graphene
100

10 ITO
ITO
1

0
200 400 600 800 0.1 1 10 100 1000
Wavelength (nm) Film thickness (nm)

Figure 6.12 The properties of graphene in comparison with the currently pre-
ferred transparent contact material, indium tin oxide (ITO).

10.6-micron receive probe

10.6-micron
5 nm top layer Graphene layer
launch transmits
surface plasmons

27 nm
h-boron
nitride
Control electrodes:
central electrode 150 nm long, gaps 100 nm

Figure 6.13 A nanostructure system based on graphene giving a voltage con-

trolled phase shift of up to ~180 at 10.6 μm wavelength.

carbon has many interesting features. It has high electrical conductivity, it is

mechanically strong whilst also being flexible and it is uniquely transparent
across the visible spectrum and beyond (Figure 6.12). In the photonics context,
graphene is obviously a candidate for touch-screen displays particularly where
highly flexible screens are required.
There is much more to thin layers of graphene than their action as conduct-
ing electrodes and many intriguing optical–electronic interactions have been
demonstrated. Figure 6.13 shows an interesting example where a tiny struc-
ture, nanometres in dimensions, is sufficient to produce a full 180 voltage
controlled phase shift for a 10.6 µm laser wavelength. The diagram also
illustrates the use of a nanodimensional probe to receive the modulated beam.
(This may even be viewed as a slightly eccentric mid-infrared transistor!) The
science and application of graphene continues to excite and intrigue – there is
little doubt that its engineering potential will eventually be realised.
The structure in Figure 6.13 also involves another intriguing material; h-BN,
hexagonal boron nitride. This has a very similar layer-by-layer crystalline
124 The Future

structure to graphene but conveniently has a large bandgap (over 5 electron

volts) and a relatively low electrical conductivity, so its action is to preserve
the plasmon (i.e. the optical-frequency electric current) within the graphene
layer. This is another example of novel material exploitation in the photonic
domain facilitated by these ever improving fabrication and manipulation
technologies.

6.8 Manipulating Light Sources

There are other innovations in photonics which promise to facilitate significant
advances in performance. All involve in some way manipulating light with
materials or, strictly speaking, material structures, since it is increasingly the
combination material properties and structural properties which facilitates
the manipulation of optical signals. Squeezed light is but one example. The
discussion of noise in the context of photonics historically centred around
the concept of ‘shot noise’, which we have already explored; it represents the
random change in the rate of arrival of photons. We have also mentioned that
this can be viewed in the ‘analogue’ domain, with the wave represented by a
vector in which the shot noise impacts equally on the amplitude and phase.
However, if a means can be found to turn the phase fluctuations into amplitude
fluctuations, or vice versa, then beating the shot noise limit by squeezing the
light (to reduce phase noise) becomes a real possibility. This assumes that after
the squeezing has been completed, the detection processes that follow are
somehow made insensitive to the noise increase incurred as a result of the
process.
Whilst this technique has been already utilised in the Laser Interferometer
Gravitational-Wave Observatory (LIGO) as mentioned in the previous chapter,
the implementation in the gravitational-wave interferometer was optically
bulky and extremely expensive. Squeezed light, however, becomes more
interesting when it can, in principle, be combined with guided wave optics
and produced using nanometre-precision machined devices. (Figure 6.14).
Thus far examples have been based on using subtle non-linear interactions or
even micro (strictly nano) mechanical resonators, in both cases to inhibit phase
fluctuations at the expense of enhancing the intensity variations. The extent of
the squeezing using these devices has thus far been modest – one or two
decibels. The integrated circuit demonstrations sometimes also involved cryo-
genic cooling, so there remains much scope for improvement towards the
LIGO exemplar of well over 10 dB. The ideas of squeezed light emerged in
 6.8 Manipulating Light Sources 125

Cryostat
environment

Light input

Squeezed light output

~ 20 mm

Figure 6.14 An exploratory mechanical structure for squeezing light in the phase
domain, based upon nanoscale micromechanical resonators. The light grey area is
a vacuum enclosure. The resonator, shown in black, is freely mounted over a
cavity in the substrate structure. The striped element coming from the right-hand
side is an optically excited resonator. The long striped rectangles facilitate
coupling between the light and the mechanical resonator.

750 nm 500 nm 400 nm

Red Green Violet

400 THz 600 THz 750 Thz

Precision frequency lock on ‘carrier’

Precision frequency lock on repetition rate –

e.g 10 GHz gives 100 ‘teeth’ in each THz

400 THz 600 THz 800 Thz

Frequency

Figure 6.15 The basics of the frequency comb utilising a 2.5 femtosecond pulse in
the green and, in this example, based upon a 50 terahertz pulse repetition rate. More
typically, a 1 Ghz repetition rate would give 400,000 ‘teeth’ over the same band-
width, in principle enabling accurately controlled spectroscopic measurements.

the 1980s but the thought of beating the shot noise limit in a compact and cost
efficient format remains a significant challenge and a source of great interest.
A frequency comb is another example of an intriguing optical source. The
concept originated as a means towards high-precision spectroscopy, where the
source is locked through some form of multiplier to a radio frequency master
oscillator (Figure 6.15). The output from such a system when viewed in the
time rather than the frequency domain is a sequence of very short pulses
separated by the master frequency. The exact techniques through which such
‘combs’ can be generated involve ultra-precision non-linear optics and opto-
electronic design. The special feature of the frequency comb is that each
component is accurately phase locked with respect to the others, resulting, in
126 The Future

the time domain, in unrivalled phase noise performance among the various
pulses within the pulse train. The applications of this continue to be explored:
optical clocks using frequency combs have been demonstrated and highly
precise spectroscopy, particularly in the time domain, is another area of
exploration. Just as in squeezing, a few tentative demonstrations of combs
realised in photonic integrated circuits have begun to emerge. In common with
squeezing, frequency combs continue to attract much attention.

6.9 Summary
The aim of this final chapter has been to give some insight, albeit superficial
and far from complete, into the immense potential which photonics offers for
the future. There are continually evolving new photonic tools. In parallel,
continuing new societal challenges emerge for which the tools offered through
photonics promise innovative and effective solutions. Figure 6.16 attempts to
encapsulate at least some of this potential on the basis of our discourse
throughout the book and the remarkably consistent recognition among numer-
ous international organisations of the challenges which society currently needs
to address. There are other indictors too, including, for example, the fact that,
at the time of writing (2018) the European production output in photonics
exceeded that in electronics and the research and development endeavour
within this sector continually gains more momentum. There is indeed some
confidence in asserting that ‘photonics is the electronics of the twenty-first
century’.

• Make solar energy economical • Engineer better medicines

• Manage the nitrogen cycle • Secure cyberspace
• Advance health informatics • Engineer the tools of scientific
• Prevent nuclear terror discovery
• Advance personalised learning • Develop carbon sequestration
methods
• Provide energy from fusion
• Restore and improve urban
• Provide access to clean water infrastructure
• Reverse engineer the brain
• Enhance virtual reality

Figure 6.16 A view of the technological challenges currently facing society and
the areas to which photonics promises to contribute (based on a US National
Academy of Engineering 2009 list, still current, and similar to the listings for
many professional organisations, e.g. Photonics 21 in the EU, the National
Photonics Initiative in the US etc.). (The prevention of nuclear terror is separately
boxed because photonics is unlikely to make a direct contribution to this large-
scale technological challenge.)
 6.10 Problems 127

6.10 Problems
Problems involving predicting the future are by their very nature encroaching
on the totally inaccessible! Answers will be educated guesswork and will vary
with time and among communities. The whole ‘future scenario’ discourse is
one to speculate about with fellow enthusiasts. So the problems here are in two
categories: in section A, the reader is invited to look in a little more depth at a
selection of current ideas within the community and in section B there is a
discursive section inviting speculation on what might happen and why.

