High Frequency Acoustics in Colloid-Based Meso - and Nanostructures by Spontaneous Brillouin Light Scattering (Springer Theses) - Tim Still

Springer Theses
Recognizing Outstanding Ph.D. Research
For further volumes:

http://www.springer.com/series/8790
Aims and Scope
The series ‘‘Springer Theses’’ brings together a selection of the very best
Ph.D. theses from around the world and across the physical sciences. Nominated
and endorsed by two recognized specialists, each published volume has been
selected for its scientific excellence and the high impact of its contents for the
pertinent field of research. For greater accessibility to non-specialists, the
published versions include an extended introduction, as well as a foreword by
the student’s supervisor explaining the special relevance of the work for the field.
As a whole, the series will provide a valuable resource both for newcomers to the
research fields described, and for other scientists seeking detailed background
information on special questions. Finally, it provides an accredited documentation
of the valuable contributions made by today’s younger generation of scientists.
Theses are accepted into the series by invited nomination only

and must fulfill all of the following criteria
• They must be written in good English.

• The topic of should fall within the confines of Chemistry, Physics and related
interdisciplinary fields such as Materials, Nanoscience, Chemical Engineering,
Complex Systems and Biophysics.
• The work reported in the thesis must represent a significant scientific advance.
• If the thesis includes previously published material, permission to reproduce this
must be gained from the respective copyright holder.
• They must have been examined and passed during the 12 months prior to
nomination.
• Each thesis should include a foreword by the supervisor outlining the
significance of its content.
• The theses should have a clearly defined structure including an introduction
accessible to scientists not expert in that particular field.
Tim Still
High Frequency Acoustics

in Colloid-Based Meso-
and Nanostructures by
Spontaneous Brillouin
Light Scattering
Doctoral Thesis performed at Max Planck Institute

for Polymer Research and accepted by
Johannes Gutenberg University,
Mainz, Germany
123
Author Supervisor
Tim Still Prof. Hans-Jürgen Butt
Max Planck Institute for Polymer Research Max Planck Institute for Polymer Research
Ackermannweg 10 Ackermannweg 10
55128, Mainz, Germany 55128, Mainz, Germany
[email protected] [email protected]
ISSN 2190-5053 e-ISSN 2190-5061
ISBN 978-3-642-13482-1 e-ISBN 978-3-642-13483-8
DOI 10.1007/978-3-642-13483-8
Springer Heidelberg Dordrecht London New York
Library of Congress Control Number: 2010929574
Springer-Verlag Berlin Heidelberg 2010
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcast-
ing, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this
publication or parts thereof is permitted only under the provisions of the German Copyright Law of
September 9, 1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
Cover design: eStudio Calamar, Berlin/Figueres
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Der Klügere gibt nach! Eine traurige Wahrheit,
sie begründet die Weltherrschaft der Dummheit.
(The wiser man gives in! A distressing truth,
it establishes the world domination of stupidity.)
Marie von Ebner-Eschenbach

Parts of this thesis have been published in the following journal
articles:
T. Still, W. Cheng, M. Retsch, R. Sainidou, J. Wang, U. Jonas, N. Stefanou,

G. Fytas, Phys. Rev. Lett. 2008, 100, 194301. Copyright 2008 APS.
T. Still, W. Cheng, M. Retsch, U. Jonas, G. Fytas, J. Phys. Condens. Matt. 2008,

20, 404203. Copyright 2008 IOP.
T. Still, R. Sainidou, M. Retsch, U. Jonas, P. Spahn, G. P. Hellmann, G. Fytas,

Nano Lett. 2008, 8, 3194. Copyright 2008 American Chemical Society.
T. Still, M. D’Acunzi, D. Vollmer, G. Fytas, J. Coll. Interf. Sci. 2009, 340, 42.
Copyright 2009, Elsevier, Inc. All rights reserved.
K. Kearns, T. Still (equal contribution), G. Fytas, M. D. Ediger, Adv. Mater. 2010,

22, 39. Copyright 2010 Wiley-VCH.
T. Still, M. Retsch, U. Jonas, R. Sainidou, P. Rembert, K. Mpoukouvalas, G. Fytas,

Macromolecules 2010, 43, 3422. Copyright 2010 American Chemical Society.
Supervisor’s Foreword
Materials that can mold the flow of elastic waves of certain energy in certain
directions are called phononic materials. The present thesis deals essentially with
such phononic systems, which are structured in the mesoscale (\1 lm), and with
their individual components. Such systems show interesting phononic properties in
the hypersonic region, i.e., at frequencies in the GHz range. It is shown that
colloidal systems are excellent model systems for the realization of such phononic
materials. Therefore, different structures and particle architectures are investigated
by Brillouin light scattering, the inelastic scattering of light by phonons.
Both the mechanical properties of the individual colloidal particles, which
manifest in their resonance vibrations (eigenmodes), as well as the acoustic
propagation in colloidal structures have been investigated. The measurement of the
eigenmodes allows for new insights into physical properties at the mesoscale, e.g.,
confinement effects, copolymer behavior, or the non-destructive determination of
nanomechanical properties of core–shell particles, supporting the working groups
aim to achieve a deeper understanding of ‘soft mechanics’ at small length scales.
Another novel contribution assigned to this thesis is the first experimental reali-
zation of a phononic band gap arising from the interaction of these particle ei-
genmodes with the effective medium band (hybridization gap). This finding
already gave new impulses to the whole field of phononics.
The thesis was performed between 03/2007 and 08/2009 at Max Planck Institute
for Polymer Research, leading to several publications and presentations in inter-
national conferences, and was honored summa cum laude by the University
of Mainz.
Mainz, March 2010 Prof. Dr. H.-J. Butt
ix
Acknowledgments
The work presented in this thesis would be nothing without the help of so many
people, who advised me, who provided samples, who eased my daily life, or
helped me in many other ways to make not only a happy but also a successful time
out of the last two and a half years.
First of all, my thanks go to Prof. Hans-Jürgen Butt, who gave me the oppor-
tunity to do my Ph.D. in his research group at Max Planck Institute for Polymer
Research. I enjoyed very much the freedom of science, which is not given
everywhere to a Ph.D. student.
Prof. Carsten Sönnichsen (J. Gutenberg-Univ. Mainz) and Prof. Alfons van
Blaaderen (University of Utrecht) kindly agreed to be the second and third referee
of my thesis.
The deepest gratitude I owe to Prof. George Fytas, who was a great supervisor,
giving me all the freedoms I needed to develop my own ideas, but on the other
hand pushing me in the best possible meaning of the word to go on with them in a
very consequent and productive way.
I would like to thank Prof. Werner Steffen for many scientific and non-scientific
discussions, technical support, proof-reading a part of this thesis, and for hiring
me.
When I started my Ph.D. in Mainz, there were two fellows that supported me a
lot by introducing me into BLS and some of the topics that accompanied me for
the next years, Dr. Cheng Wei and Dr. Eugenia Nuñez. Especially Cheng Wei was
always available to help me to solve my smaller and bigger problems with the
setup and to discuss my results. Beyond that, I enjoyed his special interpretation of
far eastern philosophy (‘You can even ride on a donkey!’).
When reading this thesis, it becomes clear that nearly all presented results are
based on diverse cooperations, since the success of my studies was up to the
continuous supply with all kinds of high quality samples. The most fruitful and
longest lasting of these cooperations was with Dr. Markus Retsch and Dr. Uli
Jonas. I really enjoyed our work together, since both of them came up with a great
deal of improvements and completely new ideas. Markus was the best conceivable
‘colloid cook’, who always delivered requested samples or realized just discussed
xi
xii Acknowledgments
ideas incredibly shortly and with impressive quality. Beyond science, he became a
good friend.
I also acknowledge the fruitful cooperation with Maria D’Acunzi, Gabriele
Schäfer and Dr. Doris Vollmer (all MPIP), leading to the investigation of PS–silica
core–shell particles.
I thank Dr. Peter Spahn and Diana Kiefer from Dr. Götz Hellmann’s group at
DKI Darmstadt for the ongoing cooperation on Silica–PMMA core–shell particles,
silica suspensions, and the phoXonic films.
Kenneth Kearns and Prof. Mark Ediger from University of Wisconsin must be
named thankfully for the cooperation on the stable organic glasses project, which
will be continued with Dr. Zahra Fakhraai.
For their theoretical support I would like to thank Dr. Rebecca Sainidou (Univ.
du Havre), Prof. Nikolaos Stefanou, and Georgios Gantzounis (both University of
Athens). Especially Rebecca was involved in many projects, and I want to thank
her not only for her calculations, but also for her patience in explaining me her
results.
I thank my office mates Akihiro Sato and Nikos Gomopoulos for many helpful
discussions, as well as Nikos for building the stretching machine.
I benefited from the great infrastructure at the Max Planck Institute for Polymer
Research. Therefore, I thank the people in the different service groups, the
mechanical shop, the electronics shop, and besides that the technical support by
Andreas Best, Gunnar Glasser and Maren Müller (SEM), Melanie Dröge and Petra
Räder (DSC), Dr. Konstantinos Mpoukouvalas (dielectric), and Uwe Rietzler
(AFM).
For the nice working atmosphere and the activities beyond the daily work I want
to thank all involved members of AK Butt, but also, in the same manner, the
people from the material research group (formerly known as AK Knoll).
This thesis was proof-read by Johannes Schmidt (MPI Golm), Daniel Szubrin
(University of Marburg), Martin Münzel (LMU Munich), and Justine Witosch
(MPI Martinsried). Thank you all.
Coming to the end of these acknowledgments, I want to thank the supervisor
of my diploma work, Prof. Wolf-Christian Pilgrim from Philipps-Universität
Marburg for introducing me into the field of inelastic scattering. Furthermore,
I thank many colleagues from all over the world that I met on several conferences
and who gave new impulses to me during many interesting discussions.
I want to thank my parents for all the support they gave me and all chances they
facilitated in my life.
And finally, I want to express my deepest thankfulness to the woman I love, who
gave me so much joy and happiness in the last years. Danke!
Mainz Tim Still

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 General Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Aims and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Basics and Brillouin Light Scattering . . . . . . . . . . . . . . . . . . . . . . . 9

2.1 Elastic Waves in Condensed Matter. . . . . . . . . . . . . . . . . . . . . . 9
2.1.1 Elasticity Theory Basics . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.2 Elastic Waves in Isotropic Media . . . . . . . . . . . . . . . . . . 14
2.1.3 Spherical Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Light Scattering Basics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Brillouin Light Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 BLS Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.2 BLS Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.3 Vibrational Modes of Individual Particles . . . . . . . . . . . . 32
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.1 Vertical Lifting Deposition and Colloidal Crystals. . . . . . . . . . . . 35
3.2 Melt Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 Polymer and Colloid Characterization Techniques . . . . . . . . . . . . 38
3.3.1 Photon Correlation Spectroscopy . . . . . . . . . . . . . . . . . . 38
3.3.2 Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.3 Differential Scanning Calorimetry . . . . . . . . . . . . . . . . . . 40
3.3.4 Density Gradient Column. . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.5 Wide Angle X-ray Scattering . . . . . . . . . . . . . . . . . . . . . 41
3.3.6 UV/VIS Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3.7 Gel Permeation Chromatography . . . . . . . . . . . . . . . . . . 42
3.4 Theoretical Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.1 Single-sphere Scattering Cross-section Calculations . . . . . 44
xiii
xiv Contents
3.4.2 The Plane Wave Method . . . . . . . . . . . . . . . . . . . . . . . . 45

3.4.3 The Multiple Scattering Method . . . . . . . . . . . . . . . . . . . 47
3.4.4 The Finite Element Method . . . . . . . . . . . . . . . . . . . . . . 51
References ......................... . . . . . . . . . . . . . . . . . . 51
4 The Vibrations of Individual Colloids . . . . . . . . . . . . . . . . . . . . . . . 53

