Nanophotonics and Plasmonics: An

Integrated View 1st Edition Ching Eng

(Jason) Png
Visit to download the full and correct content document:
https://textbookfull.com/product/nanophotonics-and-plasmonics-an-integrated-view-1s
t-edition-ching-eng-jason-png/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Diffractive Optics and Nanophotonics 1st Edition So■fer

https://textbookfull.com/product/diffractive-optics-and-
nanophotonics-1st-edition-soifer/

Plant systematics : an integrated approach Singh

https://textbookfull.com/product/plant-systematics-an-integrated-
approach-singh/

Optical MEMS, Nanophotonics, and Their Applications 1st

Edition Guangya Zhou

https://textbookfull.com/product/optical-mems-nanophotonics-and-
their-applications-1st-edition-guangya-zhou/

Drug Delivery: An Integrated Clinical and Engineering

Approach 1st Edition Yitzhak Rosen

https://textbookfull.com/product/drug-delivery-an-integrated-
clinical-and-engineering-approach-1st-edition-yitzhak-rosen/
Quantum Field Theory An Integrated Approach 1st Edition
Eduardo Fradkin

https://textbookfull.com/product/quantum-field-theory-an-
integrated-approach-1st-edition-eduardo-fradkin/

Applied Nanophotonics Sergey V Gaponenko

https://textbookfull.com/product/applied-nanophotonics-sergey-v-
gaponenko/

From Classical to Quantum Plasmonics in Three and Two

Dimensions 1st Edition Thomas Christensen (Auth.)

https://textbookfull.com/product/from-classical-to-quantum-
plasmonics-in-three-and-two-dimensions-1st-edition-thomas-
christensen-auth/

Nanophotonics in Biomedical Engineering Xiangwei Zhao

https://textbookfull.com/product/nanophotonics-in-biomedical-
engineering-xiangwei-zhao/

Coastal Wetlands: An Integrated Ecosystem Approach

Gerardo Perillo

https://textbookfull.com/product/coastal-wetlands-an-integrated-
ecosystem-approach-gerardo-perillo/
Nanophotonics
and Plasmonics
An Integrated View
SERIES IN OPTICS AND OPTOELECTRONICS

Series Editors: E Roy Pike, Kings College, London, UK

Robert G W Brown, University of California, Irvine, USA

Recent titles in the series

Handbook of GaN Semiconductor Materials and Devices
Wengang (Wayne) Bi, Hao-chung (Henry) Kuo, Pei-Cheng Ku,
and Bo Shen Piprek (Eds.)
Handbook of Optoelectronic Device Modeling and Simulation:
Fundamentals, Materials, Nanostructures, LEDs, and Amplifiers
– Volume One
Joachim Piprek (Ed.)
Handbook of Optoelectronic Device Modeling and Simulation:
Lasers, Modulators, Photodetectors, Solar Cells, and Numerical Methods
– Volume Two
Joachim Piprek (Ed.)
Nanophotonics and Plasmonics: An Integrated View
Dr. Ching Eng (Jason) Png and Dr. Yuriy Akimov
Handbook of Solid-State Lighting and LEDs
Zhe Chuan Feng (Ed.)
Optical Microring Resonators: Theory, Techniques, and Applications
V. Van
Optical Compressive Imaging
Adrian Stern
Singular Optics
Gregory J. Gbur
The Limits of Resolution
Geoffrey de Villiers and E. Roy Pike
Polarized Light and the Mueller Matrix Approach
José J Gil and Razvigor Ossikovski
Light—The Physics of the Photon
Ole Keller
Advanced Biophotonics: Tissue Optical Sectioning
Ruikang K Wang and Valery V Tuchin (Eds.)
Handbook of Silicon Photonics
Laurent Vivien and Lorenzo Pavesi (Eds.)
Microlenses: Properties, Fabrication and Liquid Lenses
Hongrui Jiang and Xuefeng Zeng
Nanophotonics
and Plasmonics
An Integrated View

Edited by
Ching Eng Png
Yuriy Akimov
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2017 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed on acid-free paper

Version Date: 20170713

International Standard Book Number-13: 978-1-4987-5867-3 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access
www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization
that provides licenses and registration for a variety of users. For organizations that have been granted
a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
 Contents

Preface xi

I Fundamentals 1
1 Electromagnetic fields in uniform media 3
by Yuriy Akimov and Pavlo Rutkevych
1.1 Electromagnetic field equations 3
1.1.1 Maxwell’s equations in medium 3
1.1.2 Material equations 6
1.1.3 Temporal and spatial dispersion 7
1.2 Local response approximation 9
1.2.1 Energy of electromagnetic field in medium 9
1.2.2 Properties of complex dielectric permittivity 10
1.2.3 Medium response modeling 12
1.3 Electromagnetic fields in medium 17
1.3.1 Electromagnetic field generation 17
1.3.2 Bulk eigenwaves 18
1.3.3 Quasi-particle classification 18
1.4 Summary 22
Bibliography 22

2 Electromagnetic waves in bounded media 25

by Yuriy Akimov
2.1 Electromagnetic fields in bounded media 25
2.1.1 Material relations for inhomogeneous media 25
2.1.2 Electromagnetic fields in inhomogeneous media 26
2.1.3 Piecewise homogeneous media 28
2.2 Boundary effects 29
2.2.1 Boundary conditions 29
2.2.2 Bulk eigenfields in real space 29
2.2.3 Eigenfields of piecewise homogeneous media 32
2.3 Polaritons of bounded media 34
2.3.1 Hybridization of bulk polaritons 34

v
vi
2.3.2 Polaritons of interface 35
2.3.3 Polaritons of slab 39
2.4 Summary 45
Bibliography 45

3 Localized polaritons of single-particle systems 47

by Yuriy Akimov
3.1 Localized polaritons 47
3.1.1 Eigenfield quantization 47
3.1.2 Wavefunctions of polariton eigenfields 50
3.2 Localized polaritons of planar systems 51
3.2.1 Planar eigenfields 51
3.2.2 Planar eigenoscillations 52
3.3 Localized polaritons of cylindrical systems 57
3.3.1 Cylindrical eigenfields 57
3.3.2 Cylindrical eigenoscillations 58
3.4 Localized polaritons of spherical systems 61
3.4.1 Spherical eigenfields 62
3.4.2 Spherical eigenoscillations 63
3.5 Summary 66
Bibliography 66

4 Localized polaritons of multi-particle systems 69

by Lin Wu, Valerian Hongjie Chen, Ping Bai, and Song Sun
4.1 Inter-particle polariton hybridization 69
4.1.1 Capacitive coupling: eigenmode hybridization theory 69
4.1.2 Conductive coupling: charge-transfer polaritons 71
4.1.3 Link between capacitive and conductive coupling 73
4.2 Polariton hybridization effects 77
4.2.1 Electric and magnetic responses 78
4.2.2 Directional radiation 82
4.2.3 Fano resonances 83
4.2.4 Chirality resonances 85
4.3 Polariton hybridization in complex systems 86
4.3.1 Image-coupled nanoparticle-on-mirror systems 86
4.3.2 Metal-dielectric hybrid systems 88
4.3.3 Periodically ordered particles 90
4.4 Summary 91
Bibliography 92

II Applications of localized eigenmodes 101

5 Nanostructural coloration 103
 vii
by Ravi S. Hegde
5.1 Introduction and background 103
5.1.1 Quantification of color 105
5.2 Coloration by nanostructures 106
5.2.1 Structural color in nanoparticle arrays 107
5.2.2 Structural color in nanoaperture arrays 113
5.2.3 Split-complementary nanostructured reflective color filters 115
5.3 Emerging materials for structural coloration 122
5.4 Summary 125
Bibliography 126

6 Nanostructure-enhanced fluorescence emission 133

by Song Sun, Lin Wu, and Ping Bai
6.1 Introduction and background 133
6.1.1 Application of fluorescent emitters 133
6.1.2 Types of fluorescent emitters 134
6.1.3 Enhancement of fluorescence with nanostructures 135
6.2 Fluorescence emission mechanism 137
6.2.1 Classical description 137
6.2.2 Two-photon absorption mechanism 138
6.3 Enhancement with metal nanostructures 140
6.3.1 Metal nanoparticles 140
6.3.2 Metal thin films 140
6.3.3 Modified emission directivity 142
6.4 Enhancement with dielectric nanostructures 143
6.4.1 Photonic crystal microcavities 144
6.4.2 Dielectric nanoantennas 144
6.4.3 Metal-dielectric hybrid structures 145
6.5 Summary 146
Bibliography 147

7 Chiral optics 151

by Eng Huat Khoo, Wee Kee Phua, Yew Li Hor, and Yan Jun Liu
7.1 Introduction and background 151
7.1.1 History of chiroptics 151
7.1.2 Natural optical activity 152
7.1.3 Chiroptical effects 154
7.2 Flat chiral nanostructures 155
7.2.1 Single-layer chiral systems 155
7.2.2 Multi-layer chiral systems 161
7.3 Biosensing with flat chiral systems 167
7.3.1 Sensing of G-actin 167
7.3.2 Sensing of F-actin 167
viii
7.3.3 Effects of superchiral fields 169
7.4 Summary 171
Bibliography 171