Section A
1. (a) How could a Mach–Zehnder interferometer in integrated optic format
become an intensity modulator (Figure 6.3)? Remember the electro-
optic effect and note that the early devices, of which this interfer-
ometer is an example, used lithium niobate as the substrate.
(b) What parameters in this device could limit its modulation bandwidth
and its possible modulation depth? How might this change if your final
system application were the design of a phase or frequency modulator?
2. (a) Figure 6.6 shows some silicon integrated optics. How does the modulator
(on the left-hand side of the figure) function in this particular case?
(b) When compared to the device in Figure 6.3 what might its limitations
be in terms of operational optical wavelength range and in achievable
phase and intensity modulation depths?
3. (a) Figure 6.8(a) shows the field distribution for a wave propagating along a
dielectric–metal–dielectric interface. Suppose that the lower layer (sub-
strate) is one face of an equilateral prism and light is incident on one of
the other faces. How does the incident angle on this face required to
produce propagation along the metallic surface depend on the propaga-
tion constant of the surface plasmon wave? How might this change if
the overlay index and thickness were varied? The clue lies in phase
matching at the interface. This prism arrangement is the basis of the
Kretchmann configuration used in overlay index measurements.
(b) why does this arrangement need a TM wave to be launched and what
would happen to a TE wave? (In a transverse magnetic (TM) wave the
electric field oscillations are perpendicular to the interface under
consideration, and in a transverse electric (TE) wave they are paralled
to the plane of the interface.) The system shown in Figure 6.8(b) has
no facility for phase-matched launching. Is this important and if so
(or, indeed, if not), why?
128 The Future

4. Figure 6.13 shows a phase modulator which has been demonstrated for the
10 micron wavelength region. Given the dimensions in the figure, and
given the observation that the signal travels from the launch electrode input
to the receive probe as a surface plasmon (i.e. an electric current), comment
on how the modulator modifies the phase delay and on any implications this
might have for the effective electron velocities in the graphene layer.
Figure 6.12 also contains some pertinent background.
5. Figure 6.14 shows a mechanical structure fabricated from silicon and
designed to perform optical squeezing, in order to reduce the phase noise
below the shot noise limit, albeit in practice to a modest extent. Consider how
this precisely micro-machined structure works Why does it need to operate in
a cryostat? How would you view the future prospects for such a concept?

Section B
6. There are many respected sources of information overviewing interesting
technical developments in photonics (Photonics 21, US National Photonics
Initiative and several others, including the professional societies in optics
such as SPIE and OSA). Look for the latest freely available reports on trends
from these organisations and compare and contrast these predicted trends for
agreements and differences. How might these trends map into the societal
needs identified by numerous other professional organisations such as the
National Academies of Engineering, the latest EU research programmes and
socio-technological organisations such as the Royal Society of Arts, UK?
7. Use collective imagination and insight (remembering that fresh views are very
valuable!) to consider other feasible photonics initiatives, such as combining
parallels from the bio-sciences, materials science, very-high-precision machin-
ing, quantum systems, electronics and information availability, for example.
8. Speculate on the detail of a few future technologically possible applications
suggested in Figure 6.16. Technological feasibility is, however, but one
factor influencing the eventual realisation of a new technology. Community
acceptance is perhaps the most important of these. Will people use it and be
willing to pay? What factors may come into this in the specific cases you
have chosen? Considerations on safety, and standards for demonstrating
safety, performance repeatability, rivalry with established practices aimed
at performing a similar function and many similar factors come into play.
Look up Porter’s five forces in this context.
 Appendix 1
More About the Polarisation of Waves

The basics of wave polarisation are very straightforward. When a polarisation

state is referred to as being linear it is simply an electromagnetic wave whose
electric field is vibrating in one plane, as shown diagrammatically in Figure
A1.1(a). Conventionally the electric vector defines the wave, and in the figure
the direction of polarisation is vertical. Figure A1.1(b) shows two waves in
phase with each other and of equal amplitude but one with the electric vector
horizontal and the other with the electric vector vertical, as in (a). Since both
these waves have the same amplitude, then what would be the amplitude and
polarisation state of the resultant when the two are added together?
In Figure A1.1(c) the horizontal and vertical components are 90 out of
phase. How would the resultant electric vector for this particular wave com-
bination evolve? If your eye is looking at it from the arrowhead point, how
would the resultant electric vector behave as it moved towards you? The
direction of rotation defines clockwise (right-handed) or anticlockwise (left-
handed) polarisation states. If the two waveforms in (c) were not exactly in
quadrature then how would the resultant state behave? As a hint the ‘not quite
in quadrature’ phase differential can be resolved into an in-phase and a
quadrature component. The net result is elliptical polarisation.
We have mentioned several times birefringent materials, in which different
polarisation components travel at different velocities. We usually consider the
case of linear components, shown diagrammatically in Figure A1.2, though
some materials are naturally circularly birefringent, depending upon the
molecular symmetries concerned – sugars are an example. In the linearly
birefringent case in Figure A1.2, an input wave (A) oriented at 45 to the
principal axis will split into slow and fast components. The input plane to
the birefringent medium is shown at top left in the diagram. The phase
difference between the two components increases as the propagation distance
through the birefringent material increases. For example, at point (B) in the

129
130 Appendices

(a) (b) (c)

Figure A1.1 The electric field vibrations in various polarisation states: (a) a simple
linear state; (b) a composite linear state with horizontal and vertical components;
and (c) a circularly polarised state with vertical components shifted by 45.

Air (A)

Input plane
(B)
Slow axis

Fast axis

Figure A1.2 Propagation of an arbitrary input polarisation state through a

birefringent medium analysed by resolving the input into states along the principal
axes of the medium.

diagram the two components are in quadrature so the output from this thick-
ness of the medium would be circularly polarised. If we were to proceed
further, the two components would become 180 out of phase, and the output
would be linearly polarised but orthogonal to the input at (A), and so on.
Birefringent materials used in this manner – often referred to as phase plates –
are extremely useful components in optical systems.
This moves us into questions concerning the influences of the input light on
the output fields. As an example suppose that the distance between the point
(A) and the point (B) in the diagram was kept constant but that the input
frequency was doubled. Assuming no net dispersion in the behaviour of light
travelling along the fast or slow axis then the output would be back to linearly
polarised but at 90 to the input.
This gives a hint on how a broadband source might behave when passed
through a phase plate of this nature in this particular format. The broadband
source would give slightly different polarisation states at the output depending
upon the wavelength of the input, and these differences will become greater
with a longer path within the birefringent medium, which can be, in principle,
extended as far as we wish. Thus the light ends up unpolarised. This assumes
that the input state is polarised, and for sources such as super-luminescent light
 Appendices 131

emitting diodes this is the case. A depolariser which operates in this manner is
often useful in conjunction with such sources.
The situation gets more complex if we remove the assumption of spatial
coherence – in other words if the input beam has components travelling in
many directions rather than simply in one direction as in Figures A1.1 and
A1.2. This is an interesting problem to ponder but in practice many situations
in optics can be visualised using the approaches that we have adopted here.

Further Reading

E. Collett, Field Guide to Polarization, SPIE Field Guides, Vol. FG05, SPIE, 2005.
 Appendix 2
The Fourier Transform Properties of Lenses

Here we aim to give an intuitive appreciation of the Fourier transform proper-

ties of a complex lens. Suppose we have (Figure A2.1) a grating with an
amplitude transmittance (the percentage of the input amplitude which emerges
after transmission through the grating) which is simply a sine wave, as
indicated. We can consider any amplitude transmittance function as a sum of
such sine waves through a straightforward Fourier transform relationship. Note
that here we are discussing amplitude transmittance throughout, although most
photodetectors, our eyes included, respond to intensity levels.
We will take the maximum transmittance as unity. The average amplitude
transmittance is shown as a dotted line. There will be interference between
amplitudes scattered from different parts of the grating; a more detailed analysis
(see for example Goodman’s excellent book) demonstrates that the two diffracted
beams indicated in the diagram have amplitudes proportional to the amplitude of
the sine wave indicated (but that this includes the ‘positive’ and ‘negative’
frequency components, on either side of the optical axis) and also that the
amplitude of the transmitted undeflected wave is proportional to the mean ampli-
tude transmittance of the grating. Furthermore, the deflection angle θd is given by
λ
sin θd ¼ (A2.1)
d
where d is the grating spacing and λ is the wavelength of the input light.
Conveniently, the quantity 1/d can also be referred to as the spatial frequency
of the grating so, for angles where sin θd can be approximated to θd the
diffraction angle is proportional to the spatial frequency. This above approxi-
mation is perhaps more versatile than might at first be guessed – even at a
diffraction angle of 45 the approximation is less than 10% in error. So we see
that the amplitudes of the diffracted components are proportional to the
amplitude of the spatial frequency component in the grating corresponding

132
 Appendices 133

Amplitude
transmittance
Parallel input Through beam
wavelength l

Position
Diffracted beams

Figure A2.1 The diffraction pattern from a grating with amplitude transmission
varying sinusoidally in one dimension.