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Elastic Vibrations in Homogeneous Polymer Colloids . . . . . . . . . 55
4.2.1 The ‘Music’ of the Spheres . . . . . . . . . . . . . . . . . . . . . . 55
4.2.2 The Influence of the Neighbors: Mixtures . . . . . . . . . . . . 61
4.2.3 The Influence of the Rigidity: Copolymers . . . . . . . . . . . 63
4.2.4 The Influence of the Wave Vector: Suspensions . . . . . . . . 69
4.3 Elastic Vibrations in Nanostructured Colloids . . . . . . . . . . . . . . . 71
4.3.1 The Influence of the Temperature: PS–SiO2 Core–Shell
Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 71
4.3.2 The Influence of the Components: SiO2-PMMA
Core–Shell Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.4 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.4.1 Homogeneous Polymer Colloids . . . . . . . . . . . . . . . . . . . 84
4.4.2 Homogeneous Silica Colloids . . . . . . . . . . . . . . . . . . . . . 85
4.4.3 PS–SiO2 Core–Shell Particles . . . . . . . . . . . . . . . . . . . . . 85
4.4.4 SiO2–PMMA Core–Shell Particles . . . . . . . . . . . . . . . . . 86
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5 Phononic Behavior of Colloidal Systems . . . . . . . . . . . . . . . . . . . . . 89

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.1.1 The Phononic Band Diagram . . . . . . . . . . . . . . . . . . . . . 91
5.1.2 The Effective Medium . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.1.3 Phononic Band Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.2 Effective Medium Velocity in Defect Doped Opals . . . . . . . . . . 100
5.3 Band Gaps in Polymer Opals and Disordered Systems . . . . . . . . 104
5.3.1 The Influence of the Order in Colloidal Systems . . . . . . 104
5.3.2 Influence of the Composition in Disordered
Colloidal Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.4 Band Gaps in SiO2 Colloidal Systems . . . . . . . . . . . . . . . . . . . 113
5.4.1 Phononic Behavior of Silica Suspensions . . . . . . . . . . . . 114
5.4.2 Silica–Poly(ethyl acrylate) Films (PhoXonics) . . . . . . . . 117
5.5 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.5.1 Melt Compressed Silica–Poly(ethyl acrylate) Films. . . . . 121
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6 Smaller than Colloids: Characterization of Stable Organic Glass. . 123

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.2 BLS Experiments on IMC . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Contents xv
6.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

6.3.1 Stable Glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.3.2 BLS Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
7.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
8 Appendix: Scattering Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . 135

8.1 Transmission Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
8.2 Reflection Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Curriculum Vitae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Abbreviations
2D Two-dimensional
3D Three-dimensional
a Angle incident laser/sample
ag Mark–Houwink parameter
b(T,S) (Isothermal, adiabatic) compressibility
Dielectric constant
Extinction coefficient
/ Glancing angle
u, w, v Scalar functions
C Central point of fcc BZ
g Viscosity
k Wave length
k Lamé coefficient
K Phonon periodicity
l Lamé coefficient
m Velocity of light in vacuum
h Scattering angle
. Mass density
r Poisson’s ratio
r Scattering cross section
r
^ Normalized scattering cross section
rik Stress tensor
s Correlation time
f Tandem FP angle
x Angular frequency
A Free energy
A Area under the curve
A Absorbance
Alml’m’ Structure constants
BLS Brillouin light scattering
nBA n-Butyl acrylate
xvii
xviii Abbreviations
BG Bragg gap
cF Coefficient of finesse
cl/t/eff (Longitudinal/transverse/effective) sound velocity
Cp/T Specific heat (at constant p/T)
Cik Components of the stiffness matrix
CCD Charge-coupled device
d Diameter
d1/2 FP mirror distances
D Diffusion coefficient
DOS Density of states
DSC Differential scanning calorimetry
DTA Differential thermo analysis
DVB Divenyl benzene
E Young’s modulus
E Electric field
E0 Field amplitude
EMT Effective medium theory
f Frequency
F Finesse
F Force
flm(r, h, /) Solution of the scalar Helmholtz equation
fcc Face centered cubic
FP Fabry–Pérot interferometer
FSR Free spectral range
G Green’s function
G Reciprocal space vector
G(q, s) Time–correlation function
g2(q, s) Second order autocorrelation
GPC Gel permeation chromatography
h, h Planck quantum (/2p)
hcp Hexagonal close packing
HG Hybridization gap
I (Scattering) intensity
I Unit tensor
IMC Indomethacin
kB Boltzmann’s constant
KPS Potassium persulfate
K Gordon–Taylor parameter
K Bulk modulus
Kg Mark–Houwink parameter
k(i/sc) (Incident/scattered) Wave vector
l Longitudinal
L Distinct point in reciprocal fcc lattice
L Shear modulus
l, m, n Three independent vectors
Abbreviations xix
LMS Layer multiple scattering method

M Distinct point in reciprocal fcc lattice
MMA Methyl methacrylate
MPIP Max Planck Institute for Polymer Research
MS Multiple scattering method
m Mass
me Electron mass
Mn Number averaged molecular weight
Mw Weight averaged molecular weight
n Refractive index
ni Unit vector
NaPSS Sodium sulfonated polystyrene
Nd:YAG Neodym doped yttrium aluminium garnet (laser)
OG Ordinary glass
p Pressure
p Momentum
PnBA Poly (n-butyl acrylate)
PCS Photon correlation spectroscopy
PDI Polydispersity
PS Polystyrene
PW Plane wave method
PMMA Poly (methyl methacrylate)
q Scattering wave vector
q Absolute value of q
qpara q Parallel to the sample plane
qperp q Perpendicular to the sample plane
Q Heat
r Radius
r Space vector
R Reflectivity
R Rigidity
R Position vector
R Absolute value of R
Rh Hydrodynamic radius
Rl(k, r) Spherical Bessel functions
Rn Bloch’s vector
RS Raman scattering
S Entropy
SCC Supercooled liquid
SEM Scanning electron microscope
SG Stable glass
T T-Matrix
t Time
t Transverse
T Temperature
xx Abbreviations
T Transmittance
TEM Transmission electron microscope
TNB aab-1,3,5-tris-Naphthylbenzene
u Displacement vector
uik Strain tensor
UV/VIS Ultraviolet/visible light (spectroscopy)
V (Scattering) volume
v Velocity
w Work
w Relative width of a band gap
wc Degree of crystallization
X Distinct point in reciprocal fcc lattice
x(n, l) Eigenmode constant
Ylm Spherical harmonics
Chapter 1
Introduction
1.1 General Introduction
One of the most excitatory and technically promising new field in physics in the
last decades was for sure the emerging possibility to control and manipulate the
flow of light by engineering of macroscopic media with periodic dielectric func-
tion, i.e., with periodic variation of the refractive index in one, two, or three
dimensions (Fig. 1.1). Since the first realization of such photonic crystals, an
abundance of publications on this topic have been released, including several new
high impact journals. Beside some quantum mechanical effects, the striking
capability of a photonic crystal is its aptitude to prevent light from propagating in
certain directions with specified frequencies, while they can operate as omni-
directional reflectors of light, independent from angle and polarization [1]. If the
propagation of electromagnetic waves is forbidden at certain frequencies, one
speaks about a photonic band gap.
The appearance of band gaps for distinct wavelengths in materials with periodic
changed dielectric constant can be rationalized as an interference effect. The
light is partially reflected at each layer interface. The multiple reflection interfere,
and if the periodicity spacing is commensurate to the wavelength of the incident
light, the destructive interference eliminates the forward propagation of the elec-
tromagnetic wave. In the simplest case of a one-dimensional (1D) photonic crystal,
i.e., with periodic modulation of only in z-direction, with normal incident light,
the gap occurs when the wavelength of the light kl is twice the crystals spatial
period a, kl = 2a. In the concept of the reciprocal space that is commonly used in
solid state physics, this belongs to a wave vector k = 2p/kl (in this case k is
identical to its component in z-direction) at the edge of the first Brillouin zone, i.e.,
k = p/a. The problem was first treated by Lord Rayleigh (1887), and such gap is
usually called Bragg gap. It can be shown that the width of such a photonic gap
depends on the dielectric contrast between the two components.
The concepts presented for the molding of electromagnetic waves can, in
principle, be transferred to mechanical waves, i.e., sound waves. A material with
T. Still, High Frequency Acoustics in Colloid-Based Meso- and Nanostructures 1

by Spontaneous Brillouin Light Scattering, Springer Theses,
DOI: 10.1007/978-3-642-13483-8_1, Ó Springer-Verlag Berlin Heidelberg 2010
2 1 Introduction
Fig. 1.1 Photonic or phononic crystals with periodicity in one, two, or three directions. In
photonic crystals, white and black segments have different refractive index, while in phononic
crystals the components must have different elastic moduli or densities
periodic modulation of the elastic properties, i.e., of the elastic moduli or the
density, is called a phononic crystal after the quantized mode of vibration in a rigid
crystal lattice—the phonon (in analogy to the photon in the photonic crystals)
[2, 3]. However, the theoretical treatment of the propagation of mechanical waves
is more complicated. While the light propagation can be fully described by a wave
with distinct polarization and velocity, which depends on the refractive index
of the passed medium, a mechanical wave can, in principle, propagate in two
different ways.
Both mechanisms are schematically shown in Fig. 1.2. The longitudinal sound
wave on the left side shows displacement along the propagation direction, while in
the transverse sound wave the displacement is perpendicular to the propagation
direction symbolized by the phonon wave vector k. Both waves propagate with
different sound velocities cl (longitudinal) or ct (transverse), respectively. The
relevant propagation mechanisms depend on the type of the material. In solid
materials, both kind of waves can propagate, while in liquid or gaseous media only
longitudinal waves are supported. Hence, the different mechanical waves are often
distinguished between elastic waves in solids and acoustic waves in fluids.
However, in this thesis the terms elastic wave, acoustic wave, or sound wave will
be used synonymously for all kinds of mechanical waves.
(a) (b)
Fig. 1.2 Propagation mechanisms of mechanical waves: a longitudinal wave with displacement
along the propagation direction (given by the direction of the wave vector k); b transverse sound
wave with displacement perpendicular to the propagation direction
1.1 General Introduction 3
On an atomic scale, the concept of the phonon is used to describe for example
the heat capacity of solids. The frequencies of the allowed mechanical waves are
quantized as a function of the distance ‘ between neighboring atoms. The atoms
(the black points in Fig. 1.2) are displaced, and the displacement of each atom is
described by the displacement vector u(r, t) as a function of position and time.
However, in this thesis the interest lays on much lower frequencies with phonon
wavelength K ‘, thus the allowed mechanical frequencies can be regarded as
continuous. Figure 1.2 holds also in this case, but now the points represent volume
elements containing many atoms. It should be noted that the spatial displacement
of volume elements as a function of time leads to areas of instantaneously
increased or decreased pressure (or local density), thus mechanical waves can be
regarded as pressure waves. This picture also rationalizes why there is no prop-
agation of transverse sound waves in fluids since fluids do not support shear (aside
from very viscous liquids).
Phononic crystals can be designed by a one-, two-, or three-dimensional peri-
odicity of their elastic properties in analogy to their electromagnetic counterparts
(Fig. 1.1). The geometry of the periodic structure has strong influence on its
mechanical properties. One of the first examples of a two-dimensional phononic
crystal is a statue in Madrid, created by the artist E. Sempere [4]. The piece of art
consist of a large number of hollow stainless-steel cylinders with diameter of
2.9 cm in a simple cubic arrangement in a a = 10 cm unit cell, the whole statue
has a diameter of 4 m. A simple transmission experiment with a sound generator
and detector in the range of audible frequencies (1–5 kHz) has shown the existence
of phononic band gaps with strong sound attenuation depending on the orientation
of source and receiver relative to the statue, i.e., as a function of the crystallo-
graphic direction.
Like in the case of photonic crystals, the Bragg gap appears at the edge of the
first Brillouin zone, i.e., the phonon wavelength K is twice the lattice parameter a,
K = 2a, and the corresponding wave vector is again k = p/a. The frequency is
given by 2pf = cl/K. Since typical sound velocities are between 102 ms-1 (air)
and 105 ms-1 (condensed matter), attenuation in the audible range requires peri-
odicities in the centimeter to meter range, like in the example of the statue.
Structures that would attenuate seismic waves would have to be constructed even
in kilometer scales, the Bragg frequency scales with 1/a. According to this,
smaller, technically handier structures with spacing in the millimeter, micrometer,
or even nanometer-scale correspond to frequencies in the MHz to THz scale.
Most of the realized phononic band gap systems are restricted to sonic and
ultrasonic crystals with macroscopic periodicity, e.g., some millimeter-sized
crystals assembled manually, which could be probed by simple acoustic trans-
mission experiments in the commensurate frequency ranges [5–7]. Only recently,
Cheng et al. [8] could show the realization of a hypersonic phononic crystal based
on the self-assembly of sub-micron colloidal polymer crystals [9].
In fact, such mesoscopic structures have some advantages. They can be easily
prepared by self-assembly methods or by interference lithography [3] on a (relative
to the lattice spacing) large scale, they can show photo-thermal or photo-acoustic
4 1 Introduction
effects, and their size is also commensurate to the wavelength of visible light.
Therefore, it is possible to create a mesoscopic artificial crystal with periodic
variation of refractive index and mechanical properties, which would act as a
photonic and phononic crystal simultaneously. Sometimes, such ‘blind and deaf’
systems are called phoXonic crystals [10, 11]. Due to their size, colloidal crystals
from mesoscopic spheres show nice colorful reflections of light depending on the
scattering angle relative to their crystallographic direction as it can be found in
natural opals. Therefore, such systems are often (also in the present thesis) called
artificial opals. Figure 1.3 shows three photographs of a polished natural opal, an
artificial opal, which consists of dense packed silica spheres (d = 170 nm) in a
organic liquid matrix, and a layer of polystyrene spheres (d = 550 nm) on a glass
substrate.
These artificial opals are examples for colloid particle systems. In general, a
colloid is a particle or droplet of typical size between 1 nm and 10 lm that is
dispersed in another gaseous, liquid, or solid medium, e.g., milk (fat droplets in
water), fume (solid particles in gas), blood (solid particles in liquid), etc. In this
thesis the word colloid is used in a less general way; here it means in particular the
well defined colloidal solid particles, which are fabricated by the methods of
material chemistry. Systems of these particles in a matrix material are referred to
as colloidal systems.
Such colloidal systems have many advantages. Due to the onward development
of colloidal and material science, nowadays it became possible to control a wide
range of properties during the synthesis of colloids. Monodisperse spherical par-
ticles can be produced in a size range from few tens of nanometer to several
microns, utilizing a wide range of materials like metals, oxides (e.g., silica), or
Fig. 1.3 Natural and artificial opals: a Natural opal (Australia, Museum Idar-Oberstein,
Germany); b dense packed silica spheres (d = 219 nm, DKI Darmstadt) in organic liquid; c PS
spheres (d = 550 nm, M. Retsch, MPIP) on glass substrate
1.1 General Introduction 5
(organic) polymers. However, colloid science is not generally limited to spherical