8 Localized polariton-based sensors 175

by Ping Bai, Xiaodong Zhou, Ten It Wong, Lin Wu, and Song Sun
8.1 Operation principles 175
8.2 Sensing structures 176
8.2.1 Nanoparticles 176
8.2.2 Periodic nanostructures 178
8.3 Light illumination effects 183
8.3.1 Front and rear illumination 184
8.3.2 Oblique illumination 186
8.4 Effects of nanostructure materials 187
8.5 Nanochip fabrication and characterization 189
8.6 Point-of-care sensing systems 193
8.6.1 System configuration 194
8.6.2 System characterization 195
8.7 Summary 197
Bibliography 197

9 Metasurfaces for flat optics 199

by Zhengtong Liu
9.1 Introduction and background 199
9.1.1 History of metasurface development 199
9.1.2 Generalized Snell’s law 200
9.2 Metasurface devices 201
9.2.1 Metasurfaces using rods as meta-atoms 202
9.2.2 Metasurfaces using V-shaped meta-atoms 204
9.2.3 Metasurfaces with other meta-atom shapes 207
9.2.4 Material selections for metasurfaces 208
9.3 Summary 209
Bibliography 211

III Applications of propagating eigenmodes 215

10 Guiding light with resonant nanoparticles 219
by Hong-Son Chu and Thomas Y.L. Ang
10.1 Introduction and background 219
10.1.1 Guiding light with coupled nanoparticles 220
10.2 Design considerations for on-chip integration 224
10.3 Nanocoupler for chain waveguide 226
 ix
10.3.1 Direct coupler 226
10.3.2 Tapered coupler 227
10.4 Nanoparticle bend chain 229
10.5 Summary 231
Bibliography 231

11 Sub-wavelength slot waveguides 235

by Hong-Son Chu
11.1 Introduction 235
11.2 Overview of sub-wavelength waveguides 236
11.3 Metal-nanoparticle double-chain waveguide 240
11.3.1 Waveguide specifications 240
11.3.2 Operation characteristics 240
11.4 Sub-wavelength slab-slot waveguides 245
11.4.1 Waveguide specifications 245
11.4.2 Operation characteristics 247
11.5 Summary 252
Bibliography 252

12 Photodetectors 257
by Ching Eng Png, Song Sun, and Ping Bai
12.1 Introduction and background 257
12.2 Semiconductor-based photodetectors 258
12.2.1 Inx Ga1−x As photodetectors 258
12.2.2 Ge-on-Si photodetectors 259
12.2.3 All-Si photodetectors 261
12.3 Photodetectors based on low-dimensional materials 262
12.3.1 Graphene-based photodetectors 262
12.3.2 Carbon nanotube-based photodetectors 263
12.4 Metal-based photodetectors 265
12.4.1 Electrode surface polaritons 265
12.4.2 Metal antenna-based photodetectors 267
12.4.3 Photodetectors without semiconductors 268
12.5 Future outlook 269
Bibliography 270

13 Integrated nonlinear photonics 275

by Jun Rong Ong
13.1 Introduction and background 275
13.1.1 Physical basis 276
13.1.2 Material platforms and properties 279
13.2 Integrated nonlinear silicon photonics 281
13.2.1 Overview and challenges 281
x
13.2.2 Application of nonlinearities 284
13.2.3 Engineering waveguide structures for nonlinear photonics 286
13.3 Integrated nonlinear quantum photonics 289
13.3.1 Photon pair generation via nonlinear silicon photonics 290
13.3.2 Experiments using entangled photon pairs 291
13.3.3 Quantum frequency conversion 291
13.4 Future outlook 292
Bibliography 293

14 Integrated nanophotonics for multi-user quantum key distribution

networks 305
by Han Chuen Lim and Mao Tong Liu
14.1 Introduction and background 305
14.1.1 Significance of QKD 306
14.1.2 Challenges 307
14.2 Multi-user QKD network 308
14.2.1 Backbone QKD links 308
14.2.2 QKD access networks 310
14.3 Wavelength-multiplexed entanglement-based QKD 312
14.3.1 Prepare-and-measure QKD 312
14.3.2 Entanglement-based QKD 312
14.3.3 Entanglement distribution in a network 313
14.3.4 Wavelength-multiplexed entanglement distribution 315
14.4 Integrated nanophotonics for QKD applications 317
14.4.1 Ideal single-photon source and weak coherent source 317
14.4.2 On-chip entangled photon generation 317
14.4.3 On-chip photon detection 324
14.4.4 On-chip photon wavelength demultiplexing 325
14.4.5 Toward fully monolithic integration 326
14.5 Future outlook 329
Bibliography 330

Index 345
 Preface

With recent developments of nanotechnologies, the photonics community found a

powerful incentive that eventually led to the formation of two new fields — nanopho-
tonics and plasmonics — studying nanostructured materials for manipulation and
control of light at nanometer scale. Started from exploration of electromagnetic
waves supported by nanosized waveguides and then moving toward the use of local-
ized eigenmodes, nanophotonics and plasmonics created a number of opportunities
in the applied sciences and even made their impact in the market.
Nanophotonics and plasmonics, being closely related fields, have common
physics, targets, and methodologies, but deal with different materials. Nanopho-
tonics is mainly focused on optically transparent materials such as dielectrics and
semiconductors, while optically opaque materials such as metals are the choices
of plasmonics. Despite strong similarity in physics of these phenomena, the two
fields were developed within different communities in slightly different ways. While
nanophotonics focused on propagating eigenwaves for integrated optics and optical
communication with a little exposure to nanoscale resonators, plasmonics studied
both propagating and localized eigenmodes for a broader range of applications. The
fact that the potential of localized eigenmodes is restricted in plasmonics by their
high ohmic losses has turned the research toward low-loss high-index dielectric res-
onators, where localized photonic eigenmodes readily demonstrated their unique ca-
pabilities to control light-matter interaction. As such, both technologies complete
their full cycles for exploration of propagating and localized eigenmodes and meet
together. To further utilize their full potentials for synergy of the two fields, the de-
velopment of a unified platform covering both areas is required.
The goal of this book is to provide a unified description of nanophotonics and
plasmonics, to highlight their similarities and advantages, and finally to use their syn-
ergy for further progress in control of light at nanometer scale. This book overviews
the state-of-the-art developments in this field from an electrodynamics point of view
enabling the joint description of both technologies. The central line of the book is
analysis of nanophotonics and plasmonics based on the use of the general theory
of electromagnetic eigenmodes. It provides the universal tools for in-depth under-
standing of localized and propagating eigenmodes in both technologies for further
implementation in various applications ranging from chiral and integrated optics to
coloration and biosensing.
The book is divided into three parts. Part I presents unified fundamentals of
nanophotonics and plasmonics. Chapter 1 introduces the theory of electromagnetic
fields in uniform media. Chapter 2 provides an introduction to electromagnetic waves

xi
xii PREFACE
in bounded media. Chapter 3 presents the theory of localized eigenmodes in sim-
ple single-particle systems. Chapter 4 discusses the hybridization of localized eigen-
modes in complex multi-particle systems.
Part II overviews numerous applications based on the use of localized eigen-
modes. Chapter 5 gives an introduction to structural coloration. Chapter 6 explores
eigenmodes for fluorescence enhancement. Chapter 7 provides an overview of the
developments in chiral optics. Chapter 8 introduces biosensing on localized eigen-
modes. Chapter 9 discusses the use of optical metasurfaces made of resonant anten-
nas for efficient light control.
Part III reviews various applications based on electromagnetic waves supported
by metallic, dielectric, and semiconductor waveguides. Chapter 10 introduces light
guiding with nanoparticle chains. Chapter 11 overviews subwavelength slot waveg-
uides. Chapter 12 discusses photodetectors. Chapter 13 introduces integrated nonlin-
ear photonics. Chapter 14 is on the use of integrated nanophotonics for multi-user
quantum key distribution networks.
We gratefully acknowledge the support from the Agency for Science, Technol-
ogy, and Research (A*STAR) of Singapore and the Institute of High Performance
Computing (IHPC). We also acknowledge all the contributors, Drs. Pavlo Rutkevych,
Lin Wu, Ping Bai, Song Sun, Ravi Hegde, Eng Huat Khoo, Wee Kee Phua, Yew Li
Hor, Yan Jun Liu, Xiaodong Zhou, Ten It Wong, Zhengtong Liu, Hong-Son Chu,
Thomas Ang, Jun Rong Ong, Han Chuen Lim, Mao Tong Liu, and Mr. Valerian
Chen, without whose significant work this book would never have been published.
We hope this book will serve as a basis for future progress in the field and will be
a valuable reference for engineers, researchers, and students in the areas of nanopho-
tonics, plasmonics, and nano-optics in general.