Amplitude
transmittance

Position

Figure A2.2 A square wave grating – only the odd harmonics appear in the
diffraction pattern with amplitudes varying as the inverse of the harmonic number
(diffraction order).

to that diffraction angle and that amplitude of the straight-through portion is

proportional to the mean amplitude transmittance.
Moving to a specific example of a more general case, in Figure A2.2 we
consider a square wave grating. Here the lowest-order diffraction angles and
amplitudes correspond to the fundamental amplitude component of the trans-
mission function for a square wave. However, a glance at the diagram shows
134 Appendices

On-axis
Front Convex Back
input
focal lens focal
laser
plane plane
light

Complex
Phase Fourier transform
and/or of
amplitude object
object

Figure A2.3 A convex lens set-up for obtaining the Fourier transform of an input
object.

that for twice the angle of diffraction for the fundamental component there is
no diffracted wave – square waves only contain odd harmonics. The third
harmonic amplitude has one-third of the amplitude and three times the deflec-
tion (in the sin θ ~ θ approximation) of the first harmonic, and so on. Further-
more, the amplitudes of the odd harmonics vary inversely as the harmonic
number (diffraction order).
Now imagine a convex lens collecting all this diffracted light (Figure A2.3).
The back focal plane of the convex lens (neglecting distortions, aberrations
and so on) will exhibit a series of dots spaced by a distance proportional to the
diffraction angle (strictly speaking, the tangent of the diffraction angle is
the focal length divided by the distance from the point where optical axis
meets the back focal plane, but we are assuming the same sin θd ~ θd approxi-
mation as before). Furthermore the amplitude of these spots in the focal plane
will be proportional to the amplitudes of these spatial frequency components. If
the lens collects all the diffracted light, an accurate image will be produced
wherever the image plane is given by the traditional lens formula. However, if
the lens misses any of this light then the corresponding amplitude components
will be reduced or even eliminated, so the image will have a different spatial
harmonic amplitude distribution from that of the original grating. For example, if
we miss the entirety of the third-order beam and above, we will be back to an
sinusoidal amplitude variation in the image plane at the fundamental spatial
frequency of the grating, since we now have first the average illumination and
the two first-order components to make up the final image.
The final nuance comes if the input object (our input grating) is placed
exactly in the front focal plane of the lens. In this case the focussed light in the
 Appendices 135

back focal plane represents both the amplitude and phase components of the
input amplitude transmittance. The concept of phase components in the amp-
litude transmittance may appear somewhat arbitrary – where is the phase
reference? The simplest approach to this is to take the mean transmittance
(zero deflection) as zero phase and define all the phase components with
respect to that. This is also effectively the standard approach in normal Fourier
transform operations and gives an insight into how an object which has only a
phase transmittance and no change in amplitude can be visualised by changing
the zero-frequency component of the Fourier transform in the back focal plane
of the lens. The reader should examine the Fourier transform expressions for a
phase-modulated carrier and the striking way in which changing the phase or
even removing the fundamental components turns the input phase distribution
into a plane amplitude output distribution.
Figure A2.3 shows a set-up that can be used to perform Fourier transforms
on any arbitrary input object and also indicates how, by manipulating the
transmission properties at the Fourier transform plane, the properties of the
resulting image, formed by the lens after the Fourier transform plane, can also
be modified. This concept has found enormous application in feature enhance-
ment (e.g. in highlighting the edges in an object and thus producing an image
with the lower spatial frequency components attenuated or sometimes
removed) and in various forms of digital image manipulation software. For
the latter, of course, the hardware between the input object’s image and the
final output is not there. However, in imaging systems from microscopes to
telescopes these phenomena are always present and this approach to hardware-
based imaging is often used. Perhaps the most familiar is ‘dark ground
imaging’ in microscopes, which enables the visualisation of low-contrast
objects, including those comprising only phase-delay detail. These phenomena
determine the quality of the final image, how this image quality can vary with
the illuminating wavelength (e.g. through chromatic aberration in lenses or the
fact that blue light is less diffracted than red) and many other features of
imaging systems.

Further Reading

J. W. Goodman, Introduction to Fourier Optics, McGraw Hill, 2017. First published in

1968. Remains the most respected text.
D. Voetz, Computational Fourier Optics, SPIE Tutorial Texts, Vol. TT89, 2011.
G. O. Reynolds et al., Tutorials in Fourier Optics, SPIE Press, Bellingham, WA, 1989.
There is also a free IPad App which performs spatial Fourier transforms on photographs
stored on the device – very instructive!
 Appendix 3
The Clausius–Mossotti Equation and the
Frequency Dependence of the Refractive Index

The key here lies in the concept of polarizability – in other words the
modifying effect of an electric field on the charge distribution in an atom or
molecule. Figure A3.1 shows that an applied field will induce a net dipole
moment on the molecule and that this in turn distorts the field distribution
around the polarised molecule. Consequently, in a dielectric medium the
effective field on each individual molecule is the sum of the applied field
and that due to the induced dipole moments within the medium – here assumed
homogeneous and isotropic.
The dielectric constant or relative permittivity, ԑ, gives a measure of the
polarisation effect. Consider the capacitor shown in Figure A3.1. By
definition,
applied field C1
ε¼ ¼ (A3.1)
nett field C0
where C1 and C0 are respectively the capacitance with and without the dielec-
tric present. The ratio ԑ can be quite large; for silicon, for example, it is nearly
12 owing to the near-cancellation of the applied field within the dielectric.
The Clausius–Mossotti equation relates the dielectric constant of a medium
to the polarisability of its molecules. For a particular frequency within the
absorption spectrum of the molecule, it can be shown that
εr  1 1 X
¼ N j αj (A3.2)
εr þ 2 3ε0 j

where Nj is the volume density of the molecules resonating at a particular

resonant mode j. The symbol αj is the polarizability (i.e. the induced dipole
moment per unit applied field) of the molecule and is a function of frequency.
The summation in the equation is over all resonances pertinent to the fre-
quency of interest at which the value of εr is to be determined.

136
 Appendices 137

Applied + + + - - -
Electric + -
+ Depolarisation -
Field Polarisation Field
+ -
E + P - -
+ + + - -

+ + + + + + + + + + + +
Applied
Electric - - - - - - - - - - - -
Field Induced charge in dielectric
E
+ + + + + + + + + + + +
- - - - - - - - - - - -

Figure A3.1 Illustrating how an applied electric field displaces the charges in a
dielectric and how this produces a depolarising field that partly cancels the applied
field within the dielectric.

The essential aspect of equation A3.2 is that αj can be derived as the

response, at the frequency of interest, of the ‘mass–spring–damper’ molecular
systems. The polarisation of a molecule measures how far an electric field at a
particular frequency stretches the molecular structure, so within αj summed
over all resonances we have the entire behaviour of the material’s dielectric
response to an electromagnetic field – in other words the variation of its
dielectric constant with frequency.
The dielectric constant is related to the refractive index n, the principal param-
eter of interest in photonics, through εr = n2, and the ‘damper’ component gives
rise to a complex solution for the dielectric constant which will vary with
frequency. The size of the ‘damper’ can be estimated through the linewidth of
the spectral absorption line. In the photonic region most of the resonances
concerned are associated with electronic transitions but many also have mixing
effects with molecular vibrations, exemplified in the line spectra of gases, as
illustrated in Figure A3.2. Spectra are also modified through changes in pressure,
introducing more molecular collisions and therefore broadening the spectral line
until individual lines are completely lost, as in the spectra of liquids and gases.
The damper element represents losses in the molecule either due to the
absorption of light and its release as heat or due to collisions and the transfer
of energy among different molecules. This loss term operates analogously to that
in all resonant systems, exemplified in the electronic LRC circuit, and is
reflected in the resonance linewidth. In some cases the loss term can be safely
neglected, for example, dielectric effects in metals above the plasma resonance.
Figure 2.7 shows the situation in a material with free electrons, each of
effective mass me, with electronic charge qe and with free electron density No.
The plasma frequency ωp is given by
138 Appendices

Principal electronic
absorption line
typically near infrared
Relative
Regularly spaced absorption
sidebands due to
molecular resonances,
typically rotational

Wavelength

Figure A3.2 A simplified line structure for absorption in gases around the near
infrared (around 1 to 3 microns wavelength). There is a central peak with
numerous rotational sidebands, typically about 2 to 3 nm apart.

ωp2 = qe2N0/(me ε0)

Here ε0 is the permittivity of free space. This relationship can be deduced by

considering the motion of a solid block of electrons, conveniently in a cube or
sphere, for which the total charge is at the centre of the cube or sphere but is
displaced from the equivalent block of stationary ions (with an equal and
opposite charge at its centre). The block of electrons experiences an electrostatic
restoring force. It then simply equivalent to a mass on a spring. This simple
approach to finding the resonant frequency ωp gives the same answer as a more
general approach which considers the charges to be in random locations!
This account has moved quickly through the basic ideas without lingering
on detailed mathematical proofs, all of which are available in a multitude of
textbooks or on-line scientific sites. On a more general front, what we have
outlined is but one example of the wide ranging insights to be gained from the
simple mass–spring–damper oscillator whether in acoustics, in the waving of
trees in the breeze, in the vibrations of car suspensions or even in atomic and
molecular spectroscopy.