particle shape. For metals, and with limitations for the form stability also for
oxides, even smaller nanoparticles can be created. Mesoscopic colloids can be
additionally structured in the nanoscale to manipulate or improve their physical
and chemical properties, leading, e.g., to spherical hybrid core–shell particles
investigated also in this thesis [12]. Another already mentioned advantage is the
possibility to realize well ordered single or multi-component systems by
self-assembly (colloidal crystals). In Fig. 3.1 (in Chap. 3) several techniques to
achieve colloidal crystals will be presented, including sedimentation [13],
electrodeposition [14], centrifugation, filtering, vertical deposition [15, 16], and
compression molding [17]. Especially the last two techniques play a role in this
thesis and will be introduced in more detail in the Sects. 3.1 and 3.2 in Chap. 3.
Since colloids exhibit outstanding physical and chemical properties (e.g., large
surface, optical properties, solubility, etc.) and tunability (size, shape, softness,
biodegradability, etc.) [18–20], this class of materials found numerous technical
applications, e.g., in suspensions as pigment or sunscreen [21], for drug-delivery
applications [22], or to improve the macroscopic mechanical properties of solids or
polymers [23].
This thesis is devoted to the investigation of the acoustic, i.e., phononic,
properties of colloidal particles and systems, which appear heterogeneous com-
pared to the sound wavelength. Hence, rich unprecedented behavior is expected
due to the large number of parameters involved and the sensitivity of elastic wave
propagation to boundary conditions. As already pointed out, the characteristic
frequencies scale with the inverse size of the observed systems, for mesoscopic
colloids hypersound, i.e., GHz frequencies, become relevant. Such high frequen-
cies are not accessible anymore by ‘simple’ sound generation and subsequent
detection with membranes (audible sound) or piezo crystals (ultrasound). In fact,
when measuring the sound propagating in a material, it is not absolutely necessary
to generate the hypersound artificially, since hypersonic phonons are thermally
excited at temperatures far away from 0 K [24]. These phonons can be probed by
inelastic light scattering, i.e., spontaneous Brillouin light scattering (BLS) in the
GHz-range (related to the mesoscopic length scale) and Raman scattering in the
THz-range (for the nanoscopic length scale) [25, 26]. Another possibility is pump-
probe spectroscopy, where a strong laser pulse impinges on a sample and a second
probe laser is used to measure the reflectivity of the sample as a function of time
[27, 28]. After Fourier transformation that approach also delivers information
about the frequencies of sound in the probed material, however, after an induced
perturbation and not spontaneously.
In this thesis, BLS is utilized to probe phonon propagation (and localization) in
micro- and nanostructures. The main principle of this unique non-destructive and
non-contact optical technique is the interaction of photon and phonon. In fact, the
photons of the probing single-frequency laser light are scattered inelastically on
the phonons in the probed material propagating along a selected direction, which is
given by the scattering wavevector q that is a function of the scattering angle and
the sample’s refractive index. In homogeneous systems the interacting phonon is
6 1 Introduction
the one whose phonon wavevector k is identical with q as a consequence of

momentum conservation. Therefore, the BLS spectrum at a given q consists of a
Doppler shift at angular frequencies x = ±cl,tk for longitudinal (l) or transverse
(t) sound velocity c. With other words, we measure the frequency shift of the
probing light due to the creation or annihilation of a phonon as a function of the
momentum (the magnitude of the wave vector) leading to the dispersion relation
x(q). The achieved frequency shifts in the GHz-range are tiny compared to the
frequency of the laser light (1014 Hz), the required high resolution is attained by a
multi-pass tandem Fabry–Pérot interferometer [29].
Of course, the use of BLS to investigate the mechanical properties of matter
is not limited to colloidal systems. The last chapter of this thesis shows how
BLS can be used to investigate the mechanics and kinetics of so-called stable
organic glasses, i.e., organic glasses that are created by very slow vapor
deposition on a temperate substrate [30, 31]. By doing so, the molecules have
more time to rearrange in an energetically more favorable way, leading to elastic
moduli comparable to normal glasses that are aged for at least several hundred
years.
1.2 Aims and Motivation
Colloids are very promising systems in many terms. They are auspicious candidates
to design hypersonic phononic systems that may be the basis of elaborated devices
that deal with the concurrent interaction of phonons and light (functional phoXonic
materials). However, so far there is no large knowledge about the details of the
nano- and mesomechanical behavior of individual colloids and the manipulation of
sound propagation in colloidal systems beyond the simple Bragg gap [8].
In this thesis a systematic approach is presented that captures the influence of
different parameters on the mechanical waves localized in spherical (as the most
accessible and theoretically best to capture model system), partially nanostruc-
tured, mesoscopic colloids as well as the propagation of sound waves in systems
composed of such particles [12, 32]. The first realization of a so-called hybrid-
ization gap in polymer based colloids will be presented, which originates from the
interaction between individual and collective colloid mechanical properties,
leading to crystalline as well as amorphous omnidirectional hypersonic band gap
systems [33].
These developments, however, must not be regarded as the end of the basic
experimental investigation of such systems. It will be shown that for silica based
systems with increasing mechanical contrast new effects occur, which will demand
further strong efforts in the theoretical and experimental conquest of this young
and promising field. Especially in consideration of the numerous appearing topics
related to photo-acoustics [34], thermo-acoustics [35], or nano- and microme-
chanics [36, 37], it seems worth to investigate colloidal model systems en detail
and the intimate relation of the heat conductivity in dielectric materials on phonons
1.2 Aims and Motivation 7
can have impact on a directional heat flow in the future. This work may be a
humble contribution to this exciting development.
1.3 Outline
The structure of this thesis is as follows:

In this chapter I introduce the theoretical and practical basics of the propagation
of elastic waves and of light scattering, in particular of Brillouin light scattering
and its applicability to colloidal particles and colloid based systems.
The second chapter is about other methods utilized in this thesis, including the
theoretical calculations that are needed to quantitatively describe the experimental
findings. If these first two chapters might appear to extend some few details
beyond the absolutely necessary length, this is caused by the aim to present the
important results in a common notation, which is not the case in the underlaying
literature.
Chapter 4 describes the work done on the vibrations of individual colloids that
can be investigated in multiple scattering or thick transparent samples. Herein, the
influences of packing in a sample consisting of many spheres and that of the sphere
material is reported as well as the discussion of general theoretical problems and
their experimental answer concerning the selection rules for eigenvibrations in
BLS and the light scattering intensity as a function of q.
The next chapter proceeds to the collective phononic behavior, i.e., the prop-
agation of sound waves, in colloidal systems embedded in a solid or liquid matrix.
Starting from a short description of the state-of-the-art, the experimentally found
effective medium velocities are discussed and a new kind of phononic band gap,
the hybridization gap, is presented. It is demonstrated by the investigation of silica
based colloidal systems that there seem to be additional mechanisms to manipulate
the phononic band diagram.
Chapter 6 is the last experimental chapter and deviates from the scope on
colloids in the other parts of the thesis since it is about the characterization of
stable organic glasses.
Finally, this thesis concludes with main results and outlook in the concluding
remarks (Chap. 7).
References
1. Joannopoulos JD, Johnson SG, Winn JN, Meade RD (2008) Photonic crystals: molding the
flow of light, 2nd edn. Princeton University Press, Princeton
2. Economou EN, Zdetsis A (1989) Phys Rev B 40:1334
3. Maldovan M, Thomas EL (2008) Periodic materials and interference lithography. Wiley-
VCH, New York
4. Martinez-Salazar R et al (1995) Nature 378:241
8 1 Introduction
5. Montero de Espinosa FR, Jiménez E, Torres M (1998) Phys Rev Lett 80:1208
6. Liu ZY et al (2000) Science 289:1734
7. Vasseur JO et al (2001) Phys Rev Lett 86:3012
8. Cheng W et al (2006) Nat Mater 5:830
9. Thomas EL, Gorishnyy T, Maldovan M (2006) Nat Mater 5:773
10. Gorishnyy T, Maldovan M, Ullal C, Thomas E (2005) Phys World 18:24
11. Bernal MP, Roussey M, Baida F, Benchabane S, Khelif A, Laude V (2008) In: Ferraro P,
Grilli S, De Natale P (eds) Ferroelectric crystals for photonic applications, vol 307. Springer,
Heidelberg, pp 307–336
12. Still T et al (2008) Nano Lett 8:3194
13. Davis KE, Russel WB, Glantschnig WJ (1989) Science 245:507
14. Trau M, Saville DA, Aksay IA (1996) Science 272:706
15. Jiang P, Bertone JF, Hwang KS, Colvin VL (1999) Chem Mater 11:2132
16. Gu ZZ, Fujishima A, Sato O (2002) Chem Mater 14:760
17. Ruhl T, Spahn P, Hellmann GP (2003) Polymer 44:7625
18. Lyklema J (1991) Fundamentals of interface and colloid science. Academic Press, London
19. Lyklema J (2005) Fundamentals of interface and colloid science IV. Elsevier, London
20. Lyklema J (2005) Fundamentals of interface and colloid science V. Elsevier, Amsterdam
21. Magdassi S, Bassa A, Vinetsky Y, Kamyshny A (2003) Chem Mater 15:2208
22. Müller RH, Mäder K, Gohla S (2000) Eur J Pharm Biopharm 50:161
23. Pichot F, Pitts JR, Gregg BA (2000) Langmuir 16:5626
24. Meschede D (2006) Gerthsen Physik, 23rd edn. Springer, Heidelberg
25. Brillouin L (1922) Ann Phys (Paris) 17:88
26. Raman VV, Krishnan KS (1928) Nature 121:501
27. Mazurenko DA et al (2007) Phys Rev B 75:161102
28. Akimov AV et al (2008) Phys Rev Lett 101:033902
29. Mock R, Hillebrands B, Sandercock JR (1987) J Phys E Sci Instrum 20:656
30. Dyre JC, Olsen NB, Christensen T (1996) Phys Rev B 53:2171
31. Swallen SF et al (2007) Science 315:353
32. Still T et al (2008) J Phys Condens Matter 20:404203
33. Still T et al (2008) Phys Rev Lett 100:194301
34. Kippenberg TJ, Vahala KJ (2008) Science 321:1172
35. Damen EPN, Arts AFM, de Wijn HW (1995) Phys Rev Lett 74:4249
36. Cheng W et al (2008) Nano Lett 8:1423
37. Schliesser A et al (2008) New J Phys 10:095015
Chapter 2
Basics and Brillouin Light Scattering
In this chapter, first the basic expressions of elasticity will be introduced, espe-
cially with regard to the propagation of elastic waves in isotropic media and their
description as spherical waves. Then the general principles of light scattering will
be presented and applied to the special case of Brillouin light scattering, including
the discussion of its technical realization.
2.1 Elastic Waves in Condensed Matter
2.1.1 Elasticity Theory Basics
In this part, fundamental concepts of the theory of elasticity are briefly introduced,
following mostly the notation of the excellent textbook of Landau and Lifschitz [1].
Essential terms are the strain and stress tensors. To introduce them, we first
define the displacement vector u that shifted a point P that can be found by
following the vector r (with the three components x1 = x, x2 = y, and x3 = z in a
Cartesian system) from the origin of the coordinate system to the point P0 with
coordinates given by the vector r0 ; i.e.,
u ¼ r r0 : ð2:1Þ
The distance between any two infinitesimally adjacent points dl is given by
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffi
dl ¼ dx21 þ dx22 þ dx23 ¼ dx2i ; ð2:2Þ
using the Einstein summation convention behind the second equal. After a
deformation it becomes