Yuriy Akimov
Ching Eng Png
 Part I

Fundamentals

1
 Chapter 1

Electromagnetic fields in uniform media

Yuriy Akimov
Institute of High Performance Computing, Singapore
Pavlo Rutkevych
Institute of High Performance Computing, Singapore

In this chapter, we will give a brief introduction to the classical electrodynamics

that constitutes the basis of modern nanophotonics and plasmonics. Section 1.1 will
consider the theory of electromagnetic fields in continuous media. In Section 1.2,
we will give an optical description of non-magnetic solids within the local response
approximation. Section 1.3 will review the theory of electromagnetic eigenwaves in
solids. The chapter will conclude with a summary in Section 1.4.

1.1 Electromagnetic field equations

1.1.1 Maxwell’s equations in medium
In the classical theory, the concept of electromagnetic field is introduced to describe
how charges interact with each other [1]. Every charge is assumed to generate elec-
tromagnetic field, via which it interacts with other charges. In the case of multi-
ple charges, the overall field is given by the sum of corresponding fields generated
by every charge, according to the superposition principle. As a result, full electro-
magnetic description of a medium requires consideration of all microscopic charges
comprising the medium, which is a complicated and troublesome procedure, as the
number of such charges in solids is ∼ 1030 per 1 m3 . To simplify the description,
the macroscopic approach is introduced [2], where all physical quantities including
fields and charges are statistically averaged in space over small volumes, the lin-
ear size of which, amacro , is much smaller compared to the wavelength, but much
larger than the lattice constant. In other words, we neglect all fluctuations appearing
at atomic scales, below amacro , and describe the medium from the macroscopic point
of view, where charges and currents are continuously distributed in space. Within
this approach, the generation of electromagnetic fields by charges is given by the

3
4 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA
macroscopic Maxwell’s equations [3, 4],
 
∂E
∇ × B = µ0 ε0 +j , (1.1)
∂t
∂B
∇×E = − , (1.2)
∂t
where ε0 = 8.85418782 · 10−12 F/m and µ0 = 4π · 10−7 H/m are the electric and
magnetic constants; j is the macroscopic current density of all charges in the medium;
and E and B are the macroscopic electric field and magnetic induction, respectively.
Usually, Maxwell’s equations are accompanied by two divergence relations,
ρ
∇·E = , (1.3)
ε0
∇ · B = 0, (1.4)

with ρ as the macroscopic charge density. These relations are used intensely in clas-
sical electrodynamics. Often, they are called the second pair of Maxwell’s equations.
However, their importance is slightly overestimated. In fact, they are derivatives of
Eqs. (1.1) and (1.2) and do not convey additional information. Therefore, they should
be considered as auxiliary and solved together with the first pair of Maxwell’s equa-
tions all the time. Otherwise, their separate solution can lead to substantial errors for
time-varying fields.1
In Eqs. (1.1)–(1.4), the charge and current densities play the role of sources for
the fields E and B. In general, charge and current densities are related by the charge
conservation law that requires
∂ρ
+ ∇ · j = 0, (1.5)
∂t
where both ρ and j should be considered total quantities, i.e., composed by both
intrinsic and extrinsic charges. As the intrinsic charges are generally represented by
positive and negative ones, it is convenient to split the total charge density ρ into two
parts:
ρ = ρpol + ρext ,
where sign-varying ρpol (r) describes fully compensated charges called polarization
charges and featuring zero total charge
Z
ρpol dV = 0,

and the second part, ρext (r), represents uncompensated charges called external
charges and exhibiting Z
ρext dV 6= 0.

1 Auxiliary relations (1.3) and (1.4) can be derived by taking divergence of the first pair of Maxwell’s

equations. Therefore, they are one order higher than original Eqs. (1.1) and (1.2). As a result, they feature
extra solutions, which are unphysical and must be excluded from consideration.
ELECTROMAGNETIC FIELD EQUATIONS 5
Both polarization and external charges are required to obey the individual continuity
equations similar to Eq. (1.5). Following it, we can similarly decompose the total
current density j,
j = jpol + jext .
Of course, such separation of polarization and external charges is relative and am-
biguous, as it can be done in different ways if the total charge is uncompensated. But
it allows us to represent the overall system as a uniform charge-compensated medium
with added external charges. In this description, the external charges and currents
introduce initial electromagnetic disturbance to the initially charge-neutral medium
with ρpol = 0, which responds with separation of its charges and ρpol 6= 0. The con-
cept of polarization charges is the fundamental idea of the classical electrodynamics
that enables the elegant and unifying description of electromagnetic response of any
charge-neutral medium regardless of its type and types of the charges composing it.
Following the Maxwell’s equation (1.2), the density of polarization charges can
be represented as the divergence of an arbitrary vector, −P,
ρpol = −∇ · P. (1.6)
The condition of full compensation of the polarization charges requires P to be non-
zero inside the medium only and to vanish outside of it. Thus, vector P describes
the polarization of the initially neutral medium and is commonly called the polar-
ization field. At the same time, the continuity equation (1.5) gives us a form for the
polarization current density,
∂P
jpol = + ∇ × M, (1.7)
∂t
where M is another arbitrary vector that describes magnetic induction of the polar-
ization current not accompanied by charge polarization. It is commonly called the
magnetization field. Introduction of the polarization and magnetization fields allows
us to rewrite the macroscopic Maxwell’s equations in a more compact form,
∂D
∇×H = + jext , (1.8)
∂t
∂B
∇×E = − , (1.9)
∂t
with the divergence relations given by
∇ · D = ρext , (1.10)
∇ · B = 0, (1.11)
where D and H are the auxiliary fields called the electric displacement and magnetic
field, introduced to account for the polarization and magnetization of the charge-
compensated medium,
D = ε0 E + P, (1.12)
1
H = B − M. (1.13)
µ0
6 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA
In this form, the Maxwell’s equations describe generation of electromagnetic field
by external currents in terms of two pairs of the electric {E, D} and magnetic {B, H}
fields. However, they do not form a closed set of equations, until we provide mate-
rial relations for the response of the charge-neutral medium to electric and magnetic
fields. In general, these relations are given by field-dependent functions for the po-
larization P = P(E) and magnetization M = M(B) vectors that eventually result in
material relations for the auxiliary fields D = D(E) and H = H(B).

1.1.2 Material equations

Establishing the relations for D(E) and H(B) takes the key place in classical elec-
trodynamics, as it describes the charge-neutral media response to electromagnetic
fields. In our considerations, we will focus on the high-frequency response in the
optical range and above, where most of solids lose their magnetic properties [2],
featuring
1
M = 0, H = B. (1.14)
µ0
Below, we will assume that condition (1.14) holds all the time and consider only mag-
netically inactive materials. The response of such materials is given by dependence
D(E) only, which is generally nonlinear. However, in most interesting cases the elec-
tric fields E are low enough, so the auxiliary field D(E) can be treated as a linear
function. It is the so-called linear electrodynamics approach, where the medium’s
polarization current jpol at a given point r and moment t is assumed to be a linear
function J(E), defined as a nonlocal response to electromagnetic fields taken at any
point of space r0 at all preceding moments t 0 < t in accordance with the causality
principle,
Zt Z
Ji (t, r) = dt 0 dr0 σi j (t,t 0 , r, r0 )E j (t 0 , r0 ). (1.15)
−∞

The tensor σi j (t,t 0 , r, r0 ) in Eqs. (1.15) characterizes transfer of the material’s re-
sponse from one point of space and time to another. For a homogeneous medium
(guaranteed by macroscopic averaging over amacro larger than the lattice constant),
the tensor σi j depends on the differences t − t 0 and r − r0 only,

σi j (t,t 0 , r, r0 ) = σi j (t − t 0 , r − r0 ). (1.16)

Now, if we perform the Fourier transform,

1
Z Z
G(t, r) = G(ω, k) ei(k·r−ωt) dω dk,
(2π)2

of J and E in the (t, r) space, we can get their relation in the frequency-wavevector
space (ω, k),
Ji (ω, k) = σi j (ω, k)E j (ω, k), (1.17)
ELECTROMAGNETIC FIELD EQUATIONS 7
where σi j (ω, k) is the tensor of complex conductivity given by
Z∞ Z
σi j (ω, k) = dτ dR σi j (τ, R) e−i(k·R−ωτ) , (1.18)
0

with τ = t − t 0 and R = r − r0 .
Now, we can use the relation J(E) in the (ω, k) space to get the material equation
for D(E) in the linear approximation,

Di (ω, k) = ε0 εi j (ω, k)E j (ω, k). (1.19)

Here, εi j (ω, k) is the tensor of complex permittivity defined as

σi j (ω, k)
εi j (ω, k) = 1 + i . (1.20)
ε0 ω
Following this definition, εi j (ω, k) characterizes the linear nonlocal response of a
charge-neutral medium to electric fields.
For an isotropic medium, properties of which are identical in any direction,
εi j (ω, k) can be composed of the unit tensor δi j and the tensor ki k j , as they are the
only two tensors of the second rank formed of the wavevector k. Thus, we can write
 
ki k j ki k j
εi j (ω, k) = δi j − 2 εt (ω, k) + 2 εl (ω, k). (1.21)
k k
Following this expression, among nine components of the tensor εi j , there are only
two independent components, εt (ω, k) and εl (ω, k). According to Eq. (1.20), these
components have clear physical meaning: εl (ω, k) gives the medium response to lon-
gitudinal electric fields (E × k = 0), while εt (ω, k) describes the response to trans-
verse electric fields (E · k = 0).