Further Reading

J. H. Hannay ‘The Clausius–Mossotti equation: an alternative derivation’, European

Journal of Physics, Volume 4, No. 3, p. 141, 1993.
C. Kittel, Introduction to Solid State Physics, 8th edition. First published in 1953,
Wiley, 2008.
A. Kumar, Introduction to Solid State Physics, PHI Learning, 2010.
 Appendix 4
The Concepts of Phase and Group Velocity

The ideas of phase and group velocity for a travelling wave have an important role
to play in photonics, and also in any other wave related transmission phenomena.
A whole chapter in Pierce’s splendid text (see the Further Reading section) is
dedicated to the topic. The phase velocity, vp, is the speed with which a wavefront
at a particular frequency travels through a medium. The group velocity, vg, is the
speed with which a wave packet containing a spread of frequencies travels though
the same medium. Implicitly this spread of frequencies is an essential feature if the
wave is to carry energy or information. Another essential feature is that these ideas
are most apparent in material structures rather than in a uniform material space
without dispersion: in such a space, the phase and group velocities are the same.
Remember though that here the term ‘structure’ can include variations in material
properties over the spread of frequencies in the wave packet, so for example when
there are material resonances the differences between phase and group velocity
will manifest themselves. Another important implication is that, whilst the speed
with which energy travels cannot exceed the speed of light in the medium, the
speed with which the wavefronts move can exceed this velocity. Recall that the
refractive index refers to a particular frequency, hence to the phase velocity; all
that’s travelling here is an important abstract concept!
Here, we will briefly explore the meaning of phase and group velocity and
their representation in what has become known as the omega–beta (ω–β)
diagram, an example of which is shown in Figure A4.1. This figure indicates
a structure which does not transmit below the cut-off frequency shown and for
which the phase velocity consistency exceeds the velocity of light c in the
transmitting material, going to infinity at frequencies below the cut off. As a
comparison the dispersion diagram for a very large section of this unstructured
material is also shown. For this baseline material, there is zero dispersion (no
velocity variation with frequency) and so both phase and group velocities are c.
In practice, this dispersion curve will only strictly apply to a vacuum.

139
140 Appendices

w (rad/s)

b = 2p/l (cm-1)

Figure A4.1 A typical ω–β dispersion diagram for a structured system, with that
of a corresponding non-dispersive uniform material shown as a dotted line.

We have seen the parameter ω, often referred to as the angular frequency,

seen many times before – it is 2πf where f is the frequency of the wave. The
parameter β, referred to as the wave number, is less familiar and is defined as
2π/λ. The ratio of these two quantities, ω/β, is often referred to as the wave
velocity. It is more correctly termed the phase velocity, vp, of the wave. This is
the velocity of a specific phase front at the frequency concerned.
The velocity at which energy flows at this frequency is in general different,
as mentioned above. In order to transmit energy, there is by definition an
implied modulation of the basic sinusoidal wave, since the wave train has to be
of finite extent; consequently the wave train is no longer at a single frequency:
there is a finite bandwidth, determined by the modulation function implicit in
the transmission of energy or data. In this case the flow of energy carried by the
total waveform travels at a velocity which is generally different and which is
defined as the slope of the ω–β diagram, dω/dβ.
As an example, we will consider the case of an electromagnetic wave
travelling between two perfectly conducting parallel plates, to illustrate how
the phase and group velocity can be very different (Figure A4.2). Whilst this is
somewhat contrived, it is conceptually close to many waveguiding principles
and provides a useful insight. It is actually a special case of a general rule – the
essential concepts apply to any structure through which a wave can propagate.
The figure shows two interfering broad equal-amplitude parallel light beams
with the wavefronts indicated. The wave front lines are one wavelength apart,
but the solid lines and the dotted lines represent the peaks and troughs of the
wavefront. Consequently, when these wave fronts interfere there will be a zero
in the interference pattern. The condition shown corresponds to a zero at the
 Appendices 141

Travelling
Paths taken in the phase
guide by the front
‘interfering
waves’

Angle Q

Perfect
conductors

Energy flow in
component beams

Figure A4.2 A conceptual diagram of the beam components of a mode in a

parallel plate waveguide surrounded by perfectly conducting walls, and the
origins of the dispersion phenomena therein.

interface with the metallic guide surfaces, with no additional zeroes in the
interference pattern between the walls.
The energy in these component wave fronts travels in the directions indi-
cated and at a velocity c dictated by the medium, which we take as a vacuum.
However, the energy propagating along the waveguide will flow at the com-
ponent of this velocity along the waveguide – in other words at c cos θ. This is
the group velocity along the guide. Note also that the precise angle required in
the interfering beams to produce the necessary patterns will vary with fre-
quency. Hence we shall see group velocity dispersion.
What about the phase fronts travelling along the guide? The points at which
the indicated interfering components cross at the waveguide-to-air- interface
are the travelling wavefronts along the guide. If we take the wavelength in
vacuum along the direction of travel of the component waves as λ then the
distance between these crossing points along the direction of the guide – the
phase wavelength – is λ/cos θ. The phase velocity is then the phase wavelength
times the optical frequency – namely c/cos θ. You might have noticed that the
product of the phase and group velocities in this case is c2. This is a general
rule, with only a very few, rarely encountered, exceptions.

Further Reading

J. R. Pierce, Almost All About Waves, Dover Books, 2006. Reprinted from the original
1974 text.
 Appendix 5
Some Fundamental Constants

Planck’s constant h = 6.62  10‒34 J s.

Energy of a photon = hf, where f is the optical frequency in Hz.
Optical power, P, is related to photon energy through P = Nhf, where N is
the rate of arrival in photons per second.
Sometimes you will see ω = 2πf referred to as the frequency. Technically it
should be the angular frequency, in radians per second.
Boltzmann’s constant kB = 1.38  10–23 J K‒1.
Permittivity of free space ε0 = 8.85  10–12 F m–1.
Permeability of free space μ0 = 4π  10‒7 H m‒1.
Mass of the electron me = 9.1  10–31 kg.
Electronic charge q = |e| = 1.6  10–19 C.
Avogadro’s number N0 = 6.02  1026 atoms per kg mole.
The electron volt, the energy change of 1 electronic charge moving through
1 volt potential, 1 eV 1.6  10–19 J.
Photon energy = hf joules ~ 1.24/λ eV for a wavelength measured in
microns.
Dielectric constant or relative permittivity for silicon, ε (Si) = 11.9.
Velocity of light in a vacuum, c = 3  108 m s–1 = 1/(μ0ε0)0.5.
Room temperature can be taken as approximately T = 300 K.

142
 Appendix 6
Comments and Hints on Chapter Problems

To start with, we give a quick recap on the comments in the Preface. To reach
solutions, there will often be a need to look up sources and to appreciate trade-
offs. Central to solving problems is the idea that you discuss any approach with
others: we can certainly learn from our fellows in addition to our formal
teachers. Often, there is no such thing as a totally right answer, especially to
any real engineering problem. So, collectively, work out the best approach and
go for the best solution you can find. Then, very importantly, critically analyse
this solution. Make sure it stands up to scrutiny! Do not forget, that there is
much source material on the web but, even here, critically assess what you
find. Very occasionally it can be misleading.
Also, exchange your views through the community; there will be web-based
forums from which you can also learn from the experiences of others.

A6.1 Chapter 1
The principal theme of all four of these Chapter 1 problems is the following
question. ‘When do photons really matter in our endeavours to usefully
understand the world around us?’ Associated with this question is the matter
of the sometimes erroneous or ambiguous use of scientific language.

Problem 1
This concerns the ideas of creating and detecting electromagnetic radiation. It
is reasonable to work with the model according to which transmission always
takes place as electromagnetic waves. However, the way to view the creation
and detection processes varies depending on circumstances. The temperature
in deep space can be rapidly determined by a question in your web browser. It

143
144 Appendices

can be safely approximated as 3 K, around 1% of room temperature. Thermal

particle energies, which manifest themselves as thermal noise, are then around
250 μeV. Taking the photon energy as about 1.24/λ, with λ in microns, gives
the wavelength of a photon with this thermal energy photon is around 5 mm
(you can also scale this from Figure 1.2). So photons will come in at about
3 mm and electronics at about 3 cm; again this can be scaled from Figure 1.2.