10 2 Basics and Brillouin Light Scattering
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi
dl0 ¼ dx02 02
1 þ dx2 þ dx3 ¼
02 dx02 i ð2:3Þ
Therefore, it is possible to write

2 oui oui oui
dl02 ¼ dx02 2
i ¼ ðdxi þ dui Þ ¼ dl þ 2 dxi dxk þ dxk dxl ð2:4Þ
oxk oxk oxl
oui oui ouk
substituting dui ¼ dxk in the last step. With dxi dxk ¼ dxi dxk and swap of
oxk oxk oxi
i and l in the last term, finally we get
dl02 ¼ dl2 þ 2uik dxi dxk : ð2:5Þ

The strain tensor uik is therein defined as

1 oui ouk oul oul 1 oui ouk
uik ¼ þ þ þ : ð2:6Þ
2 oxk oxi oxk oxi 2 oxk oxi
The last approximation is valid if second order terms can be neglected. Obviously,
uik is a symmetric tensor. Each symmetric tensor can be diagonalized in any point.
With the diagonal elements u(1), u(2), u(3) the strain in any point can be written as
the sum of three independent terms, which give the strain in three orthogonal main
directions.
dl02 ¼ ð1 þ 2uð1Þ Þdx21 þ ð1 þ 2uð2Þ Þdx22 þ ð1 þ 2uð3Þ Þdx23 : ð2:7Þ
For u(i) 1 and if higher order terms are neglected, the relative change of
elongation becomes
dx0i dxi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ 1 þ 2uðiÞ 1 uðiÞ : ð2:8Þ
dxi
With that approximation we can write the relative volume change of an infinite-
simal small volume element dV ! dV 0 as the sum of the diagonal elements of the
strain tensor [1].
dV 0 dV
¼ uii : ð2:9Þ
dV
The resulting force F on any partial volume of an elastic Rbody can be written as
the integral over the forces on any element of the volume FdV: Because of the
identity of actio and reactio for all forces between any two points within the volume
the resulting force
R can also be written as the sum over the integrals of the three
components Fi, Fi dV, which can be translated into an integral over the surface:
Z Z I
orik
Fi dV ¼ dV ¼ rik dfk : ð2:10Þ
oxk
2.1 Elastic Waves in Condensed Matter 11
Here, Fi is expressed as divergence of a second rank tensor, the stress tensor rik:
orik
Fi ¼ ð2:11Þ
oxk
In Eq. 2.10, dfi are the components of a vector that is always oriented in the
direction normal to the surfaces. The nature of the stress tensor is visualized in
Fig. 2.1. When placing the surfaces of the volume element in the principal plane of
the coordinate system (xy, yz, or xz), the component rab of the stress tensor equals
to the a-component of the force that is normal to that xb-axis.
The work dw that is achieved to perform a deformation is
dw ¼ rik duik : ð2:12Þ
For reversible, elastic deformations it follows for the free energy A (A = U - TS,
dU = TdS - dw, S: entropy) that

oA
rik ¼ : ð2:13Þ
ouik T
It can be shown that the free energy after an elastic isothermal deformation of an
isotropic body can be expressed after series expansion and neglecting higher order
terms as
k
A ¼ A0 þ u2ii þ lu2ik ; ð2:14Þ
2
introducing the Lamé coefficients k and l. Note that k and l are force con-
stants. Since it was already shown that the volume change during a deformation
is expressed by the sum of the diagonal elements uii, it is obvious that, if the
term containing the uii becomes zero, the last term expresses a pure shear
deformation. Therefore, l is also called the shear modulus, sometimes denoted
by G.
Fig. 2.1 Illustration of x

the stress tensor for the
rax-components working on a
surface normal to the x-axis.
The picture is analogously
valid for the other six com- xx
ponents when applied to the
surfaces normal to the y- and
z-axis
zx z
y
yx
By writing the following identity

1 1
uik ¼ uik dik ull þ dik ull ð2:15Þ
3 3
with uik = constdik it follows directly that every deformation can be written as a
sum of a pure shear deformation and a homogeneous dilatation. The term in
brackets is surely a pure shear deformation, because the sum of the diagonal
elements vanishes (dii = 3) and the other term is related to the homogeneous
dilatation.
For a perfectly elastic body, Hook’s law can be generalized to state that each
component of the stress tensor is linearly related to each component of the strain
tensor:
rik ¼ ciklm ulm : ð2:16Þ

Here, ciklm is the fourth-rank stiffness tensor with 34 = 81 components. Taking
into account the symmetry of the strain and stress tensors, it is possible to reduce
the number of coefficients to 62 = 36. However, these 36 components are not
independent. In the general case the following symmetry relations are valid:
ciklm ¼ ckilm ¼ cikml ¼ ckiml : ð2:17Þ
That is in an anisotropic body there are at most 21 independent components,

which are usually noted in a matrix notation, which uses the itemized identities
to abbreviate the tensor notation and to replace the tensor by a 6 9 6-matrix
with only two indices per component. The following rules are adopted for the
indices:
Tensor notation 11 22 33 23, 32 13, 31 12, 21

Matrix notation 1 2 3 4 5 6
With that notation one can rewrite Eq. 2.16 as

2 3 2 32 3
r1 C11 C12 C13 C14 C15 C16 u1
6 r2 7 6 C12 C22 C23 C24 C25 C26 7 6 u2 7
6 7 6 76 7
6 r3 7 6 C13 C23 C33 C34 C35 C36 7 6 7
6 7¼6 7 6 u3 7 : ð2:18Þ
6 r4 7 6 C14 C24 C34 C44 C45 7
C46 76 7
6 7 6 6 u4 7
4 r5 5 4 C15 C25 C35 C45 C55 C56 5 4 u5 5
r6 C16 C26 C36 C46 C56 C66 u6
While the discussion of the stiffness matrix for different crystal symmetries is
done in specialized treatments, here only the isotropic case, which is a good
approximation for many cases, including also most polymer systems, should be
discussed in the rest of this section. For an isotropic body, it can be shown that
due to further symmetry considerations the stiffness matrix in Eq. 2.16 has the
following form [2]:
2 3 2 32 3
r1 C11 C12 C12 0 0 0 u1
6 r2 7 6 C12 C11 C12 0 0 0 7 6 u2 7
6 7 6 76 7
6 r3 7 6 C12 C12 C11 0 0 0 7 6 7
6 7¼6 7 6 u3 7 : ð2:19Þ
6 r4 7 6 0 0 0 C 0 0 7 6 7
6 7 6 44 7 6 u4 7
4 r5 5 4 0 0 0 0 C44 0 5 4 u5 5
r6 0 0 0 0 0 C44 u6
Because the three remaining constants are related by the following relation
C11 ¼ C12 þ 2C44 ; ð2:20Þ

there are, indeed, only two independent constants remaining. These two constants
can be identified with the Lamé coefficients
k ¼ C12 and l ¼ C44 ; ð2:21Þ

while C11 is also called the longitudinal modulus. Beside the shear modulus
G = l, some other elastic parameters are often encountered, namely the bulk
modulus, the Young’s modulus, or the Poisson’s ratio.
The bulk modulus K measures a substance’s resistance to uniform compres-
sions. It is defined as the ratio of the hydrostatic pressure to the fractional volume
change,
op 2
K¼ ¼ k þ l: ð2:22Þ
ouii 3
K should not be confused with the longitudinal bulk modulus
M ¼ C11 : ð2:23Þ
The Young’s modulus E, also known as modulus of elasticity or tensile mod-
ulus, is a measure of the stiffness of an isotropic elastic material. It is given by the
ratio of longitudinal stress (which has units of pressure) and the dimensionless
longitudinal strain,
r11 lð3k þ 2lÞ
E¼ ¼ : ð2:24Þ
u11 kþl
The ratio of the lateral strain to the longitudinal strain defines the Poisson’s
ratio,
u22 k
r¼ ¼ : ð2:25Þ
u11 2ðk þ lÞ
2.1.2 Elastic Waves in Isotropic Media
When a body is deformed, the deformation usually goes along with changes in the
temperature. However, the heat transport is normally much slower compared to
periods of vibrations in the body. Therefore, one can regard the movements as
(quasi)adiabatic. It can be shown that in the adiabatic case the values for the Young’s
modulus and the Poisson’s ratio, given above for the isothermal case, change into
Ta2 Ta2
Eadiabatic ¼ E þ E2 and radiabatic ¼ r þ ð1 þ rÞE ; ð2:26Þ
9Cp 9Cp
where a ¼ ðoV=oTÞp and Cp is the specific heat at constant pressure [1]. In the
following, r and E mean their adiabatic values.
The general equation of movement can be written as
o2 ui orik
. ¼ ð2:27Þ
ot2 oxk
with mass density .: For the isotropic elastic medium the equation of movement
becomes
o2 ui E E
. 2
¼ r2 u þ : ð2:28Þ
ot 2ð1 þ rÞ 2ð1 þ rÞð1 2rÞgrad divu
Regarding a plane elastic wave in x-direction in an infinite medium, i.e.,
the deformation u depends only on the x-coordinate (all derivatives with respect
to y and z become zero), the components of the vector u become
o2 ux 1 o2 ux o2 uy 1 o2 uy o2 uz 1 o2 uz
¼ 0; ¼ 0; ¼ 0: ð2:29Þ
ox2 c2l ot2 ox2 c2t ot2 ox2 c2t ot2
The Eq. 2.29 are one-dimensional wave equations, cl and ct are their velocities
of propagation. Obviously, the propagation in x-direction is different from that in
the other directions. If the displacement ux lies in the direction of propagation of the
wave, the wave is called a longitudinal wave with longitudinal sound velocity cl:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffi
Eð1 rÞ k þ 2l C11
cl ¼ ¼ ¼ : ð2:30Þ
.ð1 þ rÞð1 2rÞ . .
In the other directions, the displacement (uy, uz) lies in a plane normal to the
direction of propagation of the wave. Such a wave is called a transverse wave with
transverse sound velocity ct:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffi sffiffiffiffiffiffiffi
E l C44
ct ¼ ¼ ¼ : ð2:31Þ
2.ð1 þ rÞ . .
Accordingly, from Eqs. 2.21 and 2.23 it is clear that we can express longitudinal
and shear modulus simply by
M ¼ .c2l ð2:32Þ
and
G ¼ .c2t ; ð2:33Þ
respectively, while the bulk modulus K depends on the longitudinal as well as on

the transverse sound velocity, since after Eqs. 2.20 and 2.22 it becomes

2 4 4
K ¼ M 2G þ G ¼ M G ¼ . c2l c2t : ð2:34Þ
3 3 3
The longitudinal sound velocity is always larger than the transverse sound
velocity. For r B 0.5,
pffiffiffi
c l [ c t 2: ð2:35Þ
One should note that the transverse wave has two possible polarizations, which
are orthogonal to each other. Furthermore, it is notable that longitudinal waves
involve changes in the volume of the medium, i.e., dilatation or compression of a
local volume element, while the transverse waves cause no volume change.
2.1.3 Spherical Waves
Because this thesis treats in many parts experiments on colloids acting as spherical
scatterers, it is meaningful to introduce briefly the principle solution of the elastic
wave equation for an isotropic medium. The equations presented here will be
required again in the discussion of the vibrational eigenmodes of spherical
particles.
In general, for a harmonic elastic wave of angular frequency x, the displace-
ment vector u can be written as
uðr; tÞ ¼ <½uðrÞ expðixt: ð2:36Þ
Using that one can write the general equation of motion in the following time-
independent form [3]:
ðk þ 2lÞrðr uÞ lr r u þ .x2 u ¼ 0: ð2:37Þ