1.1.3 Temporal and spatial dispersion

In general case, tensor εi j depends on the frequency ω and the wavevector k. Even-
tually, any electromagnetic pulse disperses by propagating in the medium, as the
Fourier components
G(ω, k) ei(k·r−ωt)
with different ω and k (that comprise the pulse in accordance with the Fourier trans-
form) propagate with different phase velocities ω/k. Thus, the materials with fre-
quency and wavevector dependence are dispersive. The frequency dependence of
the tensor εi j describes the temporal dispersion of electromagnetic fields, while the
wavevector dependence gives the spatial dispersion.
Temporal dispersion of solids arises due to the inertia and friction of intrinsic
charges that make the polarization inertial to electric field. Thus, the medium re-
sponse at a given moment t depends on the electric field values at all preceding
moments t 0 ≤ t.
8 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA
The time interval τ = t − t 0 , over which the previous history still has a significant
effect, is defined by the characteristic frequencies ωs . It is obvious that for electro-
magnetic fields oscillating at a very high frequency ω  ωs , the intrinsic charges do
not have enough time to form any significant polarization. Eventually, the result is
very weak temporal dispersion with
εi j (ω → ∞) = δi j .
However, at frequencies ω below or close to the characteristic frequencies ωs , the
temporal dispersion increases and cannot be ignored anymore.
In contrast to the temporal dispersion, the spatial one comes from nonlocality of
the medium response to electric fields. Physically, it is given by the dependence of
the polarization vector P(t, r) on the electric fields E(t, r0 ) in the vicinity of the point
r. The region over which the nonlocality takes place is defined by the characteristic
length as = amacro + adyn given by the field averaging and charge dynamics.
The first nonlocal mechanism is purely artificial and introduced through statisti-
cal averaging of all fields in space over amacro . It should be considered the inherent
error of the macroscopic approach that does not allow us to look at smaller scales, but
is necessary to fulfill the condition of the medium’s uniformity. The second mech-
anism is physical, caused by the dynamics of charges in the medium. At high fre-
quencies, it is given by the traveling distance of the electrons at the highest occupied
orbital r
2EHOMO
adyn = ,
mω 2
where EHOMO and m are the energy and effective mass of those electrons.
As adyn decreases with the frequency very fast, the response of typical solids in
the optical range is highly localized (kas ≈ kamacro  1) with negligibly small spatial
dispersion. It allows us to treat εi j (ω, k) independent of the wavevector k, when
εi j (ω, k) = δi j ε(ω) (1.22)
with
εt (ω, k) = εl (ω, k) = ε(ω).
Thus, magnetically inactive uniform solids are well described at optical frequen-
cies and above within the local response approximation, where the material rela-
tions are completely given by the scalar complex dielectric permittivity ε(ω) =
ε 0 (ω) + iε 00 (ω),
D(ω, k) = ε0 ε(ω)E(ω, k), (1.23)
B(ω, k) = µ0 H(ω, k). (1.24)
Note that in Eq. (1.24) we wrote the relation B(H) instead of H(B). This formal
change is due to the symmetry of the Maxwell’s equations, following which it is
more natural to describe magnetic properties in terms of H rather than B. Namely for
this reason, the field H is commonly called the magnetic field by analogy with the
electric field E although it is actually an auxiliary quantity. Hereinafter, electromag-
netic fields will be described in terms of vectors E and H only.
LOCAL RESPONSE APPROXIMATION 9
1.2 Local response approximation
1.2.1 Energy of electromagnetic field in medium
According to the Maxwell’s equations, distribution of electromagnetic field in a
medium obeys the energy conservation law. For magnetically inactive materials, it
can be written as [3, 4]:
1 ∂
(ε0 E 2 + µ0 H 2 ) + ∇ · (E × H) + jpol · E = −jext · E, (1.25)
2 ∂t
describing the energy balance between electromagnetic fields and different types of
charges composing the medium. The right-hand side of this equation represents the
power density of the external currents spent on the polarization of the medium’s
element dV and the excitation of the electric and magnetic fields inside it,

Aext = −jext · E, (1.26)

while the left-hand side gives us different mechanisms for that energy expenditure.
The first one is the increase of the energy density
1
U0 = (ε0 E 2 + µ0 H 2 ) (1.27)
2
of the electric E and magnetic H fields over the time dt. The second mechanism is
the radiation of the energy out of the volume dV ; its contribution is given by the
divergence of the Poynting vector

S = E × H, (1.28)

which is considered as the energy flux density of the electromagnetic field. The third
mechanism is the work done by the polarization charges, described with the power
density
Qmed = jpol · E. (1.29)
In general, Qmed is comprised of the dissipation power density Qdis and the change
of the energy density Upol stored in the medium in the form of the polarization field
P,
∂Upol
Qmed = + Qdis . (1.30)
∂t
Thus, the conservation law for the overall electromagnetic field in the medium can
be written as
∂U
+ ∇ · S + Qdis = Aext , (1.31)
∂t
where U = U0 +Upol is the total energy density of the three fields E, H, and P.
To derive the energy characteristics in the local response approximation, we cal-
culate the time-averaged value of Qmed ,
 
1 1 ∂P ∗
Qmed = Re(jpol · E∗ ) = Re ·E , (1.32)
2 2 ∂t
10 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA
assuming that the electromagnetic field is not purely monochromatic, but has a small
frequency width,
E = E0 (t) e−iωt , H = H0 (t) e−iωt ,
where E0 (t) and H0 (t) are slowly varying functions, as compared to e−iωt . By doing
that, we can capture the effect of temporal dispersion that gives a significant contri-
bution to the energy stored in a medium [3, 5],
∂ ε0 |E|2 ∂ (ωε 0 )
 
ω
Qmed = − 1 + ε0 ε 00 |E|2 . (1.33)
∂t 4 ∂ω 2
By comparing with Eq. (1.30), we find that the time-averaged energy density of the
overall electromagnetic field (including the polarization one) is
1 ∂ (ωε 0 )
 
2 2
U= ε0 |E| + µ0 |H| (1.34)
4 ∂ω
with the dispersion effect given by the function ∂ (ωε 0 )/∂ ω, where ε 0 is the real part
of the complex dielectric permittivity ε(ω). For the time-averaged dissipation rate,
we obtain
ω
Qdis = ε0 ε 00 |E|2 , (1.35)
2
where ε 00 is the imaginary part of ε(ω). In this way, we conclude that the real part of
complex dielectric permittivity defines the energy of electromagnetic field, while the
imaginary part describes its dissipation.

1.2.2 Properties of complex dielectric permittivity

As we have already seen, scalar dielectric permittivity plays an essential role in de-
scription of the electrodynamic field interaction with a medium. Therefore, it is vital
to understand the main properties of ε(ω).
Some common qualities of the scalar dielectric permittivity can be obtained from
the general definition of σi j (ω, k) given by Eq. (1.18). Note the difference between
the tensors σi j (t, r) and σi j (ω, k) in that equation — the former is a purely real func-
tion, as it relates two real vectors J(t, r) and E(t, r), while the latter is generally com-
plex. Following this peculiarity, we can write the symmetry relation for σi j (ω, k),
σi j (−ω, −k) = σi∗j (ω, k),
which can be translated to the identical relation for εi j (ω, k) with Eq. (1.20),
εi j (−ω, −k) = εi∗j (ω, k).
Thus, for isotropic materials in the high-frequency range, we have
ε(−ω) = ε ∗ (ω).
In terms of the real ε 0 (ω) and imaginary ε 00 (ω) parts of ε(ω), it can be rewritten as
follows:
ε 0 (−ω) = ε 0 (ω), ε 00 (−ω) = −ε 00 (ω), (1.36)
LOCAL RESPONSE APPROXIMATION 11
exhibiting evenness for the real part and oddness for the imaginary part of scalar
permittivity.
Another property can be obtained if we consider the limiting case of ω → ∞.
This is the case of highly localized response with as → 0, when all intrinsic charges
oscillate at such a high frequency and short scale, so they do not feel each other. In
this regime, nuclei are immobile due to their high masses, so the medium’s response
is given by light electrons only driven by the electric field,

∂ v(t, r) e e
= − E(t, r) = − E0 (r) e−iωt ,
∂t m m
where v, −e, and m are the directional velocity, charge, and mass of the electrons.
Oscillating in the electric field E, electrons create the polarization current jpol given
by
jpol (t, r) = −en0 v(t, r),
where n0 is the total density of the medium’s electrons. Performing the Fourier trans-
form for v, E, and j, we derive

e2 n0
jpol (ω) = i E(ω).
mω
At the same time, according to the definition of the polarization vector P, the internal
current is given by

jpol (ω) = −iωP(ω) = −iωε0 [ε(ω) − 1]E(ω).