Problem 2
Microwave photonics is an ambiguous term: it is used occasionally in situ-
ations when microwaves can be viewed as photons (and you may dig around
on the web for suitable examples). More often, though, it refers to the
applications and technologies around modulating an optical signal at micro-
wave frequencies (typically in the 1 to 100 GHz band). Optical communication
applications in deep space are beginning to look promising – so is one photon
modulating another? Is this a combination of electromagnetics and electronics?
This is a fruitful topic for discussion!

Problem 3
This is a straightforward extension of the thought processes above . . .

Problem 4
This highlights the much vaunted wave–particle duality discourse. Look up
X–ray diffraction, then think of the implications – and how hot would the
world need to be for X-rays to be normally viewed as being propagated as
electromagnetic waves in a medium? This estimate also illustrates the prag-
matic approach to waves, particles and electric currents explored in Chapter 2.

A6.2 Chapter 2
For the remaining chapters, we will discuss some of the problems, maybe a
sample or maybe all of them, and then encourage you to explore answers
among yourselves.
In this chapter, perhaps problem 4 is the most pertinent. In problem 4(a) –
deriving Snell’s law – the key here is that the projections of the wavefronts
onto the interface must match on each side and also that the phase velocity of
the incident and refracted rays on each side is determined by the relevant
refractive index. Another way of saying this is that the optical phase velocities
 Appendices 145

are the same on each side of the interface between the two different materials;
as mentioned before this is an important generic observation. Also, when you
sketch the situation, notice that the wavelength components along the interface
exceed those in the input beam itself. If the input is from air or vacuum, then
the phase velocity at the interface exceeds c! This is a useful conceptual tool to
visualise situations when vphase> c.
For problem 4(b), look around and think about everyway phenomena – and
these can be acoustic or mechanical as well as visual.
In Problem 4(c), full derivations for the Fresnel reflection coefficients are
quite involved, but the final results are very definitely of interest. In terms of
amplitude reflection coefficients and transmission coefficients we have, for
waves incident from a region of index n1 entering a region of index n2 via a
planar interface;
ts = 2n1 cos θ1 / (n1 cos θ1 + n2 cos θ2)

rs = (n1 cos θ1 – n2 cos θ2) / (n1 cos θ1 + n2 cos θ2)

for the S polarisation, for which the electric vector is perpendicular to the plane
of incidence. For the P polarisation,
tp = 2n1 cos θ1 / (n2 cos θ1 + n1 cos θ2)

rp = (n2 cos θ1 – n1 cos θ2) / (n2 cos θ1 + n1 cos θ2)

Note the subscripts in the P and S cases for the transmitted and reflected
amplitudes. . .
Other points to note here:
– The phase differences depend on whether the reflection is from high to low
index or vice versa. There is a 180 difference (multiply by –1)
– Check that the sum of the transmitted and reflected power coefficients is
what you would expect it to be – namely unity. Power is proportional to the
square of amplitude.
– You are looking through a glass window, with refractive index 1.5. How
much light will be reflected at normal incidence (our eyes, like almost every
other photodetector, respond to power)? Consider viewing an illuminated
object from different angles?

A6.3 Chapter 3
The problems here all explore what happens when light passes through a
material. Here, the basic possibilities are absorption and scattering, absorbed
146 Appendices

light turning into heat, light at other wavelengths, electronic carrier generation,
refractive index changes. Consider what happens at interfaces, as already
mentioned in Chapter 2. In some places the discussion here is a preamble to
topics discussed in Chapter 4, for example, birefringence.

Problem 1
This concerns the possible ways in which radiation from the sun (assume a
black body at the temperature mentioned in the text) is altered by passing
through the earth’s atmosphere. First, calculate the spectrum as a function of
wavelength. The consider the various phenomena which can modify the
spectrum between its arrival at our atmosphere and its arrival at earth’s surface.
You will need to explore absorption and scattering and also the ways in which
these phenomena vary with weather conditions, where for example both scatter
(from clouds in particular) and atmospheric content (water vapour) can change.
There are also implications regarding altitude – how would things change at
8000 m?
And, as for most of our problems, there is scope for teamwork in this
problems. Again, there is no totally right answer, so a good basic understand-
ing is essential in order to weigh up the different possibilities!

Problem 2
The approach here could be to start by taking 1 mW as the same power level all
along the fibre, for 50 km, and see what comes out of that. Thereafter, examine
the situation where the average loss is a typical value (look this up – or assume,
for a start, that it is 0.5 dB per km). What is the input power needed to average
1 mW over the fibre length? Integrate this distribution over the length to obtain
the total Kerr-induced phase change. Next, examine the same process for, say
1 dB/km and 0.1 dB/km.
Part 2 of the problem is about the mixing processes which occur in non-
linear media and the generation of sum and difference signals. You will need to
think this process through, but the clue lies in the fact that changes induced by
frequency 1 will also impact on the transmission properties of frequency 2. As
for ‘significant’ – you can access the literature on four-wave mixing in optical
communications when two signals very close to each other in wavelength are
transmitted along a fibre, or you can decide on appropriate sounding criteria
and then compare with actual practice. An appropriate starting point might be
that the amplitude of the sum and difference signals is 10% of the separate
 Appendices 147

signals (20 dB down), for example. In cases like this it is always helpful to
initially establish some guidelines to get a feel for the situation of interest.

Problem 3
This is about detecting light and how detection is almost inevitably a function
of optical wavelength. The differences between bandgap (photon) and bolo-
meter (power-into-temperature-change) systems are discussed in the text. You
might like to analyse this further, however, and think about the impact of the
modulation bandwidth on the performance of the two types of system and also
the achievable signal to noise ratios for each type – how small an input power
change could be detected by the bolometer and the (shot noise limited)
bandgap detector? How might these thresholds vary with the optical
incident power?
The final part of this problem, concerning the design of an ‘ideal’ photo-
detector, requires some thought on what is ideal. Yes, it needs to respond over
the wavelength region specified. But over what input power range is it not
specified? Here, making some assumptions seems appropriate. Start with an
input of 1 mW and then consider the design implications for inputs of 1 micro-
watt and 1 watt? Also, we have not stated the operational bandwidth for
modulated optical signal. However, here it is reasonable to assume for a start
that the source is continuous.
Working from these assumptions leads into a starting point based on a small
thermally well insulated bolometer with appropriate thermal conductivity
coefficients (look up some reasonable values) designed into a suitable resist-
ance measurement bridge. In contrast you will have a shot noise limited
bandgap detector with the implications on responsivity previously explored.
Could band gap detectors be stacked to enhance the lower wavelength
responsivity?
There’s much to explore in this problem, and again it is a discussion topic
among a group more than an individual endeavour.

Problem 4
This is about Fresnel reflection from interfaces and optimising the combination
of the reflection from the lens-to-air interface together with the LED-to-glass
interface with a view to getting the maximum power release (remember that we
are usually looking at amplitude reflection (see the comments above on the
Chapter 2 problems).
148 Appendices

The other aspect here is that it is possible to design coatings made from a
material with an intermediate index, which for one single wavelength will
allow though all the light from one medium to another. There is a Wikipedia
article on optical coatings, and you may wish to demonstrate the expression for
the matching-layer thickness mentioned there using the Fresnel relationships.
So, in principle, for a single-wavelength LED, with careful design all the light
could be emitted.
As for the laser issue: coatings are an important factor here, but this time
they are used to reflect, rather than match for transmission. The basic aim with
this question is to encourage an understanding of the lasing process, the
thinking that eventually leads to the laser equation and the concept of recom-
bination rates. Figure A6.1 may provide a useful starting point.

(a) (b) (c)

Spontaneous Smulated
Absorpon emission
Excited emission
state
2 Photons
Photon Photon Photon hn
hn Ground hn hn
state

Figure A6.1 A representation of the three processes possible in photon generation

and absorption. (a) Straightforward absorption of an input photon which results in
thermal generation or sometimes the generation of a lower energy photon via
fluorescence for example. (b) represents spontaneous emission where an excited
state, for example produced by electrical current injection, produces a photon
which is not correlated with other photons produced by the input excitation.
(c) - stimulated emission, the basis of optical amplification and lasing, where
an input photon causes the emission another one of the same wavelength in
an already excited material.