In a spherical coordinate system, the displacement vector can be written as the
sum of three vectors u ¼ l þ m þ n such that Eq. 2.37 can be broken into three
independent vector Helmholtz equations [4]
ðr2 þ kl2 Þl ¼ 0; ðr2 þ kt2 Þm ¼ 0; ðr2 þ kt2 Þn ¼ 0; ð2:38Þ
where l represents the displacement associated with longitudinal wave and n and m
represent the transverse displacements, which are orthogonal to each other. These
three vectors can be related to scalar functions u; w; and v that are solutions of a
scalar Helmholtz equation (r2 + k2)f = 0:
1
l¼ ru; ð2:39Þ
kl
m ¼ r rw; ð2:40Þ
1
n¼ r r rv: ð2:41Þ
kt
The transverse displacement vectors m and n are herein expressed as the product
of the constant position vector r and the scalar functions mentioned above [4]. In
polar coordinates the solution of the scalar Helmholtz equation is known. It is
flm ðr; h; /Þ ¼ Rl ðkrÞYlm ðh; /Þ: ð2:42Þ
Rl(kr) are n-th order spherical Bessel functions, which represent the radial dis-
placement. Ylm(h, /) are the n-th order spherical harmonics (Legendre functions)
with l = 0, 1, 2, 3, ... and m being an integer -l B m B +l. Knowing this,
the vector solutions of Eq. 2.38 can be written as
1
llm ðR; kl Þ ¼ r½Rl ðkl rÞYlm ð^r Þ; l ¼ 0; 1; 2; 3; . . .; ð2:43Þ
kl
mlm ðR; kt Þ ¼ r ½rRl ðkt rÞYlm ð^r Þ; l ¼ 1; 2; 3; . . .; ð2:44Þ
1
nlm ðR; kt Þ ¼ r r ½rRl ðkt rÞYlm ð^r Þ; l ¼ 1; 2; 3; . . . ð2:45Þ
kt
2.2 Light Scattering Basics
This section is a brief introduction into the basic principles of light scattering.
While the derivation of scattering theory on the basis of quantum field theory is
possible, herein the scattering medium as well as the light are treated classically,
leading to practically the same results within the scope of this thesis. Of course
there are light scattering effects, as for example in the well-known Raman scat-
tering technique, which deals with rotational and vibrational transitions of single
atoms or molecules, i.e., effects in a quantum length scale, that must be treated
(at least partially) quantum mechanically. However, this thesis deals with the
investigation of phonons by Brillouin light scattering (Sect. 2.3) in condensed
2.2 Light Scattering Basics 17
matter. Thus the investigated phonons are classical waves with wavelengths in the
order of some nanometer up to micrometer, it is fully justified to apply a classic
theory.
The classical theory of light scattering in dense media developed by Einstein [5]
and Smoluchowski considers the sample as divided into small volume elements
large enough to contain many molecules (i.e., following classical physics), but
of linear dimension small compared to the wavelength of light. An incident
light wave induces a dipole moment in each volume element, which becomes
the source of scattered radiation. Provided that the induced polarization is
constant through the medium, the net scattered radiation in all directions but the
forward will be zero due to destructive interference, because the wavelets
scattered from each subregion differ only by a phase factor that depends on the
relative position of the small volumes. Therefore, neglecting small surface
effects, it is possible to pair each small volume with another small volume
whose scattered field is identical in amplitude but opposite in phase, thus they
cancel out [4, 6, 7].
However, in real media there will always be small random fluctuations in the
local dielectric constant due to the thermal motion of the atoms and molecules
in the sample. Because these fluctuations should be uncorrelated from one
volume element to the next, these regions are optically different, and therefore
also the amplitudes of the scattered light are uncorrelated. That means that now
light is also scattered in other directions than forward due to only partial
interference.
The local dielectric constant at a point at position r and time t; ðr; tÞ; is gen-
erally described by the dielectric constant fluctuation tensor dðr; tÞ: It describes
the relation between the local and the average dielectric constant 0 ;
ðr; tÞ ¼ 0 I þ dðr; tÞ: ð2:46Þ
I is the second-rank unit tensor.

When the incident light is a plane wave with field amplitude E0, angular
frequency xi, and incident propagation vector ki ; the incident electric field
can be written as
Ei ðr; tÞ ¼ ni E0 exp iðki r xi tÞ ð2:47Þ
with ni being the unit vector in the direction of the incident field.
At large distance R from the scattering volume, the scattered electric field
Es ðR; tÞ can be obtained from the fact that the Maxwell equations must hold for the
total electric field, the incident electric field and as well for the scattered electric
field, where the total field is just the sum of the two others:
E ¼ Ei þ Es : ð2:48Þ
Indeed, the same should be valid for the electric displacement field D and the
magnetizing field H: The solution of the Maxwell equations is lengthy and
performed elsewhere [7]. In the end it follows for the component of the scattered
electric field Es(R, t) that
Z
E0
Es ðR; tÞ ¼ exp iks R d3 r
4pR0 V ð2:49Þ
exp iðqr xi t½ns ½ks ðks ðdðr; tÞni Þ;
where ns is the polarization, ks is the propagation vector and xs is the frequency of

the scattered plane wave field that reaches the detector, while V indicates that the
integral is over the whole scattering volume. The scattering wave vector q is
defined as the difference between incident and scattered propagation vector,
q ¼ ki ks : ð2:50Þ
The angle between ki and ks is called the scattering angle h (see Fig. 2.2).
Figure 2.2b makes clear that the magnitude of the scattering wave vector q can
be computed as
q2 ¼ jqj2 ¼ ki2 þ ks2 2ki ks : ð2:51Þ
When—like in most light scattering experiments—ki & ks, i.e., the scattering is
quasi-elastic, the cosine rule becomes applicable and Eq. 2.51 becomes
h
q2 ¼ ki2 þ ks2 2ki ks ¼ 2ki2 ð1 cos hÞ ¼ 4ki2 sin2 : ð2:52Þ
2
Or, with ks = ki = 2pn/k, where the length of the incident wave vector is written
pffiffiffiffi
as function of the refractive index nðn ¼ 0 Þ and the incident wave length k:
h 4pn h
q ¼ 2ki sin ¼ sin : ð2:53Þ
2 k 2
This is the Bragg condition.
(a) (b)
Fig. 2.2 a At the detector, the total radiated field is the sum of all fields radiated from any
infinitesimal volume d3r at position r from the center O of the illuminated volume. The detector is
at position R: b General light scattering setup: incident light of polarization ni ; frequency xi, and
wave vector ki is scattered. Although it is scattered in all directions (not shown for clarity), the
figure shows only the light with ns and wave vector ks that can reach the detector after passing an
analyzer. The wave vector q is shown in gray as ki ks (according to [7])
2.2 Light Scattering Basics 19
Using the spatial Fourier transform of the dielectric fluctuation

Z
dðq; tÞ ¼ d3 r exp iqrdðr; tÞ; ð2:54Þ
V
Eq. 2.49 can be rewritten as

E0
Es ðR; tÞ ¼ exp iðks R xi tÞ½ns ½ks ks ðdðq; tÞni Þ: ð2:55Þ
4pR0
This can be simplified [7]:
ks2 E0
Es ðR; tÞ ¼ exp iðks R xi tÞdis ðq; tÞ; ð2:56Þ
4pR0
where
dis ðq; tÞ ns dðq; tÞni ð2:57Þ
is the component of the dielectric constant fluctuation tensor along the initial and
final polarization direction.
The time-correlated function of Es can then be written as
k4 jE0 j2
Es ðR; 0ÞEs ðR; tÞ ¼ s 2 2 2 hdis ðq; tÞi expðixi tÞ: ð2:58Þ
16p R 0
Then, the spectral density

Z þ1
1
IE ðxÞ ¼ dshE ðtÞEðt þ sÞi expðixsÞ ð2:59Þ
2p 1
of light reaching the detector with ns ; ks ; and xs can be computed as

Z þ1
I0 ks4 1
Iis ðq; xs ; RÞ ¼ 2 2 2
dthdis ðq; 0Þdis ðq; tÞi expðiðxÞtÞ ð2:60Þ
16p R 0 2p 1
with I0 jE0 j2 and

x xi xs : ð2:61Þ
The latter means that the spectral density (the intensity measured by the
detector) depends only on the difference between the incident and the scattered
frequency. Furthermore from Eq. 2.60 one can learn that Iis / ks4 (i.e., Iis / k4 )
and Iis / R2 : The k-4 dependence means that electromagnetic radiation with
short wavelengths is scattered more than that with longer wavelength, as, e.g.,
apparent in the blue color of the sky. The R-2 dependence is expected, because
it just expresses the attenuation of a spherical wave.
For a given experiment, the coefficient in Eq. 2.60 becomes a constant, and the
scattering intensity is then only affected by the spectral density of the dielectric
constant fluctuations, i.e., the integral
Z þ1

Iis ðq; xÞ / dt d is ðq; 0Þdis ðq; tÞ : ð2:62Þ
1
The integral of Eq. 2.62 over frequency, the integrated intensity at all fre-
quencies, provides information about the q-dependent mean-square fluctuations :
D E
Iis ðqÞ ¼ jdis ðqÞj2 : ð2:63Þ
Harking back on the definition of the scattering wave vector q (Eq. 2.50) and
the frequency shift x (Eq. 2.61), the scattering event can be considered in terms of
energy and momentum conservation. Most generally, during a scattering process,
the scattered photon sustains an energy change from hxi to hxs and a momentum
change from hki to
hks : This must be related to the creation or annihilation of an
excitation in the scattering medium. It is:
hx ¼
hxs hxi ð2:64Þ
hk ¼
hks
hk i : ð2:65Þ
Note that in these equations the energy and momentum of the excitation can have
both positive and negative sign.
2.3 Brillouin Light Scattering
Brillouin light scattering (BLS) is the inelastic scattering of monochromatic laser

light by phonons in the GHz frequency range. The required high resolution is
obtained by the use of multipass tandem Fabry–Pérot interferometers. In BLS
spectroscopy of transparent samples, the desired dispersion relations are obtained
by recording the phonon frequencies as a function of the scattering wave vector q,
which varies with the scattering angle. However, in samples exhibiting strong
multiple light scattering, as for the dry colloidal crystals (opals), q is ill-defined
and hence q-dependent acoustic-like modes become inaccessible. Though, local-
ized in space, i.e., q-independent modes can be recorded in the BLS spectrum as it
was demonstrated for sub-micron colloidal silica [8] and polymer crystals [9].
These q-independent frequencies have been identified as the resonance modes of
the individual colloidal particles, i.e., BLS can record numerous thermally excited
elastic resonances in one measurement (Chap. 4). These eigenfrequencies are
uniquely defined by the geometrical and elastic characteristics of the particles.
Based on these data, the elastic properties of the materials can be calculated at the
nanoscale. Thus the combination of q-independent BLS spectroscopy of multiply
2.3 Brillouin Light Scattering 21
light scattering (opaque) samples and the dispersion relations from the q-depen-
dent BLS spectroscopy on transparent samples is a powerful methodology to
investigate the elastic behavior of nanostructured materials.
This section will introduce the basic principles of BLS following a simple
approach. Further important results obtained by thermodynamical considerations
will be introduced very briefly. Then a description of the BLS setup with
selective attention on the main principles of the tandem Fabry–Pérot interfero-
meter is given.
2.3.1 BLS Basics
The main principle of BLS is the scattering of photons on sound waves and the
constructive interference of the multiply reflected light beam as sketched in
Fig. 2.3. A plane elastic wave creates a periodic change of density and hence of
the local dielectric constant in a medium, symbolized by the black/white layers.
The typical velocity of an acoustic wave is between 103 and 104 ms-1, while the
velocity of the probing light c is &3 9 108 ms-1. Because of this great dis-
crepancy the dielectric inhomogeneities can be regarded as a quasi-static (i.e.,
‘frozen’) lattice on which the photons of the probing light are scattered. Thus, it is
justified to treat the medium as a periodic multilayer stack with periodicity K, the
wavelength of the phonon, as shown in Fig. 2.3. The probing laser light is multiply
reflected (under the scattering angle h) on these layers and the reflected light
interferes on the detector, whose distant to the sample R is much larger than the
periodicity of the scattering planes ðR
KÞ: The reflected intensity reaches its
maximum when the interference is constructive, i.e.,
h
2nK sin ¼ k ð2:66Þ
2
with refractive index n and wavelength of light k.
Fig. 2.3 BLS scattering

process: the probing light k i, i
beam is multiply reflected on ks , s

the planes of a quasi-static
periodic multilayer stack of
modulated local dielectric
constant with periodicity K.
sound wave
Only reflected beams that

interfere constructively
account for the scattering
intensity
Insertion of Eq. 2.66 in the Bragg condition (Eq. 2.53) leads to