Thus, we get
e2 n0
ε(ω) = 1 −
ε0 mω 2
for the asymptotic behavior of scalar dielectric permittivity at extremely high fre-
quencies. In the limit of ω → ∞, it gives us

lim ε(ω) = 1,
ω→∞

or alternatively
lim ε 0 (ω) = 1, lim ε 00 (ω) = 0. (1.37)
ω→∞ ω→∞

Other, thermodynamic, properties of the scalar dielectric permittivity can be ob-

tained if we consider the energy density of the overall electromagnetic field U and the
energy dissipation rate Qdis given by Eqs. (1.34) and (1.35), respectively. Following
them, under the thermodynamic equilibrium, ε 0 (ω) and ε 00 (ω) should feature

∂  0
ωε 00 (ω) ≥ 0.

ωε (ω) ≥ 0, (1.38)
∂ω
The former condition imposes a restriction on temporal dispersion, while the latter
defines the sign of ε 00 (ω) for optically passive media.
12 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA
1.2.3 Medium response modeling
Up to this point, we gave the description of a charge-compensated medium with-
out mentioning its nature. It was the unified theory covering magnetically inactive
materials ranging from dielectrics to semiconductors and metals. Now, let us dis-
cuss different groups of materials and review common models developed for their
description in the local response approximation.
Within the local response approximation, the scalar complex dielectric permittiv-
ity can be written in the following form

ε(ω) = 1 + χ(ω), (1.39)

where χ(ω) is the scalar susceptibility of the neutral-charge medium that gives us
the material relation for the polarization field

P(ω) = ε0 χ(ω)E(ω). (1.40)

If there are independent mechanisms for polarization of the medium, we can write
the susceptibility as a sum over them

χ(ω) = ∑ χα (ω). (1.41)

α

Following it, understanding of typical dependences of χα (ω) given by polarization of

different groups α of intrinsic charges is of critical importance for medium response
modeling.
Below we overview most commonly used classical models developed for de-
scriptions of different media. They all are based on the relation of the polarization
field and current,
∂P
= jpol = ∑ qα nα vα , (1.42)
∂t α

where the polarization current is contributed by different species of the polarization

charges α characterized with the charge qα , volume density nα , and directional ve-
locity vα . After additional differentiation, we get

∂ 2P ∂ vα
2
= ∑ qα nα , (1.43)
∂t α ∂t

where ∂ vα /∂t is given by the motion equation different for every species.
Within the local response approximation (valid at high frequencies only), the
polarization current is mainly contributed by conduction and bound electrons due
to their light weight. Heavy nuclei can be considered immobile at these frequencies
without significant loss of accuracy, as their contribution to the polarization current
is at least a thousand times smaller than those of conduction and bound electrons.
LOCAL RESPONSE APPROXIMATION 13

Figure 1.1: Fitting of |χ 0 | with the dissipationless Drude model for silicon (Si), sil-
icon nitride (Si3 N4 ), and gold (Au). The fitted values of ω p0 are 31, 38, and 80
eV, respectively. The obtained total electron density is in excellent agreement with
the mass densities of the corresponding materials, which are 2.329, 3.44, and 19.32
g/cm3 . The optical data fitted are taken from Refs. [6, 7].

Dissipationless Drude model

The dissipationless Drude model has already been used in the previous section when
we considered properties of the scalar dielectric permittivity at extremely high fre-
quencies. In this case, all electrons (regardless of whether they are conduction or
bound) behave as free: they oscillate with the directional velocity driven by the high-
frequency electric field only,
∂v e
= − E, (1.44)
∂t m
without feeling each other and nuclei. As a result, the polarization of the medium is
given by the differential equation
∂ 2P 2
= ω p0 ε0 E, (1.45)
∂t 2
p
where ω p0 = n0 e2 /(mε0 ) is the plasma frequency of all electrons in the medium,
characterized with the total electron density n0 . The modeled polarization field and
scalar susceptibility have the following form
2 ε
ω p0 2
ω p0
0
P(ω) = − E, χ(ω) = − . (1.46)
ω2 ω2
This model describes the limiting case of highly localized response of any
medium, regardless of its electronic structure. The applicability of this model is re-
stricted to extremely high frequencies, above all characteristic frequencies of the
medium, ω  ωs . Fitting of experimental data with this model allows us to get ω p0
and estimate the total density of electrons in the medium, as shown in Fig. 1.1.
14 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA

Figure 1.2: Fitting of the scalar dielectric permittivity of silver (Ag) with the dissipa-
tive Drude model. The estimated ω p and γ correspond to 9.19 and 0.02 eV. The fitted
data are taken from Ref. [8].

Dissipative Drude model

In the dissipationless Drude model, we ignored interactions between electrons. If
we add friction to the equation of motion, we can get the dissipative Drude model
that describes the response of free electrons with significant contribution of electron-
electron scattering,
∂v e
= − E − γv, (1.47)
∂t m
where γ is the scattering rate. In this case, the equation for polarization field P has
the following form,
∂ 2P ∂P
2
+γ = ω p2 ε0 E, (1.48)
∂t ∂t
resulting in
ω p2 ε0 ω p2
P(ω) = − E, χ(ω) = − , (1.49)
ω(ω + iγ) ω(ω + iγ)
p
where ω p = ne2 /(mε0 ) is the plasma frequency of the free electrons with volume
density n.
This model was first proposed by Paul Drude in 1900 to explain electrical conduc-
tion of metals. It provides an adequate description of good conductors with dominat-
ing contribution of conduction electrons. As a rule, Drude-like response is observed
at lower frequencies, below the electron inter-band transitions where the polariza-
tion by conduction electrons dominates (Fig. 1.2). At higher frequencies, additional
terms accounting for the inter-band transitions should be considered as well. Through
fitting of ω p and γ, this model allows us to determine density of the conduction elec-
trons together with their relaxation time.
LOCAL RESPONSE APPROXIMATION 15

Figure 1.3: Characteristics of the oscillator susceptibility χi (ω) in the Lorentz model.

Lorentz oscillators model

Although the dissipative Drude model accounts for the electron-electron scattering,
it describes free electrons only and does not consider electron-nucleus bonds. Of
course, a full description of bound electrons response is very complicated; it requires
a quantum-mechanical consideration of all possible electron transitions. Therefore,
it is common to use the methods of classical mechanics to model the quantum re-
sponses of bound electrons. The most widely used model is of Lorentz oscillators.
It describes bound electrons as a system of N decoupled oscillators. Every oscillator
represents a set of identical bound electrons of density ni featuring the same bond
potential Ui and scattering rate γi , where i = 1, 2, ..., N. Then, the electron dynamics
of the i-th oscillator is given by
∂ vi e ∇Ui
= − E − γi vi − , (1.50)
∂t m m
so the corresponding partial polarization Pi obeys the equation

∂ 2 Pi ∂ Pi
+ γi + ω0i2 Pi = Ω2i ε0 E, (1.51)
∂t 2 ∂t
where Ω2i = e2 ni (mε0 )−1 and ω0i2 = (E · ∇)2Ui |rmin (|E|2 m)−1 . By solving Eq. (1.51)
we get the polarization field

Ω2i ε0
Pi (ω) = − E, (1.52)
ω 2 − ω0i2 + iγi ω
and susceptibility of the i-th Lorentz oscillator,

Ω2i
χi (ω) = − . (1.53)
ω 2 − ω0i2 + iγi ω
16 ELECTROMAGNETIC FIELDS IN UNIFORM MEDIA

Figure 1.4: Fitting of the scalar dielectric permittivity of (a) aluminium (Al) and (b)
silicon (Si) with the model of Lorentz oscillators. Optical data of Al are fitted well
with N = 3 oscillators, while Si data require N > 10 for satisfactory matching. The
fitted data are taken from Ref. [9].

Finally, to get the total polarization field and susceptibility, we need to sum the re-
spective partial contributions over the oscillator index i
N N
P(ω) = ∑ Pi (ω), χ(ω) = ∑ χi (ω). (1.54)
i=1 i=1

Although the Lorentz oscillators model is based on classical mechanics princi-

ples, it provides a very good fitting for bound electron responses. More accurate
quantum-mechanical considerations confirm the Lorentz shape of scalar suscepti-
bility caused by electron transitions [10], although the model’s parameters Ωi , ω0i ,
γi bear different meanings. In addition, the Lorentz oscillators model covers the re-
sponse of conduction electrons as well. If we choose ω0i = 0 and N = 1, we get the
Drude model where ω p = Ωi . All this makes χi (ω) a universal spectroscopic func-
tion that can be used in composition of any type of dielectric permittivity. Indeed,
Another random document with
no related content on Scribd:
layer"; these are the "collencytes" and "scleroblasts"; secondly, it
contains "archaeocytes," cells of independent origin.

Collencytes are cells with clear protoplasm and thread-like

pseudopodial processes; they are distinguished as stellate or bipolar,
according as these processes are many or only two. Scleroblasts or
spicule cells are at first rounded, but become elongated with the
growth of the spicule they secrete, and when fully grown are
consequently fusiform.

Fig. 65.—Diagrammatic section of a siliceous Sponge. a.p, Apopyle; d.o, dermal

ostia; ex.c, excurrent, or exhalant canal; in.c, incurrent canal; o, osculum.
(Modified from Wilson.)