Problem 5
This problem is about the way in which plastics induce stress-related birefrin-
gence between two crossed polarisers, so that a component comes through the
second polariser. In general this component will be coloured, since the
birefringence will change (almost double) in going from long (red) to shorter
wavelengths.
With careful analysis this can be quantified. However, accurate knowledge
of the stress-optic coefficients involved is sparse (but how would they be
used?), and in practice a very rough guide can be derived from the variations
in colour. Blue is more sensitive (less stress needed to introduce this colour)
than red. Why is this the case?
 Appendices 149

Problem 6
Here are some hints on this problem.
– In depletion layer an applied voltage will remove all the free carriers. This
leaves behind the charges on the remaining lattice at a charge density
equivalent to the doping (free-carrier) density times the electronic charge.
This can then be put into Gauss’s law to obtain a field slope. However, for
an intrinsic region it is a reasonable approximation to assume that remaining
charge density is low enough that the slope of the field is zero in this region.
In the p+ and n+ (substrate) regions it is also safe to assume zero field slopes.
This simplification gives the need for a bias voltage of 20 kV/cm times
10 microns to achieve saturated velocity. The calculated bias voltage is
worthy of comment!
– The depletion layer width has to be traversed in a time which is small
compared with any modulation on the optical signal. You may like to
consider why this is the case, and also what might happen if, for example,
the modulation frequency is twice the inverse of the transit time. The carrier
velocity can be taken as 107 cm/s.
– The next issue lies in how far the photons will reach into the depletion
region before the carrier generation rate drops below, say, 10% of the
maximum reached very close to the surface. The p+ contact is 1 micron
thick. Can this thickness be ignored? How significant is this absorption of
photons in the overall quantum efficiency of the device? How do all these
answers change with wavelength at photon energies above the band gap?
The graph at the end of this section shows carrier velocity plots in silicon
and gallium arsenide (see Figure A6.2).

Figure A6.2 Carrier velocity vs Electric Field Characteristics for Si and GaAs.
150 Appendices

– This discussion should have indicated that, while the basic design of a
photodetector is relatively straightforward, the proverbial devil lies in the
detail. In particular, compatibility with external circuit values (bias voltages
and dimensions especially) and operational bandwidth come to the fore,
coupled strongly with the operating wavelength. These considerations form
another topic for discussion!

A6.4 Chapter 4
Problem 1
For a thin lens, perhaps the simplest way to conceive this might be to consider
a stack of thin isosceles-triangle prisms made from the lens glass, and then to
calculate the angle of exit from this prism as a function of distance from the
principal axis of the lens. The angles in the isosceles triangle will be deter-
mined by the slope of the curved surface of the lens at a particular point.
To get the perfect curvature for focussing regardless of the distance from the
axis, i.e. away from the paraxial approximation for a thin lens, all that is needed
is to generalise the paraxial approach used above – but to use it ‘backwards’ in
order to obtain the necessary slope as a function of distance from the axis.
This will not give the exact answer or a prescription for a practical lens, but
it will give the general idea. And it is interesting to discuss the reasons why it
won’t be exact.

Problem 2
This is about a generalised approach to Snell’s law, but in a variety of possible
planes of incidence. This would be a good problem for group discussion, using
a large sheet of paper on which to sketch the various possible situations.

Problem 3
The first part concerns Rayleigh scattering, the route through the atmosphere
and the inverse fourth-power law. But the story for clouds will need you to
explore a somewhat different type of scattering – Mie scattering.

Problem 4
The first part is about the geometry of a parabola and ‘specular’ reflection, for
which the angle of incidence equals the angle of reflection. The second part is
about image receptor sizes, and what to do to compensate for the many sources
 Appendices 151

of turbulence. When one considers the logistics of realising these ideas, things
really become interesting.

Problem 5
This is simply diffraction – with a laser pointer and a piece of cloth. Quantify
the process, measure distances, find the wavelength for the pointer (or make an
informed estimate) and explain the multiple spots which appear in terms of a
Fourier transform which could describe the essential features of the cloth. And
try a different sample – a very sharp edge for example.

Problem 6
This is basically a manipulation to find the Fourier series for A + B cos kz, for
A < B and even for A << B. Thereafter there are hints in the main text.

Problem 7
The first part of this question is about integrating all the possible diffraction
contributors at an ‘input angle = output angle’ condition (see Figure 4.12).
Incidentally, why should these two angles be equal? Does the idea work for
unequal angles? The integration process is somewhat involved though
straightforward.
As for the second part, the answer is that the frequency of the directed beam
is shifted by an amount equal to the ultrasonic frequency in the travelling
wave. The wavelength of the ultrasound in water (sound velocity ~1500m/s)
needs to be small compared with the optical beam width and also small enough
to produce a diffracted beam well separated from the input. The normal
Doppler-shift concepts apply, but they do need to be carefully configured to
fully appreciate what's happening here!

Problem 8
The first part implicitly points to an assumption that the electric fields in the
waveguide are zero at the interface between the high-index (core) and low-index
(cladding) materials. You will then need to look at a fixed guide width and
construct angles at which these conditions can be met using two interfering beams
directed symmetrically about the guide axis. There will be a set of discrete angles
for the emerging beams, and there will also be a threshold at which these angles
152 Appendices

will correspond to incident beams at angles exceeding the critical angle between
core and cladding. Then the cut-off for that waveguide has been reached.
The second part asks you to critically analyse these results and deduce why,
and by roughly how much, these first intuitive estimates need to be modified to
get closer to actuality. This is another topic for exploration and discussion
within your group.

A6.5 Chapter 5
Problem 1
This problem concerns the design for an optical receiver. First you will need to
calculate the shot noise (along the lines of the calculations underlying
Figure 5.1 but for a different wavelength). Then there is the question of how
much shot noise there is in the specified bandwidth. However, is it really
necessary to consider this aspect at this stage? In fact all that is needed is the
minimum load resistance needed for the shot noise to exceed the thermal noise.
This defines the resistance needed for the load, which should be as low as
possible to get the necessary RC, i.e. load resistance times local capacitance
(the latter assumed here to be the photodiode operational capacitance). Is this
value of capacitance, needed for a 1 GHz bandwidth, compatible with a viable
photodiode? If not, then what can be done about it? Lots of scope here for
critical insights and productive discussion.

Problem 2
This requires a combination of appropriate SNR (signal to noise ratio) esti-
mates for shot and thermal noise, but in the context of a photodiode array. You
will also need to make assumptions about the total power from the illuminating
source and the spectral uniformity of this white light source.
The other aspects of this problem include consideration of the aperture of
the lens used to focus the absorption spectrum onto the array and the resulting
‘diffraction-limited blur’ on the spot on each detector. Is it OK to assume that
the lens will be diffraction limited, and if not what would you do about it? The
final point is to highlight the immense potential contributions from the use of
long integration times, reducing the bandwidths to a few Hz or often far less, in
the detection process. Looked at in reverse, there are many atomic processes
which can be best understood using transient (nanosecond and better) obser-
vation. What are the implications here?
 Appendices 153

Problem 3
The whole of this story can be deduced from Figure 5.3, the text around it and
some basics on Fourier transforms: notably, what determines the resolution
and dynamic range in a Fourier transform calculation? The only distraction is
that the resolution in the first part of the question is specified in wavelengths,
whereas the Fourier transform system fundamentally works in optical frequen-
cies. Additionally, the frequency range is in wavelength terms (sometimes you
will see spectroscopic results in terms of ‘wavenumbers’ – the inverse of
wavelengths).

Problem 4
Part (a) of this problem is in principle straightforward – what is the minimum
bandwidth on a 1 micron source required to achieve a coherence length of less
than 5 microns? There is some juggling to be done with the units (time, distance,
frequency, wavelength in microns and bandwidths in nm) used here. There is
also a very important but unstated need here in that the source needs to be
spatially coherent. Another unstated need is depth resolution, but in what
medium? It is probably safe to assume that this medium is air, since all the other
media will have a refractive index greater than or equal to 1. Why is this the case?
The second part of the problem brings together the mechanics and drives, the
mirrors, the data storage and processing for the whole system and the need to
put numerical requirements on them all. Hence explore among your group how
practical it might be to go to 1 micron, or even better, spatial resolution.
The last part of the problem is about illumination wavelengths, the transpar-
ency of the sample and the wavelengths which might characterise artefacts of
interest in a particular sample. There is some interesting and useful exploration
to be done around these issues.