2p 4pn h
¼ sin ð2:67Þ
K k 2
and
2p
q¼ : ð2:68Þ
K
That is to say the scattering wave vector q is equivalent to the wave vector k of the
sound wave (k = 2p/K), viz. by changing h; q can be chosen, too, and at every q
the corresponding sound waves are probed selectively. The identity q ¼ k shows
that the momentum conservation (cf. Eq. 2.64) during the scattering process is
realized by a momentum transfer between the sound wave and the photon. From
this point of view it is clear that the scattered light changes its angular frequency
from xi to xs by inelastic interaction, in which a phonon, the acoustic quantum,
can be either created (Stokes process) or annihilated (anti-Stokes process),
depending on the direction of motion of the acoustic wave. In terms of energy
conservation considerations (Eq. 2.65) it is obvious that the exchanged energy hx
comes from the angular frequency of the phonon, xphonon = |x| = xs - xi, which
is related to the phonon wavelength and the velocity c of the acoustic wave by
2pc
xphonon ¼ x ¼ : ð2:69Þ
K
x
Therefore the frequency f ¼ undergoes a Doppler shift:
2p
cq c 4pn h
fs ¼ fi ¼ fi sin ; ð2:70Þ
2p 2p k 2
using the definition of q given in Eq. 2.53. Here, the minus corresponds to a
motion of the acoustic wave away from the detector leading to a frequency
decrease (or phonon creation; Stokes), and the plus corresponds to a propagation
direction of the acoustic waves towards the detector leading to an increasing
photon frequency (or phonon annihilation; anti-Stokes). Obviously, it is
x ¼ xs xi ¼ cq: ð2:71Þ
Therefore the BLS spectrum consists of doublets centered at the elastic frequency
with frequencies f = ±cq/2p (frequency shifts).
Equations 2.64 and 2.65 can now be rewritten for the Brillouin scattering
process as follows:
hxs hxi ¼
hx ð2:72Þ
hks
hki ¼
hq: ð2:73Þ
Note that in these equations momentum and energy (frequency) of the acoustic
wave are positively defined and plus- and minus-signs correspond again to anti-
Stokes and Stokes process, respectively.
So far, this easy approach explains satisfactorily the appearance of the doublet
as well as the position of the frequency shifts. However, when performing BLS
experiments a central line always appears that is not explained by the above
considerations. Also the intensity of the signals can not be predicted. To elucidate
these features of a BLS spectra it is necessary to take some ideas from thermo-
dynamics into account.
The total scattering intensity of a Brillouin spectrum was first derived by
Einstein in 1910 [5, 6]. He started with considering as a function of density and
temperature ¼ ð.; TÞ: For . and T are statistically independent, one can write
the total differential

o o
d ¼ d. þ dT: ð2:74Þ
o. T oT .
Then
D E o 2 D E o 2 D E
2 2
ðdÞ ¼ ðd.Þ þ ðdTÞ2 : ð2:75Þ
o. T oT .
Using the assumption

o o

ð2:76Þ
o. T oT .
and
D E
ðd.Þ2 kB TbT
¼ ð2:77Þ
.2 v
for the mean-square fluctuation in density in the volume element v, where kB is
Boltzmann’s constant and bT is the isothermal compressibility, Eq. 2.75 becomes
D E o 2 D E o 2 k Tb
2 2 B T
ðdÞ ðd.Þ ¼ . : ð2:78Þ
o. T o. T v
Comparison with Eq. 2.63 shows the main result of these calculations,
I / bT ; ð2:79Þ
i.e., the total scattering intensity depends on the isothermal compressibility.
Without going into details, the full vectorial expression for IðqÞ is given by [7]
2
2 o
Iis ðqÞ ¼ ðni ns Þ V.2 kB TbT ð2:80Þ
o. T
with scattering volume V. If V(h) & const, which may be valid depending on the
experimental details, the total scattering intensity is also independent from the
scattering angle.
Although Eq. 2.80 gives the total scattering intensity, experiments showed that
only a part of the intensity belongs to the Brillouin doublets while the rest makes
for a central line. These findings could be explained by Landau and Placzek
regarding (S, p) as function of entropy S and pressure p [1]. In analogy to
Eq. 2.75 one can write
D E 2 D E o 2 D E
2 o 2
ðdÞ ¼ ðdSÞ þ ðdpÞ2 ð2:81Þ
oS p op S
because fluctuations in S and p are independent, too.

The first term, also called the Rayleigh term, represents local entropy fluctua-
tions, which do not propagate in normal liquids and are the source of the unshifted
component of the scattered light with intensity IC. The other term, the Brillouin
term, represents the isoentropic pressure fluctuations, ergo, sound waves that
contribute to the Brillouin doublet. With
D E k T
B
ðdpÞ2 ¼ ; ð2:82Þ
vbS
where bS is the adiabatic compressibility, and by writing

o o o. o
¼ ¼ .bS ð2:83Þ
op S o. S op S o. S
the Brillouin term can be transformed into

D E o 2
o 2 2 2 kB T o 2 kB TbS
ðdpÞ ¼ . bS ¼ . : ð2:84Þ
op S o. S vbS o. S v
It becomes clear that the intensity of the Brillouin doublet 2IB depends on the
adiabatic compressibility bS.
After further simplifications it can be shown that the ratio between total scat-
tering intensity (IC + 2IB) and the intensity of the Brillouin doublet is simply given
by the ratio of isothermal and adiabatic compressibility or of the specific heats at
constant pressure or volume [1, 6, 7]:
IC þ 2IB bT Cp
¼ ¼ : ð2:85Þ
2IB bS CV
In the form
IC b bS Cp CV
¼ T ¼ ¼c1 ð2:86Þ
2IB bS CV
it is known as the Landau–Placzek equation.
2.3.2 BLS Instrumentation
It was already pointed out that the relative shift of the photon frequency in BLS
spectroscopy (*108 to 1011 Hz) is quite subtle compared to the initial frequency
of the probing light (1014 Hz) or to the resolution obtained by Raman spectros-
copy, where frequency shifts in the order of 1013 Hz are measured frequently by
diffraction grating spectrometers. In order to achieve such high resolution, Fabry–
Pérot interferometers (FPs) are used in BLS. In combination with an highly
monochromatic laser light source a single FP or, even better, a design composed of
two FPs with multiple light pass can give excellent results. Here the FP principle is
elucidated and the construction details of the used tandem FP as well as the general
features of the whole BLS setup are illuminated.
In principle, a single FP is not much more than an etalon as it is used in laser
resonant cavities. The FP consists of two plane mirrors with reflectivity R mounted
accurately parallel to one another. The spacing between the mirrors is given by d and
can be varied by moving one of the mirrors. When the laser light enters the etalon
through the first mirror, which is usually a plane glass plate with a thin layer of metal
on the inner side, it is reflected numerously between the two mirrors. However, the
light with intensity II is already reflected when it enters the etalon, so that we have
only the intensity I0 = II(1 - R) inside the FP before the first internal reflection.
During the reflections within the FP, the intensity decreases after x reflections to
Ix ¼ I0 Rx ¼ II ð1 RÞRx ; ð2:87Þ
because every time a part of the light is also transmitted; a typical value for BLS
experiments is R & 0.93. Note that Eq. 2.87 is not fully true as it does not take
into account the dissipation of energy, which is usually considered by the
absorptance A, i.e., R + T + A = 1, with transmittance T. Anyhow, it implies that
with increasing reflectivity the number of reflections inside the FP increases
rapidly, and this is important for the working principle of the system. When the
light is reflected many times between the mirrors, the reflected beams interfere.
Only light that fulfills the equation
mk ¼ 2nFP d cos hFP ; ð2:88Þ
with m being an integer and nFP being the refractive index inside the FP, and hFP
being the angle between the light in the FP and the normal to the mirrors, is
transmitted losslessly due to constructive interference. Thus usually the interfer-
ometry is performed by moving one of the mirrors, there is just air between the
mirrors. (The other possibility would be to vary n during a measurement.) With
normal incidence (cos 0 = 1) and n = 1 we can therefore simplify to
2d
k¼ : ð2:89Þ
m
That means that only light with wavelength 2d/m is transmitted, and the trans-
mitted k can be easily scanned when scanning different ds by moving a mirror,
e.g., by a piezo transducer. The separation between two adjacent transmission

maxima for a given d is called the free spectral range (FSRk = Dk). Differenti-
ating Eq. 2.89 leads to
Dk Dm
¼ ð2:90Þ
k m
with initial wavelength k. For m ± 1 it follows in the described setup (normal
incidence, nFP = 1)
k2
FSRk ¼ ; ð2:91Þ
2d
or in the frequency domain
m
FSR ¼ ; ð2:92Þ
2d
with m being the velocity of light in vacuum.
The linewidth of the transmitted line depends strongly on R. As denoted above,
the more often the beam is reflected before it transmits the second beam of the FP
(or the first, but this light is uninteresting for the further analysis) the more often it
can interfere with itself. Therefore the destructive interference will be enforced for
wavelengths not satisfying Eq. 2.89 with increasing reflectivity and the function
I(k) becomes more narrow.
Indeed, the phase difference dp between each succeeding reflection is
2p
dp ¼ 2d: ð2:93Þ
k
The transmission function of the etalon T(dp) is found to be (http://en.wikipedia.
org/wiki/Fabry_Perot, April 2009)
ð 1 RÞ 2 1
Tðdp Þ ¼ ¼ ; ð2:94Þ
1 þ R2 2R cos dp 1 þ cF sin2 dp
2
where cF is the so-called coefficient of finesse F:
4R
cF : ð2:95Þ
ð1 RÞ2
F is defined as the ratio between FSRk and the full width at half maximum
(FWHMk) of a transmission peak in the T(k)-function, i.e., F gives the relative
separation between nearest transmission peaks (cf. Fig. 2.4) and is also the most
important parameter for the practical resolution of a spectrometer. As it is
approximately
Fig. 2.4 High and low

finesse, FWHMk at
fixed FSRk
4
FWHMk pffiffiffiffiffi; ð2:96Þ
cF
the finesse is found to be

pffiffiffiffiffi pffiffiffi
FSRk p cF p R
F¼ ¼ : ð2:97Þ
FWHMk 2 1R
However, in real experiments the ‘practical finesse’ depends not only on the
reflectivity of the mirrors but also, e.g., on the mirror flatness. In general, a high
F is intended, but by practical limitations it cannot be made much greater than
about 100 [10].
Single-pass FPs have been used to realized the first measurements of traveling
acoustic waves in solids and liquids. However, it turned out quite early that the
contrast was too low to resolve weaker signals and also the interference between
neighboring orders could create easily very complicated spectra. Both problems
were solved by John Sandercock when he introduced his multi-pass tandem
Fabry–Pérot interferometer (multi-pass modes are also possible in single FPs).
The idea behind is quite simple: two FPs with slightly different FSR are con-
nected in series (tandem). An intelligent use of additional mirrors assures that the
light passes each FP several times (multi-pass). In the tandem operation, both FPs
must transmit the same wavelength simultaneously by an appropriate scanning
technique. Due to the differing FSRs of both individual FPs, always one of the FPs
blocks the neighboring interferences. Let us assume FP1 to have a free spectral
range FSRk1 and FP2 FSRk2. Then only wavelengths are transmitted that simul-
taneously satisfy
ktrans ¼ 2pd1 and ktrans ¼ 2qd2 ; ð2:98Þ
where d1 and d2 are the mirror distances in FP1 and FP2 and p and q are integers.
The effect of the tandem operation is sketched in Fig. 2.5b. The next order of light
(a) (b)
d 2= d 1cos
transmission
FP1
1
FP2
FP2
2
FP1 tandem FP
translation
stage
d1 wavelength
Fig. 2.5 a Basic mechanism of a Sandercock multipass tandem Fabry–Pérot interferometer. The
light passes FP1 and then FP2. Two mirrors (one of FP1 and one of FP2) are moved simulta-
neously on a piezoelectric translation stage. The angle f of the light path between the two FPs is
chosen in that way that the ratio between d1 and d2 is constant. b Effect of the multi-pass through
two FPs: the neighboring orders (viewed from the harmonized central signals) that can pass each
single FP are suppressed. Only the common multiples of Dk1 and Dk2 will be transmitted
gainlessly
transmitted through FP1 (i.e., Dp = 1) is nearly completely suppressed by FP2 and