Each spicule consists of an organic filamentar axis or axial fibre

around which sheaths of silica hydrate are deposited successively by
the scleroblast. Over the greater length of the spicule the sheaths
are cylindrical, but at each end they taper to a point. The axial canal
in which the axial fibre lies is open at both ends, and the fibre is
continuous at these two points with an organic sheath, which invests
the entire spicule. From this structure we may conclude that the
spicule grows at both ends—i.e. it grows in two opposite directions
along one line—it has two rays lying in one axis, and is classed
among uniaxial diactinal spicules. Being pointed at both ends it
receives the special name oxea. The lamination of the spicule is
rendered much more distinct by heating or treatment with caustic
potash.[197]
 Fig. 66.—Cut end of a length of a siliceous spicule from Hyalonema sieboldii,
with the lamellar structure revealed by solution. × 104. (After Sollas.)

The archaeocytes are rounded amoeboid cells early set apart in the
larva; they are practically undifferentiated blastomeres. Some of
them become reproductive elements, and thus afford a good
instance of "continuity of germ plasm," others probably perform
excretory functions.[198]

Fig. 67.—Free-swimming larva of Gellius varius, in optical section. a, Outer

epithelium; pi, pigment; x, hinder pole. (After Maas.)

The reproductive elements are ova and spermatozoa, and are to be

found in all stages in the dermal jelly. Dendy states that the eggs are
fertilised in the inhalant canals, to which position they migrate by
amoeboid movements, and there become suspended by a peduncle.

The larva has unfortunately not been described, but as the course of
development among the near relatives of H. panicea is known to be
fairly constant, it will be convenient to give a description of a
"Halichondrine type" of larva based on Maas' account of the
development of Gellius varius.[199] The free-swimming larvae escape
by the osculum; they are minute oval bodies moving rapidly by
means of a covering of cilia. The greater part of the body is a
dazzling white, while the hinder pole is of a brown violet colour. This
coloured patch is non-ciliate, the general covering of cilia ending at
its edge in a ring of cilia twice the length of the others. Forward
movement takes place in a screw line; when this ceases the larva
rests on its hinder pole, and the cilia cause it to turn round on its
axis.

Sections show that the larva is built up of two layers:—

1. "The inner mass," consisting of various kinds of cells in a

gelatinous matrix.

2. A high flagellated epithelium, which entirely covers the larva with

the exception of the hinder pole.

Fig. 68.—Longitudinal section through the hinder pole of the larva of G. varius. a,
Flagellated cells; ma1, undifferentiated cell; ma2, differentiated cell; pi,
pigment; x, surface of hinder pole. (After Maas.)

The cells in the inner mass are classified into (1) undifferentiated
cells, recognised by their nucleus, which possesses a nucleolus;
these are the archaeocytes; (2) differentiated cells, of which the
nucleus contains a chromatin net; these give rise to pinacocytes,
collencytes, and scleroblasts. Some of them form a flat epithelium,
which covers the hinder pole. Some of the scleroblasts already
contain spicules. Fixation occurs very early. The front pole is used
for attachment, the pigmented pole becoming the distal end (Fig.
69). The larva flattens out, the margin of the attached end is
produced into radiating pseudopodial processes. The flagellated
cells retreat to the interior, leaving the inner mass exposed, and
some of its cells thereupon form a flat outer epithelium. This is the
most important process of the metamorphosis; it is followed by a
pause in the outward changes, coinciding in time with
rearrangements of the internal cells to give rise to the canal system;
that is to say, lacunae arise in the inner mass, pinacocytes pass to
the surface of the lacunae, and form their lining; the flagellated cells,
which have lain in confusion, become grouped in small clusters.
These become flagellated chambers, communications are
established between the various portions of the canal system, and its
external apertures arise. There is at first only one osculum. The
larvae may be obtained by keeping the parent sponge in a dish of
sea water, shielded from too bright a light, and surrounded by a
second dish of water to keep the temperature constant. They will
undergo metamorphosis in sea water which is constantly changed,
and will live for some days.

We have said that the young sponge has only one osculum. This is
the only organ which is present in unit number, and it is natural to
ask whether perhaps the osculum may not be taken as a mark of the
individual; whether the fistular specimens, for example, of H. panicea
may not be solitary individuals, and the cockscomb and other forms
colonies in which the individuals are merged to different degrees.
Into the metaphysics of such a view we cannot enter here. We must
be content to refer to the views of Huxley and of Spencer on
Individuality.

But it is advisable to avoid speaking of a multi-osculate sponge as a

colony of many individuals, even in the sense in which it is usual to
speak of a colony of polyps as formed of individuals. The repetition
of oscula is probably to be regarded as an example of the
phenomenon of repetition of parts, the almost universal occurrence
of which has been emphasised by Bateson.[200] Delage[201] has
shown that when two sponge larvae fixed side by side fuse together,
the resulting product has but one osculum. This, though seeming to
bear out our point of view, loses weight in this connexion, when it is
recalled that two Echinoderm larvae fused together give rise in a
later stage to but one individual.
 Fig. 69.—Larva of Gellius varius shortly after fixation. The pigmented pole,
originally posterior, is turned towards the reader. R, Marginal membrane
with pseudopodia; x, hinder pole. (After Maas.)

Ephydatia fluviatilis.
In the fresh water of our rivers, ponds, and lakes, sponges are
represented very commonly by Ephydatia (Spongilla) fluviatilis, a
cosmopolitan species. The search for specimens is most likely to be
successful if perpendicular timbers such as lock-gates are examined,
or the underside of floating logs or barges, or overhanging branches
of trees which dip beneath the surface of the water.

The sponge is sessile and massive, seldom forming branches, and is

often to be found in great luxuriance of growth, masses of many
pounds weight having been taken off barges in the Thames. The
colour ranges from flesh-tint to green, according to the exposure to
light. This fact is dealt with in a most interesting paper by Professor
Lankester,[202] who has shown not only that the green colour is due
to the presence of chlorophyll, but that the colouring matter is
contained in corpuscles similar to the chlorophyll corpuscles of green
plants, and, further, that the flesh-coloured specimens contain
colourless corpuscles, which, though differing in shape from those
which contain the green pigment, are in all probability converted into
these latter under the influence of sufficient light. The corpuscles,
both green and colourless, are contained in amoeboid cells of the
dermal layer;[203] and in the same cells but not in the corpuscles are
to be found amyloid substances.
The anatomy of Ephydatia fluviatilis is very similar to that of
Halichondria panicea, differing only in one or two points of
importance. The ectosome is an aspiculous membrane of dermal
tissue covering the whole exterior of the sponge and forming the roof
of a continuous subdermal space. This dermal membrane is
perforated by innumerable ostia, and is supported above the
subdermal cavity by means of skeletal strands, which traverse the
subdermal cavity and raise the dermal membrane into tent-like
elevations, termed conuli. The inhalant canals which arise from the
floor of the subdermal cavity are as irregular as in H. panicea, and
interdigitate with equally irregular exhalant canals; these latter
communicate with the oscular tubes. Between the two sets of canals
are the thin folds of the choanosome with its small subspherical
chambers provided with widely open apopyles (Fig. 70). The soft
parts are supported on a siliceous skeleton of oxeas, which may
have a quite smooth surface or may be covered in various degrees
with minute conical spines (Fig. 72, a, b). These spicules are
connected by means of a substance termed spongin deposited
around their overlapping ends, so as to form an irregular network of
strands, of which some may be distinguished as main strands or
fibres, others as connecting fibres. In the main fibres several
spicules lie side by side, while in the connecting fibres fewer or
frequently single spicules form the thickness of the fibre. The fibres
are continuous at the base with a plate or skin of spongin, which is
secreted over the lower surface of the sponge and intervenes
between it and the substratum. Of the chemical composition of
spongin we shall speak later (see p. 237). It is a substance which
reaches a great importance in some of the higher sponges, and
forms the entire skeleton of certain kinds of bath sponge. Lying loose
in the soft parts and hence termed flesh spicules, or microscleres,
are minute spicules of peculiar form. These are the amphidiscs,
consisting of a shaft with a many-rayed disc at each end (Fig. 72).
 Fig. 70.—Ephydatia fluviatilis. Section of flagellated chamber, showing the
choanocytes passing through the apopyle. (After Vosmaer and
Pekelharing.)

In addition to its habitat the fresh-water sponge is worthy of attention

on account of its methods of reproduction, which have arisen in
adaptation to the habitat. A similar adaptation is widespread among
fresh-water members of most aquatic invertebrates.[204]

Ephydatia fluviatilis normally produces not only free-swimming larvae

of sexual origin, but also internal gemmules arising asexually. These
bodies appear in autumn, distributed throughout the sponge, often
more densely in the deeper layers, and they come into activity only
after the death of the parent, an event which happens in this climate
at the approach of winter.

Fig. 71.—Portion of the skeletal framework of E. fluviatilis. a, Main fibres; b,

connecting fibres. (After Weltner.)
 Fig. 72.—Spicules of E. fluviatilis. a. b. c. Oxeas, spined and smooth; d. e,
amphidiscs, side and end views. (After Potts.)