Problem 5
The first part of this question is basically a standard derivation, but the ideas
should be within the grasp of readers without any need for hunting elsewhere.
Think about the basic situation. The light is injected in both directions at once
from a particular point (the beam splitter or directional coupler). It then
propagates around the loop, and one can assume for simplicity that it propa-
gates in air. After a delay in the loop, the light then meets itself coming in the
opposite direction, as it were, but the lengths of the clockwise and counter-
clockwise paths will be slightly different by an amount corresponding to the
154 Appendices

rotation rate times the time delay (multiplied by 2 or not? Think that one
through carefully).
This time delay corresponds to a phase delay and the rest is straightforward.
Interestingly, the result is independent of the refractive index of the medium
through which the light is travelling around the loop.
The second part concerns making reasonable estimates. How long could the
fibre be without too much loss, allowing for consistent bending? How sensitive
could such a precision fibre interferometer be, when designed by a skilled
person?

Problem 6
Straightforward radiation pressure calculations are needed for part (a), and
some careful descriptive discussion around the situation which pertains for part
(b), including the very pertinent query, in which direction does or should
gravity work?

A6.6 Chapter 6
Problem 1
The key to this problem lies in the fact that lithium niobate is electro-optic: its
dielectric constant can be changed by applying an electric field via the surface
electrodes. Therefore, the phase delay in the upper path in the diagram can be
modulated. This is an interferometer and, from the symmetry in the design, it
can be assumed that equal optical power goes along each path. Consequently,
if the two paths emerge via the ‘light out’ path in phase, all the optical power
goes into that port. (The two paths will be in antiphase at the other port, so
nothing exits.) Consequently, electrically changing the phase difference
changes the output. There are more subtleties in this when the output ports
are not in phase.
In part (b) the frequency limits are to do with transit times through the
modulator and what RC time constants are possible. What would be needed for
(i) a phase modulation system and (ii) a frequency modulator? You may also
like to consider a frequency shifter system. Although this ventures into com-
munication system theory, it is relevant since such systems are in use in optical
communications.
Finally, another aspect of this – what might happen if these functions were
implemented in silicon photonics? This leads into the next problem.
 Appendices 155

Problem 2
The answers to this problem (and more background on the foundations of
silicon photonics) can be gleaned from G.T. Reed et al. ‘Silicon optical
modulators’ in Nature Photonics, vol. 4, pp. 518–526, August 2010, and other
articles in the ‘Focus’ special section on silicon photonics in this issue.

Problem 3
The first part of this problem is about phase matching along the interface. You
will need to dig a little to derive an accurate value for the phase velocity of the
wave along the interface. However, some useful insight into how this all works
is gained by making some ‘first guesses’.
Note that, whilst this is rarely stated, the top overlay (often referred to as the
‘analyte’) is often water based and is implicitly of a lower refractive index than
the substrate material. The phase velocity along the interface is consequently
higher than in the bulk material. We also have a mirror in the metallic layer and
this will preferentially reflect the incident wave, unless there is an exact
coupling into this hybrid ‘substrate–metal–analyte-layer’ wave structure (as
indicated, and with an effective index close to the average value between the
two sides of the very thin metal layer). At this point there will be a dip in the
reflected wave when this coupling takes place, corresponding to phase match-
ing between the substrate incident wave and the surface wave defined by the
properties of the interfaces of the analyte, thin metallic and substrate layers.
There’s a nice review (30 pages) of the principles of all this in K. M. Mayer
and J. H. Hafner, ‘Localised Surface Plasmon Resonance Sensors’, Chemical
Reviews 2011 (dx.doi.org/10.1021/cr100313v | Chem. Rev. 2011, 111,
3828–3857).
The second part of this problem on TM or TE polarisation states at the
interface was discussed in the text. As for the ‘needle’ structure in (b) in the
figure: think about the incoming direction of the incident wave, the dimensions
of the ‘needles’ and the operation of a lightning conductor!

Problem 4
The basic feature here is that, by changing the bias on the central electrode, it
is possible to change the delay time for the current flowing through the
graphene by the equivalent of 180 at 10.6 microns, which corresponds to
about 28 THz or a period of about 35 fs. Thus the time-delay change needed
across the ~300 nm device corresponding to this change in transit time is
156 Appendices

equivalent to a change in the electron velocity of ~20  106 m/s! This should
be checked carefully. It gives some insight into the velocity of the carriers
in graphene compared with the velocity of the carriers in silicon (see
Figure A6.2).
You could also work through the sheet resistance figures in Figure 6.12 to
gain some insight into the mobility of electrons as carriers in graphene.
Investigation into the electron density in the graphene sheet is needed in order
to gain this insight. You will need investigate the basic ideas of sheet
resistance.

Problem 5
For this problem, look at the order of magnitude of the dimensions of the
resonator as shown in the diagram. Might the structure have a resonance in the
optical range? The overall device does need to be cooled down to work;
thermally induced vibrations in such a tiny structure could cause considerable
trouble. They would also reduce the effective optical Q factor of the cavity
(why?). Even so, only 0.5 dB of squeezing is produced.
The original idea was published in A.H. Safavi-Naeini et al., Nature,
Vol. 500, p.185, August 2013. There was also a follow up based on silicon
but with a somewhat better performance, this time using very high Q factor
resonators. You can find this at arXiv1309_6371v1 where it is openly
available.
As for the future – well – what do you think?

Problems 6–8
These problems are intended as joint research projects with your colleagues.
They give an opportunity to explore factors which will come into play as
photonics continues to expand both its contributions into society and its ‘tool
kit’ of esoteric devices performing intriguing, in many cases as yet undreamt
of, tasks. There is a fascinating future yet to evolve – we are just at the
beginning!
 Further Reading

A. On Background Tools and Concepts

1. J. R. Pierce, Almost All about Waves, Dover Publications, originally 1974, 200
pp. Remains available.
2. R. Bracewell, The Fourier Transform and its Applications, McGraw Hill, originally
1965, 375 pp. Remains available.
3. J. F. James, A Student’s Guide to Fourier Transforms, Cambridge University Press,
2002, 130 pp.
4. D. Fleisch, A Student’s Guide to Maxwell’s Equations, Cambridge University Press,
2008, 130 pp.
5. E. Collett, Field Guide to Polarization, SPIE Field Guides, Vol. FG05, SPIE, 2005.
6. J. W. Goodman, Introduction to Fourier Optics, first published 1968, 4th edition,
McGraw Hill, 2017, Remains the most respected text.
7. R. Feynman, R. Leighton and M. Sands, The Feynman Lectures on Physics. Three
volumes, 1964, 1966. Library of Congress Catalog Card No. 63-20717
nISBN 0-8053-9045-6 (2006 the definitive edition, 2nd printing, hardcover). Well
presented, well argued, accounts of subtle concepts in optics and photonics (and
most of the rest of essential physics too!).
8. The ‘Optics for Kids’ web site (https://www.optics4kids.org/home) run by the
Optical Society of America (OSA) also has some interesting accounts of many basic
ideas in optics – can be fun to look through!

B. Publications Triggered by the International Year of Light

1. M. Garcia-Matos and L Torner, The Wonders of Light, Cambridge University Press,
2015.
2. K. Arcand and M. Watzke, Light – The Visible Spectrum and Beyond, Black Dog,
2015.
3. Celebrating Light, SPIE Press, 2015.
4. Inspired by Light, SPIE Press, 2016.

157
158 Further Reading

C. More Details on Photonics Principles

1. G. R. Fowles, Introduction to Modern Optics, Dover Books, 1990.
2. B. Saleh and M. Teich, Fundamentals of Photonics, Wiley, 2007, 1175, pp.
3. J. Donnelly and N. Massa, Light – Introduction to Optics and Photonics, New
England Board of Higher Education, Boston, 2007.
4. M. Mansuripur, Classical Optics and its Applications, Cambridge University Press,
2009.
5. E. Hecht, Optics, 5th edition, Pearson Press. This book first appeared in 1974 and
continues as an authoritative and comprehensible guide to wave and ray optics.
6. D. Voetz, Computational Fourier Optics, SPIE Tutorial Texts, Vol. TT89,
7. G. O. Reynolds et al., Tutorials in Fourier Optics, SPIE Press, Bellingham, WA,
1989.
8. There is also a free IPad App which performs spatial Fourier transforms on photo-
graphs stored on the device – very instructive!

D. Some Experiments
1. R. N. Zare, B. H. Spencer and M. P. Jacobson, Laser Experiments for Beginners,
University Science Books, Sausalito, CA, 1995.
2. R. N. Compton and M. A. Duncan, Laser Experiments for Chemistry and Physics,
Oxford University Press, 2016.

E. Some Examples of Online Resources on Photonic

Principles and Applications
1. http://spie.org/?&ID=x10&SSO=1 for free texts including educational CDs and
videos.
2. http://spie.org/publications/fundamentals-of-photonics-modules for photonics prin-
ciples modules.
3. http://spie.org/education/education-outreach-resources/online-resources lists a rich
variety of educational material available free from SPIE.