vice versa, depending on the reflectivity of the mirrors. In our case the transmit-
tance of wrong wavelengths is about 10-3. According to that, the total FSR of the
tandem interferometer is the distance between the common multiples of Dk1 and
Dk2. Therefore, with tandem FPIs much higher finesses can be achieved.
Although the principle idea is not complicated at all, the technical realization
was problematic. Both FPs have to be scanned synchronously, and from Eqs. 2.89
and 2.98 it is clear that the ratio between the mirror distances in both FPs must be
constant. With other words
Dd1 d1
¼ : ð2:99Þ
Dd2 d2
Indeed, this condition was hard to satisfy as the total scanning distance is usually
below 1 lm and the scan is repeated many times during one measurement, so that
even very small asymmetries in the nanometer-range would disable the
experiment.
This problem was solved in 1987 by a new, elegant design, presented in
Fig. 2.5a [11]. The two FPs used in Sandercock’s ‘parallelogram geometry’
consist of one fixed mirror and one movable mirror, each. The two movable
mirrors are placed on a shared panel, the translation stage, which is the part of the
setup moved during the scan. Thus, both FPs change the mirror distance always at
the same time. The FPs are mounted that the two mirrors of each would touch at
the same time if the translation stage would be moved to the very left. In order to
satisfy Eq. 2.99 the relative orientation of the two FPs is chosen that way that the
angle f between FP1 and FP2 is fixed and the mirror distances have then to be
adjusted in a way that
d2 ¼ d1 cos f: ð2:100Þ
A movement of the translation stage to the right shifts the spacings between the
mirrors simultaneously by Dd1 and Dd1cos f, i.e., the ratio keeps constant.
In my experiments, a six-pass tandem Fabry–Pérot interferometer was used.
The path of the light inside the FP as well as the other details of the applied BLS
setup are sketched in Fig. 2.6. The sample is mounted in the center of a goniometer
(Huber) in a custom-made sample holder with or without oven. A solid state
pumped frequency-doubled Nd:YAG laser (coherence; 150 mW at 532 nm) is
fixed on the goniometer and can be rotated so that scattering angles between 0° and
*160° can be chosen either in transmission or reflection geometry. In difference
to pure BLS backscattering techniques, i.e., h = 180°, this technique has the
advantage that not only the components of q but the whole wave vector is changed
according to Eq. 2.53.
Before the light reaches the sample, it passes a Glan polarizer (extinction ratio
10-5) with vertical polarization (V), i.e., perpendicular to the scattering plane, to
ensure fully polarized incident light. Behind the sample the light scattered in the
direction of the detector is collected by an aperture and focused into the entrance
pinhole of the tandem Fabry–Pérot interferometer (JRS Scientific Instruments) by
some lenses. Before entering the FP, a Glan–Thompson analyzer (extinction ratio
10-8) is passed that selects either vertically (V), i.e., perpendicular to the scat-
tering plane, or horizontally, i.e., parallel to the scattering plane, polarized light
(or, of course, everything in between). After passing the FP the transmitted light is
detected by an avalanche photo diode (APD) and processed by an multi-channel
analyzer with 1,024 channels. The further processing is performed by a computer
software.
The stability of the alignment is greatly enhanced by the use of a reference
beam. Therefore, a small amount of the laser light is diverted via the reflection
Fig. 2.6 BLS setup with six-

pass tandem Fabry–Perot
AP
interferometer. Electronic
D
stabilization over days to

weeks is achieved by perma-
nent compensation using a
POLARIZER
diverted part of the unscat-

LA
tered light as reference beam.

APERTURE
ANALZYER
SE
The goniometer allows to

R
record dispersion relations

with continuous q-range.
For the eigenmode spectra
the scattering angle is not
relevant [12] SAMPLE
TANDEM FABRY-PEROT
GONIOMETER
(5%) of a parallel plate on a beam splitter and an optical fiber from the incident
laser beam and introduced as a reference beam that gives the central line via an
optical fiber. Therefore, a mechanical shutter is used, which switches periodically
the entrance to the FP between reference beam and scattered light and excludes the
central elastic line. In order to avoid mechanical disruptions the whole setup is
placed on an optical table with active vibration damping.
In some experiments dealing with temperature-dependent effects, e.g., glass
transition experiments (Sect. 4.3.1 in Chap. 4) or measurements on kinetically
stable organic glasses (Chap. 6), an oven is used in order to control the tem-
perature of the sample in the range between ca. 10 and 200 °C. The oven is a
metal cylinder with filament and coolant tubes in the wall [13]. A small slit at the
hight of the laser and a cylindrical quartz glass insert in order to avoid heat
transfer through the slit allow the light to reach the sample and the detector
afterwards. The temperature is controlled electronically (built in-house) by two
Pt-100 temperature sensors inside the wall and inside the heat chamber, close to
the sample. The temperature can be stabilized within better than ±0.2 K in the
given range.
Most experiments in this thesis deal with samples on a plane substrate, i.e.,
‘films’. When performing the BLS experiment on such samples, there are in
principle two different scattering geometries, the transmission and the reflection
geometry. In transmission geometry the light scattered on the other side of the film
than that of the incident laser beam is investigated, while in reflection geometry
the light scattered on the side of the incident beam is regarded.
The two geometries are sketched in Fig. 2.7 and a full geometrical derivation is
given in Appendix.
Fig. 2.7 Principle BLS scat- (a) reflection

tering geometries for a film
incident
sample (view atop the scat-
n0=1 laser
tering plane). a Transmission
geometry, b reflection geom-
etry. On the right side the n ki ks qperp q
vector decomposition for the q qpara
wave vector q in its compo- n0=1
nents perpendicular ðqperp Þ
and parallel to the film transmission SCATTERING
plane ðqpara Þ
(b) SCATTERING incident

reflection laser
n0=1 qperp
ks
n q q
ki
n0=1 qpara
transmission
In the transmission case, it is found that the magnitude of the scattering wave
vector becomes

4pn 1 1 1 1 1
q¼ sin sin sin ðh aÞ þ sin sin a ð2:101Þ
k 2 n n
with the angles a and h given in Fig. 2.7a. It is shown in the appendix that
the length of the component of the scattering wave vector parallel to the film,
qpara, is
2p
qpara ¼ ðsin a þ sin ðh aÞÞ: ð2:102Þ
k
Obviously, in this equation the refractive index n of the sample is eliminated.
A special transmission geometry in which q ¼ qpara exists for h = 2a. In this
case q is given simply by
4p h
q ¼ qpara ¼ sin ðfor h ¼ 2aÞ: ð2:103Þ
k 2
To use this special geometry is advantageous in several ways, indeed nearly all the
experiments in this thesis are performed using it. First of all, it facilitates the
calculation of q, as the refractive index of the sample has not to be taken into
account. Although n of most ‘standard materials’ (e.g., typical polymers like
polystyrene) is well-known, the exact determination of n for unknown materials or,
even more, for composite materials, as they are widely used in this thesis, may
become complicated and time consuming. The second great benefit of this
geometry is that in this case the direction of q is well defined parallel to the film.
When discussing colloidal crystals [14] it is important to know which crystallo-
graphic direction is probed by the BLS experiment.
In the reflection geometry the general expression for q as a function of n, h, and
a becomes

4pn 1 1 1
q¼ cos sin1 sin a þ sin1 sin ðh þ aÞ : ð2:104Þ
k 2 n n

Also for the reflection case there is a special scattering geometry. If a ¼ 1802 h the
scattering wave vector is identical to its component perpendicular to the film, i.e.,
q ¼ qperp in Fig. 2.7b.
In a cylindrical sample, e.g., a liquid in a NMR-tube, q has simply the
magnitude
4pn h
qcylinder ¼ sin : ð2:105Þ
k 2
2.3.3 Vibrational Modes of Individual Particles
In the foregone section it was pointed out that the result of a BLS experiment is
usually the plot of longitudinal or transverse [or mixed, especially in thin films,
e.g., the so-called Lamb waves (http://en.wikipedia.org/wiki/Lamb_wave, May
2009)] sound waves as a function of the scattering wave vector q; while the
frequencies f(q) are given as a Doppler shift of the probing laser light around its
elastic line. The phase sound velocities are given by the slope of the f(q)-diagram.
Thus, in the case of a nonlinear function f(q), i.e., if the acoustic mode is
dispersive, the result of the BLS experiment is the dispersion relation in the
investigated q-range.
However, the appearance of a dispersion relation is only meaningful if it
is possible to determine q. If strong multiple scattering occurs in a sample,
q becomes ill-defined. In such samples f(q) is no longer accessible. Anyhow, now
the inelastic scattering from localized modes, i.e., the vibrational resonance
modes, can lead to incoherent BLS in analogy to the Raman scattering.
Indeed, in BLS experiments such samples show a spectrum where scattering at
all possible qs between zero degree incident and backscattering case contributes to
the final result, independent from the experimental scattering angle h. Further-
more, also the polarization information is lost.
Nevertheless, the Brillouin spectrum of multiple scattering samples with well
defined shape still gives useful information, which are not accessible by other
techniques. In Chap. 4 of this thesis, several experiments are discussed dealing
with the detection of vibrational resonance modes of spherical colloidal samples.
BLS spectra of such samples deliver many useful information about the individual
colloids and, in case of hybrid materials, about the mechanical properties of the
components, too. This section will give a short introduction into the mathematical
description of the vibrations in spheres, their theoretical calculation, and their
detection by BLS.
Under stress-free boundary conditions, the vibrational modes are usually
called eigenmodes. The eigenmodes for free homogeneous elastic spheres have
been derived by Lamb in the nineteenth century [15]. The modes can be clas-
sified as torsional and spheroidal ones, both labeled by the indices n, l, and m,
which describe the radial (n) and the angular (l, m) dependence of the dis-
placement (cf. Sect. 2.1.3). The torsional modes are fully tangential, i.e., they
involve only shear motions and do not cause changes in the sphere volume—they
do not contribute to the BLS intensity. The spheroidal modes involve usually
both shear and stretching motions, and they can be fully specified by two indices
in analogy to the atomic orbitals as the n-th order radial solution (n = 1, 2, 3,...)
for angular momentum ‘quantum number’ l (l = 0, 1, 2,...) [4, 12, 16]. Only
spheroidal modes with l = 0 have purely radial displacement (breathing modes)
[17].
By solving the wave equation for elastic and isotropic media (Eq. 2.37), Lamb
found the frequencies of the eigenmodes (n, l) to be (in a very general expression):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
Rðcl ; ct Þ
f ðn; lÞ ¼ xðn; lÞ d; ð2:106Þ
.
with x(n, l) being a constant for each individual mode, R(cl, ct) being the rigidity,
and d being the diameter of the sphere. The rigidity is a function of the longitu-
dinal and transverse sound velocities and has the dimension of pressure, thus it is
simply a modulus. Indeed, e.g., for the (1,2)-mode, R is identical with the shear
modulus l(=C44).
The experiments discussed in Chap. 4 are all related to the vibrational modes of
colloidal spheres in air. Because of the large density (and elastic) mismatch
between air and the colloids, the boundary-conditions are quasi that of a stress-
free, undamped vibration, thus these modes are practically real eigenmodes and
the values calculated by Lamb’s theory are a good approximation. However, in the
more general case of a elastic vibrator embedded in an elastic matrix, some
coupling should occur between the eigenmodes and the propagating acoustic
waves in the matrix. The amount of coupling depends on the elastic mismatch
between spheres and matrix. While the limit case of large mismatch corresponds to
the free eigenmodes, where the elastic energy is totally localized in the spheres, the
other limit case is a zero-mismatch, which means that all energy is coupled
between sphere and matrix and an elastic wave can travel the system like a
homogeneous medium.
To calculate the distinct vibrational modes of elastic spheres with radius rs and
density .s in an elastic matrix with density .m , one has to introduce boundary
conditions. At the surface of the sphere, the displacement u has to be the same
inside and outside the sphere [4, 18]:
ui jr¼rs ¼ um jr¼rs : ð2:107Þ
The same must be fulfilled for the surface traction sðs ¼ rðrÞ r^; with stress tensor
r and outgoing unit vector normal to the sphere surface r^Þ :
si jr¼rs ¼ sm jr¼rs : ð2:108Þ
The field can be written as a linear sum of spherical waves as discussed in