Weltner[205] has shown that on the death and disintegration of the

mother sponge some of the gemmules remain attached to the old
skeleton, some sink and some float. Those which remain attached
are well known to reclothe the dead fibres with living tissue. They
inherit, as it were, the advantages of position, which contributed to
the survival of the parent, as one of the selected fittest. The
gemmules which sink are doubtless rolled short distances along the
bottom, while those which float have the opportunity of widely
distributing the species with the risk of being washed out to sea. But
even these floating gemmules are exposed to far less dangers than
the delicate free-swimming larvae, for their soft parts are protected
from shocks by a thick coat armed with amphidiscs.

The gemmules are likewise remarkable for their powers of resistance

to climatic conditions, powers which must contribute in no small way
to the survival of a species exposed to the variable temperatures of
fresh water. Thus, if the floating gemmules or the parent skeleton
with its attached and dormant offspring should chance to be included
in the surface layer of ice during the winter, so far from suffering any
evil consequences they appear to benefit by these conditions. Both
Potts and Weltner have confirmed the truth of this statement by
experiments. Weltner succeeded in rearing young from gemmules
which had suffered a total exposure of 17 days to a temperature
"under 0° C."
Of important bearing on the question of the utility of the gemmules
are certain instances in which E. fluviatilis has been recorded as
existing in a perennial condition.[206] The perennial individuals may
or may not bear gemmules, which makes it evident that, with the
acquisition of the power to survive the winter cold, the prime
necessity of forming these bodies vanishes.

The perennial specimens are described as exhibiting a diminished

vegetative activity in winter, the flagellated chambers may be absent
(Lieberkühn), or present in unusually small numbers (Weltner), the
entire canal system may be absent (Metschnikoff), or, on the other
hand, it may be complete except for the osculum.

Fig. 73.—Gemmule of E. fluviatilis. b, Amphidisc. (After Potts.)

In tropical countries gemmulation occurs as a defence against the

ravages caused by the dry season when the waters recede down
their banks, exposing all or most of their sponge inhabitants to the
direct rays of the sun. The sponges are at once killed, but the
contained gemmules being thoroughly dried, become efficient
distributing agents of the species; they are light enough to be carried
on the wind. It is probable that those individual sponges which
escape desiccation survive the dry season without forming
gemmules.

It has been shown experimentally that gemmules are not injured by

drying—Zykoff found that gemmules kept dry for a period of two
years had not lost the power of germination.

The mature gemmules consist of a more or less spherical mass of

cells, which we shall refer to as yolk cells, and of a complex coat.
The latter is provided with a pore or pore tube (Fig. 74) which is
closed in winter by an organic membrane.

There are three layers in the coat: an inner chitinous layer

surrounded by an air-chamber layer, which is finely vesicular,
showing a structure recalling plant tissue, and containing amphidiscs
arranged along radii passing through the centre of the gemmule.
One of the discs of each amphidisc lies in the inner chitinous coat,
while the other lies in a similar membrane which envelopes the air-
chamber layer and is termed the outer chitinous coat.

Marshall has suggested that one function of the amphidiscs is to

weight the gemmules and thus protect them against the force of the
river current; and no doubt the sinking or floating of individual
gemmules depends on the relative degree of development of the air-
chambers and of the amphidiscs.

A study of the development of Ephydatia gemmules vividly illustrates

various characters of the inner processes of sponges. Specially
noteworthy are the migrations of cells and the slight extent to which
division of labour is carried: one and the same cell will be found to
perform various functions.

Fig. 74.—Part of a longitudinal section of a gemmule of Ephydatia sp. passing

through the pore (a). (After Potts.)

The beginning of a gemmule is first recognisable[207] as a small

cluster of amoeboid archaeocytes in the dermal membrane. These
move into the deeper parts of the sponge to form larger groups.
They are the essential part of the gemmule, the yolk cells, which,
when germination takes place, give rise to a new sponge. They are
followed by two distinct troops of actively moving cells. Those
forming the first troop arrange themselves round the yolk cells and
ultimately assume a columnar form so that they make an epithelioid
layer. They then secrete the inner chitinous coat. The cells of the
second troop are entrusted with the nutrition of the gemmule.
Consequently they pass in among the yolk cells, distribute their food
supplies, and make their escape by returning into the tissues of the
mother sponge, before the columnar cells have completed the
chitinous coat. Yet another migration now occurs, the cells
—"scleroblasts"—which have been occupied in secreting amphidiscs
at various stations in the sponge, carry the fully formed spicules to
the gemmules and place them radially round the yolk cells between
the radially lying cells of the columnar layer. The scleroblasts
themselves remain with the amphidiscs, and becoming modified,
contribute to the formation of the air-chamber layer. The columnar
cells now creep out between the amphidiscs till their inner ends rest
on the outer ends of these spicules. They then secrete the outer
chitinous coat and return to the mother sponge.

Carter gives directions[208] for obtaining young sponges from the

gemmules. The latter should be removed from the parent, cleaned
by rolling in a handkerchief, and then placed in water in a watch-
glass, protected with a glass cover and exposed to sunlight. In a few
days the contents of the gemmule issue from the foramen and can
be seen as a white speck. A few hours later the young sponge is
already active and may be watched producing aqueous currents. At
this age the sponge is an excellent object for studying in the living
condition: being both small and transparent it affords us an
opportunity of watching the movements of particles of carmine as
they are carried by the current through the chambers.

Potts[209] describes how he has followed the transportal of spicules

by dermal cells, the end of each spicule multiplying the motion,
swaying like an oscillating rod.

In E. fluviatilis reproduction also occurs during the warmer months in

this climate by means of sexual larvae. These are interesting for
certain aberrant features in their metamorphosis.[210] While some of
the flagellated chambers are formed in the normal way from the
flagellated cells of the larva, others arise each by division of a single
archaeocyte. This, it is suggested, is correlated with the acquisition
of the method of reproduction by gemmules, the peculiarities (i.e.
development of organs from archaeocytes) of which are appearing in
the larvae.

Definition.—We may now define sponges as multicellular, two-

layered animals; with pores perforating the body-walls and admitting
a current of water, which is set up by the collared cells of the
"gastral" layer.

Position in the Animal Kingdom.—Sponges are the only

multicellular animals which possess choanocytes, and their mode of
feeding is unique. Since they are two-layered it has been sought to
associate them with the Metazoan phylum Coelenterata, but they are
destitute of nematocysts or any other form of stinging cell, and their
generative cells arise from a class of embryonic cells set apart from
the first, while the generative cells of Coelenterata are derived from
the ectoderm, or in other cases from the endoderm. These weighty
differences between sponges and that group of Metazoa to which
they would, if of Metazoan nature at all, be most likely to show
resemblance, suggest that we should seek a separate origin for
sponges and Metazoa. We naturally turn to the Choanoflagellate
Infusorian stock (see p. 121) as the source of Porifera, leaving the
Ciliate stock as the progenitors of Metazoa.

That both Porifera and Metazoa are reproduced by ova and

spermatozoa is no objection to this view, seeing that the occurrence
of similar reproductive cells has been demonstrated in certain
Protozoa (see pp. 100, 128).

Let us now see which view is borne out by facts of embryology.

Suppose, for the moment, we regard sponges as Metazoa, then if
the sponge larva be compared with the Metazoan larva we must
assign the large granular cells to the endoderm; the flagellated cells
to the ectoderm; and we are led to the anomalous statement that the
digestive cells in the adult are ectodermal, the covering, outer cells
endodermal; or conversely, if we start our comparisons with the
adults, then it follows that the larval ectoderm has the characters of
an endoderm, and the larval endoderm those of an ectoderm.

Thus both embryology and morphology lead us to the same point,

they both show that in the absence of any fundamental agreement
between Porifera and Metazoa it is necessary to regard the two
stocks as independent from the very first, and hence the name
Parazoa (Sollas) has been given to the group which contains the
Porifera as its only known phylum.

Interesting in connexion with the phylogeny of Parazoa is the

Choanoflagellate genus Proterospongia (Fig. 75), described by
Saville Kent, and since rediscovered both in England and abroad.
[211] This is a colony of unicellular individuals embedded in a
common jelly. The individuals at the surface are choanoflagellate,
while in the interior the cells are rounded or amoeboid, and some of
them undergo multiple fission to form reproductive cells. This is just
such a creature as we might imagine that ancestral stage to have
been of which the free-swimming sponge larva is a reminiscence: for
we have seen that the flagellated cells of the larva are potential
choanocytes.
 Fig. 75.—Proterospongia haeckeli. a, Amoeboid cell; b, a cell dividing; c, cell with
small collar; z, jelly. × 800. (After S. Kent.)

CHAPTER VIII

PORIFERA (CONTINUED): FORMS OF SPICULES—CALCAREA—

HOMOCOELA—HETEROCOELA—HEXACTINELLIDA—DEMOSPONGIAE—
TETRACTINELLIDA—MONAXONIDA—CERATOSA—KEY TO BRITISH
GENERA OF SPONGES

Sponges fall naturally into two branches differing in the size of their
choanocytes: in the Megamastictora these cells are relatively
large, varying from 5µ to 9µ in diameter; in Micromastictora they
are about 3µ in diameter.[212] For further subdivision of the group the
spicules are such important weapons in the hands of the
systematist that it is convenient to name them according to a
common scheme. This has been arrived at by considering first the
number of axes along which the main branches of the spicules are
distributed, and secondly whether growth has occurred in each of
these axes in one or both directions from a point of origin.[213]

I. Monaxons.—Spicules of rod-like form, in which growth is directed

from a single origin in one or both directions along a single axis. The
axis of any spicule is not necessarily straight, it may be curved or
undulating. The ray or rays are known as actines.