F. The Troubles About Waves, Rays and Photons

1. F. Sellen, Wave Particle Duality, A compilation from numerous authors totalling
324 pages, originally published in 1992 by Springer; published in paperback, 2013.
2. R. Chandrasekar Causal Physics: Photons by Non-Interactions of Waves, CRC
press, 2014. A single-author perspective targeted towards optical science and
engineering.
 Further Reading 159

G. Some More Advanced Resources

1. A. Cutolo, A. G. Mignani and A. Tajani (eds.), Photonics for Safety and Security,
World Scientific, 2014. An excellent compilation of photonic technologies and
applications.
2. SPIE Tutorial texts in Optical Engineering – a series of over 100 straightforward
tutorial texts in topics from spectroscopy to nanophotonics to plasmonics to optical
design. See also SPIE Spotlights and SPIE Field Guides. These books cover all the
photonic principles and applications and are invariably presented in a concise and
readable manner.
3. J. P. Dowling, Schrödinger’s Killer App: Race to Build the World’s First Quantum
Computer, CRC Press, 2003.
4. O. Solgaard, Photonic Microsystems, Springer, 2009.
5. M. Ohtsu and H. Hori, Near-Field Nano-Optics: From Basic Principles to Nano-
Fabrication and Nano-Photonics (Lasers, Photonics, and Electro-Optics),
Springer, 2012.
6. S. V. Gaponenko, Introduction to Nanophotonics, Cambridge University Press,
2010.
7. L. Novotny and B. Hecht, Principles of Nano-Optics, 2nd edition, Cambridge
University Press, 2012.
8. K. Sakoda and M. Van de Voorde, Micro and Nano Photonic Technologies, Wiley,
2017.
9. E. Diamanti, ‘Quantum signals could soon span the globe’, Nature, Vol. 549, p. 41,
September 2017.
10. P. N. Prasad, Introduction to Biophotonics, Wiley, 2003.
11. G. Keiser, Optical Fiber Communications, 5th edition, Tata McGraw-Hill Educa-
tion 2013.
12. J. Hecht, City of Light, Oxford University Press, 1999.
13. SPIE Tutorial texts in Optical Engineering – a series of over 100 straightforward
tutorial texts in topics from spectroscopy to nanophotonics to plasmonics to optical
design.
14. J. H. Hannay, ‘The Clausius–Mossotti equation: an alternative derivation’, Euro-
pean Journal of Physics, Vol. 4, No. 3, p. 141, 1993.
15. C. Kittel, Introduction to Solid State Physics, now in its 8th edition after first being
published in 1953, Wiley.
16. A. Kumar, Introduction to Solid State Physics, PHI Learning, 2000.

H. Sources of Latest News and Developments

1. Nature Photonics often features special issues and tutorial articles, for example on
plasmonics, graphene, silicon photonics and many other topics.
2. The Professional Societies, most notably SPIE and OSA, disseminate technology
reviews though their member journals, SPIE Professional and Optics and Photonics
News.
160 Further Reading

3. SPIE Newsroom combines scientific and commercial announcements in optics and

photonics.
4. Trade Journals, available on free subscription, such as Photonics Spectra, Laser
Focus World, and Biophotonics, regularly publish easily digested technology
updates.
5. Consult web sites for Photonics 21(Europe) and the National Photonics Initiative
(US), and also download their reports, which present technology and applications
developments important for the future of society and aimed at a general, technology-
literate, audience.
 Index

absorption spectra, 28 dispersion, 14

absorption spectroscopy, 99 dispersion, normal, 69
acousto-optic phase modulator, 52 dispersion in waveguides, 76
adaptive optics, astronomical telescopes,
103 edge enhancement, 68
anisotropic materials, 13 electro-gyration, 53
anti-Stokes wavelengths, 30 electronics, criteria for, 4
atmospheric turbulence, 63 electro-optic effect, 53
avalanche photodiode, 38, 40, 41 energy gap, direct, indirect, 39
energy levels, 18
bandgap, 37 entanglement, concept, 120
birefringence, 13 evanescent field in dielectric optical
birefringence, circular, 53 waveguide, 74
black body, 36
blue sky, 78 far-field region, 64
bolometer, 36 float glass, 64
Boltzmann relationship, 86 fluorescence, 32
Bragg diffraction, 70 fluorescent lamp, 44
Brewster angle, 19 Fourier transform optics, 66
Brillouin scattering, 31 Fourier transform plane, 67, 91
Fourier transform spectroscopy, 89
chiral materials, 58 Fraunhofer region, 65
chromatic aberration (chromatic dispersion), frequency comb, 125
61 Fresnel equation, 19, 71
Clausius–Mossotti equation, 12, 136 Fresnel reflection, 19
coherence, 22 future societal challenges, 126
coherence, spatial, 49
coherence, temporal, 49 gas laser, 51
colloidal gold, 22, 80 global warming, 43
complex refractive index, 16 graphene, 122
concave mirror, parabolic, 62 guide star, 103

depletion region, 42, 57, 149 hollow core fibre, Bragg guiding condition,
depletion length, 38 77
dielectric constant, 12 holographic imaging applications, 92
diffraction gratings, 65 holography, 69

161
162 Index

hybrid silicon lasers, 114 photo-acoustics, 33

hyperspectral imaging, 100 photodetection, 36
photodiode, 37
interference patterns, 21 photomultiplier, 36
interferometry, quadrature off-set, 87 photon energy, 18
invisibility cloak, 78 photonic crystal fibre waveguide, 76
photonic materials and structures, conceptual
Kerr effect, 29 building blocks, 120
photonics, criteria for, 4
laser cutting, 98 photons and waves, 19
laser generated ultrasound, 34 photostriction, 42
laser induced breakdown spectroscopy, 35 Planck’s law, 43
laser machining, 34 plasma resonance, 17
laser surgery, 99 plasmonics, 17, 80
lens resolution, criteria for, 61 Pockels effect, 53
lens, aspherical, 62 polarisation states, 13
lens, spherical convex, 60 precision mechanical
light emitting diode, 45 coordinate measuring machine, 94
light, wave or particle, 5 principal axes, 14
LIGO use of squeezed light, 88
lumens per watt, 48 quantum communications, 121
quantum dot, 81
mass–spring–damper oscillator, 12 quantum mechanical energy levels for quantum
metamaterials, 78, 117 dot, 81
metastable energy state, 30
microscopy, confocal, 110 radiation pressure, 105
microscopy, fluorescence, 112 Raman scattering, 31
microscopy, near field scanning, 111 ray optics, 11
molecular polarizability, 136 Rayleigh scattering, 76
molecular resonance, 51 recombination of carriers, 38
refractive index, 12
nanophotonic circuits, 116 resolution, spatial filtering of images, 66
nanophotonic resonator, 118 resonant structures, 9
nanophotonics, 78 responsitivity, photodiode, 37
negative-index metamaterials, 78 retro-reflecting mirrors and lenses, 60
Newton’s rings, 95
noise vector, phase noise, amplitude noise, Sagnac interferometer, 95
87 semiconductor laser, 49
shot noise limit, 87
oil on water, thin films, 71 signal to noise ratio for optical systems, 85
optical coherence tomography, 92 silica, losses and dispersion, 74
optical fibre, parabolic index, 97 silicon photonics, 113
optical tweezers, 105 silicon waveguides, 114
optical waveguide, 73 single-mode fibre dispersion compensated
optically flat surface, 62 W profile, 97
opticqal fibre, single-mode, 74 single-photon avalanche detector (SPAD),
overmoded single-mode waveguide, 75 121
single-photon sources, 120
phase contrast, 68 skin depth, 28
phase plate, 59 solar cell, 42
phonons, 31 spatial filter, 77
 Index 163

spatial resolution, 134 three-dimensional imaging, photographic, 93

spectra, 26 transparent objects, imaging through spatial
spectral nephelometry, 100 filter, 68
spectroscopy, 27
squeezed light, photonic circuit, 124 virtual energy state, 30
squeezed light noise reduction, 87 visual sensitivity, 47
Stokes wavelengths, 30 volume hologram, 92
stress-optic coefficient, 52
subwavelength structure, 22 wave equation, 16
surface plasmons, Kretchmann configuration, waveguide dispersion, chromatic, 74
116 waveguide dispersion, structural, 73
wavelength resolution of spectrometer, 89
thermal diffusion, 34 Wien displacement relationship, 43
thin film optics, 72
thin lens paraxial approximation, 61 Young’s slits, 21