Biradiate monaxon spicules are termed "rhabdi" (Fig. 76, a). A

rhabdus pointed at both ends is an "oxea," rounded at both ends a
"strongyle," knobbed at both ends a "tylote." By branching a rhabdus
may become a "triaene" (Fig. 110, k, l).

Uniradiate monaxon spicules are termed "styli."

II. Tetraxons.—Spicules in which growth proceeds from an origin in
one direction only, along four axes arranged as normals to the faces
of a regular tetrahedron. Forms produced by growth from an origin in
one direction along three axes lying in one plane are classed with
tetraxons.

III. Triaxons.—Spicules in which growth is directed from an origin in

both directions along three rectangular axes. One or more actines or
one or two axes may be suppressed.

IV. Polyaxons.—Spicules in which radiate growth from a centre

proceeds in several directions.

V. Spheres.—Spicules in which growth is concentric about the origin.

A distinction more fundamental than that of form is afforded by the

chemical composition: all sponges having spicules composed of
calcium carbonate belong to a single class, Calcarea, which stands
alone in the branch Megamastictora.

Fig. 76.—Types of megascleres. a, Rhabdus (monaxon diactine); b, stylus

(monaxon monactine); c, triod (tetraxon triactine); d, calthrop (tetraxon
tetractine); e, triaxon hexactine; f, euaster.

BRANCH I. MEGAMASTICTORA
CLASS CALCAREA
Calcarea are marine shallow-water forms attached for the most part
directly by the basal part of the body or occasionally by the
intervention of a stalk formed of dermal tissue. They are almost all
white or pale grey brown in colour. Their spicules are either monaxon
or tetraxon or both. The tetraxons are either quadriradiate and then
called "calthrops," or triradiate when the fourth actine is absent. The
triradiates always lie more or less tangentially in the body-wall;
similarly three rays of a calthrop are tangentially placed, the fourth
lying across the thickness of the wall. It is convenient to include the
triradiate and the three tangentially placed rays of a calthrop under
the common term "triradiate system" (Minchin). The three rays of
one of these systems may all be equal in length and meet at equal
angles: in this case the system is "regular." Or one ray or one angle
may differ in size from the other rays or angles respectively, which
are equal: in either of these two cases the system is bilaterally
symmetrical and is termed "sagittal." A special name "alate" is given
to those systems which are sagittal in consequence of the inequality
in the angles. Thus all equiangular systems whether sagittal or not
are opposed to those which are alate. This is the natural
classification.[214]

Sub-Class I. Homocoela.
The Homocoela or Ascons possess the simplest known type of canal
system, and by this they are defined. The body is a sac, branched in
the adult, but simple in the young; its continuous cavity is
everywhere lined with choanocytes, its wall is traversed by inhalant
pores, and its cavity opens to the exterior at the distal end by an
osculum. The simple sac-like young is the well-known Olynthus of
Haeckel—the starting-point from which all sponges seem to have set
out. Two processes are involved in the passage from the young to
the adult, namely, multiplication of oscula and branching of the
original Olynthus tube or sac. If the formation of a new osculum is
accompanied by fission of the sac, and the branching of the latter is
slight, there arises an adult formed of a number of erect, well
separated main tubes, each with one osculum and lateral branches.
Such is the case in the Leucosoleniidae. In the Clathrinidae, on
the other hand, branching of the Olynthus is complicated, giving rise
to what is termed reticulate body form, that is, a sponge body
consisting of a network of tubules with several oscula, but with no
external indication of the limits between the portions drained by each
osculum. These outward characters form a safe basis for
classification, because they are correlated with other fundamental
differences in structure and development.[215]

As in Halichondria, and in fact all sponges, the body-wall is formed of

two layers; the gastral layer, as we have said, forming a continuous
lining to the Ascon tube and its branches. The dermal layer includes
a complete outer covering of pinacocytes, which is reflected over the
oscular rim to meet the gastral layer at the distal end of the tube; a
deeper gelatinous stratum in which lie scleroblasts and their
secreted products—calcareous spicules; and finally porocytes.[216]
These last are cells which traverse the whole thickness of the thin
body-wall, and are perforated by a duct or pore. The porocytes are
contractile, and so the pores may be opened or closed; they are a
type of cell which is known only in Calcarea. It will be noticed that the
fusiform or stellate "connective tissue cells" are absent. The layer of
pinacocytes as a whole is highly contractile, and is capable of
diminishing the size of the sponge to such an extent as quite to
obliterate temporarily the gastral cavity.[217]

The choanocytes show certain constant differences in structure in

the families Clathrinidae and Leucosoleniidae respectively. In the
former, the nucleus of the choanocyte is basal; in the latter, it is
apical, and the flagellum can be traced down to it (Fig. 77).
 Fig. 77.—The two types of Asconid collar cells. A, of Clathrina, nucleus basal; B,
of Leucosolenia, nucleus not basal, flagellum arising from the nuclear
membrane. (A, after Minchin; B, after Bidder.)

The tetraxon spicules have "equiangular" triradiate systems in the

Clathrinidae, while in Leucosoleniidae they are "alate." Finally, the
larva of Clathrinidae is a "parenchymula" (see p. 226), that of
Leucosoleniidae an "amphiblastula."

The fact that it is possible to classify the Calcarea Homocoela largely

by means of histological characters is in accordance with the
importance of the individual cell as opposed to the cell-layers
generally throughout the Porifera, and is interesting in serving to
emphasise the low grade of organisation of the Phylum. The organs
of sponges are often unicellular (pores), or the products of the
activity of a single cell (many skeletal elements); and even in the
gastral layer, which approaches nearly to an epithelium, comparable
with the epithelia of Metazoa, the component cells still seem to
assert their independence, the flagella not lashing in concert,[218] but
each in its own time and direction.

Sub-Class II. Heterocoela.

Fig. 78.—Transverse section of the body-wall of Sycon carteri, showing articulate

tubar skeleton, gastric ostia (a.p), tufts of oxeas at the distal ends of the
chambers (fl.ch), and pores (p). (After Dendy.)
 Fig. 79.—Sycon coronatum. At a a portion of the wall is removed, exposing the
paragaster and the gastric ostia of the chambers opening into it.

The Heterocoela present a series of forms of successive grades of

complexity, all derivable from the Ascons, from which they differ in
having a discontinuous gastral layer. The simplest Heterocoela are
included in the family Sycettidae, of which the British representative
is Sycon (Fig. 79). In Sycon numerous tubular flagellated chambers
are arranged radially round a central cavity, the "paragaster," into
which they open (Figs. 78, 79). The chambers, which are here often
called radial tubes, are close set, leaving more or less quadrangular
tubular spaces, the inhalant canals, between them; and where the
walls of adjacent chambers come in contact, fusion may take place.
Pores guarded by porocytes put the inhalant canals into
communication with the flagellated chambers. The paragaster is
lined by pinacocytes; choanocytes are confined to the flagellated
chambers.

The skeleton is partly defensive, partly supporting; one set of

spicules strengthens the walls of the radial tubes and forms
collectively the "tubar skeleton." It is characteristic of Sycettidae that
the tubar skeleton is of the type known as "articulate"—i.e. it is
formed of a number of successive rings of spicules, instead of
consisting of a single ring of large spicules which run the whole
length of the tube.

Fig. 80.—Sycon setosum. Young Sponge. × 200. d, Dermal cell; g, gastral cell; o,
osculum; p, pore cell; sp1, monaxon; sp3, triradiate spicule. (After Maas.)

The walls of the paragaster are known as the "gastral cortex"; they
contain quadriradiate spicules, of which the triradiate systems lie
tangentially in the gastral cortex, while the apical ray projects into the
paragaster, and is no doubt defensive. The distal ends of the
chambers bristle with tufts of oxeate spicules, and the separate
chambers are distinguishable in surface view. It is interesting to
notice that in some species of Sycon, the gaps between the distal
ends of the chambers are covered over by a delicate perforated
membrane, thus leading on, as we shall see presently, to the next
stage of advance.[219] The larva of Sycon is an amphiblastula (see
p. 227). Fig. 80 is a drawing of the young sponge soon after fixation;
it would pass equally well for an ideally simple Ascon or, neglecting
the arrangement of the spicules, for an isolated radial tube of Sycon.
Figs. 81, 82 show the same sponge, somewhat older. From them it is
seen that the Sycon type is produced from the young individual, in
what may be called its Ascon stage, by a process of outgrowth of
tubes from its walls, followed by restriction of choanocytes to the
flagellated chambers. Minute observation has shown[220] that this
latter event is brought about by immigration of pinacocytes from the
exterior. These cells creep through the jelly of the dermal layer and