Sect. 2.1.3. The field inside the spheres becomes
X
ui ðrÞ ¼ aLlm llm ðR; kli Þ þ aM i N i
lm mlm ðR; kt Þ þ alm nlm ðR; kt Þ ð2:109Þ
lm
with the coefficients aLlm ; aM N

lm and alm : In the same way we can write for the field in
the matrix
X
um ðrÞ ¼ bLlm lm M m N m
lm ðR; kl Þ þ blm mlm ðR; kt Þ þ blm nlm ðR; kt Þ ð2:110Þ
lm
with the coefficients bLlm ; bM N

lm and blm : With Eqs. 2.43–2.45 one can write
X 1
uðrÞ ¼ xLlm r½Rl ðkl rÞYlm ð^
rÞ þ xM
lm r ½rRl ðkt rÞYlm ð^
rÞ
lm
kl
ð2:111Þ
1
þ xNlm r r ½rRl ðkt rÞYlm ð^
rÞ;
kt
where the x replace either the coefficients a inside or b outside the spheres and kl
and kt are the longitudinal wavenumbers (k = x/c) inside (ki) or outside (km)
the sphere, respectively.
With Eq. 2.111 it becomes clear that Eqs. 2.107 and 2.108 are equivalent to
three scalar equations, each, and the boundary conditions can be expressed as a
system of six homogeneous equations with an infinite numbers of unknowns. Due
to the orthonormality over the spherical surface of the Legendre functions Ylm, it is
possible to decompose each of these equations into l equations, but independent
from m. That is also the coefficients a and b depend only on l, leading to a system
of six equations with six unknowns for each l. It can be broken into two smaller
systems, considering the orthogonality of mlm to nlm and llm ; which contain two
equations involving the coefficients aM M
l and bl and four equations containing the
L L N N
coefficients al ; bl ; al ; and bl ; respectively. Non-trivial solutions for both systems
are only achieved if their determinants are zero, leading to a set of discrete modes
with angular frequencies xnl for each l and n-th order, corresponding to the
expression given in Eq. 2.106.
The considerations presented in this section are the fundamental of theoretical
methods dealing with the scattering of plane waves on single and multiple spheres
presented in Sect. 3.4 in Chap. 3.
References
1. Landau LD, Lifshitz EM (1983) Elastizitätstheorie. Lehrbuch der theoretischen Physik.

Akademie-Verlag, Berlin
2. Beltzer AI (1988) Acoustics of solids. Springer, Berlin
3. Liu ZY et al (2000) Phys Rev B 62:2446
4. Cheng W (2007) Hypersonic elastic excitations in soft mesoscopic structures. PhD thesis
5. Einstein A (1910) Ann Phys 338:1275
6. Cummins HZ, Gammon RW (1966) J Chem Phys 44:2785
7. Berne BJ, Pecora R (2000) Dynamic light scattering. Dover, Mineola
8. Kuok MH et al (2003) Phys Rev Lett 90:255502
9. Cheng W et al (2005) J Chem Phys 123:121104
10. Sandercock JR (2008)
11. Mock R, Hillebrands B, Sandercock JR (1987) J Phys E: Sci Instrum 20:656
12. Still T et al (2008) J Phys: Condens Matter 20:404203
13. Cheng W et al (2007) Macromolecules 40:7283
15. Lamb H (1882) Proc Lond Math Soc S1–13:189
16. Montagna M (2008) Phys Rev B 77:045418
18. Psarobas IE, Stefanou N, Modinos A (2000) Phys Rev B 62:278
Chapter 3
Methods
3.1 Vertical Lifting Deposition and Colloidal Crystals
The assembly of colloids into well defined coherent structures commonly occurs
under the influence of external fields, e.g., gravitational sedimentation [1], elec-
trophoretic deposition [2], or vertical deposition either by evaporation [3] or lifting
the substrate [4] (Fig. 3.1).
In the vertical lifting deposition method upon immersion of a hydrophilic
substrate into a colloidal dispersion a meniscus at the substrate is formed. Evap-
oration takes place at the three phase contact lines (air, dispersion, and substrate),
which causes a solvent flux towards the meniscus, as shown in Fig. 3.2. Thus,
colloidal particles are constantly transported with the liquid to the crystallization
front. An interplay of long-range attractive and short-range repulsive forces causes
the self-assembly of the colloidal material into a face-centered cubic (fcc) or
hexagonally close packed (hcp) crystal [5, 6]. In order to control the thickness of
such a colloidal crystal—a critical parameter for further use in ensuing applica-
tions—the substrate is withdrawn from the dispersion at a certain speed at given
environmental parameters such as temperature and humidity. Besides the fabri-
cation of such ‘simple’ colloidal crystals also more complex systems like binary
[7, 8] and ternary colloidal crystals [9] were demonstrated. The colloidal crystals
that are topic of this article are produced using an enhanced vertical lifting
apparatus by Retsch [10].
Alongside with this research, also the counterpart to the highly ordered crystals—
colloidal glasses [11, 12]—comprising of two distinct, monodisperse latex particles
have already been developed [13]. Doping of the colloidal crystal with removable
moieties (sacrificial templates) opens a pathway to the designed introduction of
defects, which are highly interesting with respect to their contribution to phononic
properties. At the same time, colloidal crystals and glasses serve as templates for the
fabrication of so-called inverse opals [16]. These materials exhibit an interconnected
3D network with high surface area, since the constituent spheres of the colloidal
crystals have been removed. Inverse opals have attracted strong interest in photonic

36 3 Methods
(a) (c) (e)
(b) (d) (f)
Fig. 3.1 Self assembly of colloidal crystals by a sedimentation, b electrodeposition, c centri-

fugation, d vertical deposition (cf. Fig. 3.2), e filtering, or f compression molding (according
to [13])
Fig. 3.2 Schematic draw of climate chamber:

the colloidal crystallization temperature, humidity
process in a vertical lifting
deposition apparatus. The lifting direction
black arrow symbolizes the
moving direction of the sub- evapo-
strate by an adjustable stepper ration
motor with small step size
(not drawn). The surrounding
climate chamber allows full
control over air temperature
and humidity, which are colloidal
important parameter in the dispersion
crystallization process
(according to [13])
crystal research, due to their full photonic band gap [16, 17], and are also promising
but unexplored materials for phononic experiments.
Complementary to the vertical deposition method a technique for the fabrica-
tion of large area colloidal monolayers has been recently established. This new
method itself gives access to highly interesting new materials. Multiple stacks of
colloids of various diameters as well as deposition on uneven or curved substrates
have been demonstrated. Controlled etching of such a colloidal monolayer allows
for the first time the fabrication of large areas of non-close-packed structures [18].
Finally, colloidal monolayers are well known and widely used for surface
patterning. Vertical lifting deposition, colloidal monolayer fabrication and the
3.1 Vertical Lifting Deposition and Colloidal Crystals 37
combination of both are powerful tools in order to rationally design new periodic
functional materials with generic possibilities for defect tuning, creation of multi-
layers, colloidal glasses, and inverse opals.
3.2 Melt Compression
Vertical lifting deposition is a method that delivers high quality colloidal crystals
several tens of microns thick and with cm2 areas. Although, in principle it would
be possible to upscale the method, it would be unpractical and slow to produce
much larger and thicker samples by that. A technique that is much better suited if
fast and large scale hybrid material films of thickness up to the millimeter range is
needed is the melt compression (or compression molding) technique that is
schematically shown in Fig. 3.1f and in more detail in Fig. 3.3a.
Melt compression can be utilized to prepare films out of hard core/soft shell
core–shell particles, where the soft shell becomes the matrix and the hard cores
form a crystalline (usually fcc) lattice, whose spacing depends on the initial vol-
ume ratio of core and shell materials. The preparation starts with the coagulation of
the latex dispersion in a rubbery mass. When the soft shell is an elastomer with low
Tg, e.g., poly(ethyl acrylate) (PEA), the shells form a continuous matrix in which
the cores, e.g., harder polymers like polystyrene or oxides like silica, are dispersed.
At DKI the mass is shaped into a cylindrical sample, which is uniaxially com-
pressed at 170 °C in a Collin 300 press, with an initial pressure of 1 bar and a final
pressure of 50 bar [19]. In the beginning of the pressing process, the first crys-
talline layers appear on the plates of the press. As shown in Fig. 3.3a, the pressure
creates a horizontal flow perpendicular to the direction of compression. The order
increases with pressure by the sequent formation of new crystalline layers above
the already existing layers in order to achieve the densest packing. Finally, the
whole film should be crystalline. It could be shown only recently that in such films
the direction corresponding to the (111) plane of the fcc lattice is radial from the
center of the round press cylinder, i.e., multi-domain ordering must be assumed,
however, the crystallographic directions are well defined [20].
Fig. 3.3 a The flowing melt crystallizes along the plates of the press under uniaxial compression
[19]. b Melt compressed polymer opal film with PS core and PEA shell [20]
38 3 Methods
Figure 3.3 shows a typical film prepared consisting of PS cores and PEA
matrix. It is obvious that in this case the particle size is chosen that way the film
acts as a photonic crystal, since against the dark background the film reflects the
green light under Bragg conditions. Where it is bended, the angle to the light
changes and it is reflected on another crystallographic plane, hence the color
changes.
3.3 Polymer and Colloid Characterization Techniques
The aim of this section is to introduce very briefly the different auxiliary tech-
niques applied in this thesis to characterize polymer and colloid samples, including
spectroscopy, electron microscopy, calorimetry, density determination, and size
exclusion methods.
3.3.1 Photon Correlation Spectroscopy
The photon correlation spectroscopy (PCS), also known as dynamic light scat-
tering, is a light scattering technique that can be used to obtain the size distribution
of small particles (d \ k) in diluted solution. When laser light hits small particles,
Rayleigh scattering occurs in all directions. Because the small particles undergo
Brownian motion in the liquid, the relative position of the scatterers within the
scattering volume is constantly changing. This results in a time-dependent scat-
tering intensity fluctuation, because the monochromatic and coherent laser light is
scattered on the moving scatterers. The light scattered from the individual scat-
terers interferes with the light scattered on all the other particles. By using a
correlator at a given scattering angle h, i.e., at a given wave vector q, one can
record the time–correlation function
Z
1 1
Gðq; sÞ ¼ hIðq; tÞIðq; t þ sÞi ¼ lim Iðq; tÞIðq; t þ sÞdt ð3:1Þ
T!1 T 0
with I(q, t) being the scattering intensity at time t and wave vector of magnitude q.
Thus I(q, t) is randomly fluctuating with time due to particle motion, G(q, s) is
decreasing exponentially
from its initial value G(q, 0) to zero for s ! 1—or, to
be exact, from I 2 i to I 2 : The so-called normalized second order autocorrelation
curve is generated from the intensity trace as follows:
hIðtÞIðt þ sÞi
g2 ðq; sÞ ¼ : ð3:2Þ
hIðtÞi2
In the simplest case of monodisperse spheres, the autocorrelation function decays as

a single exponential decay, which is directly related to the diffusion coefficient D.
3.3 Polymer and Colloid Characterization Techniques 39
When the viscosity g of the liquid is known, D gives the radius of the spherical
particles r by the Stokes–Einstein relation
kB T
D¼ : ð3:3Þ
6pgr
All PCS experiments in this thesis were performed by myself on a setup built
in-house at MPIP.
3.3.2 Electron Microscopy
The common characteristics of all types of electron microscopes is that electron

(particle) beams are used to illuminate a sample and to create a magnified image.
Fast electrons are generated by electron guns, consisting of an electron source
(cathode), an accelerating anode, and electron lenses to collimate and focus
the electron beam. The de Broglie wavelength of a particle with particle
momentum p is
h h
k¼ ¼ ð3:4Þ
p cme v
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where h is Planck’s constant and c ¼ 1 v2 =m2 the Lorentz factor for very high
velocities v. For an electron with mass me, the wavelength of the electron particle
beam depends on the velocity of the electrons, i.e., on the acceleration voltage
(usually in the order of several kilovolts). Usual electron wavelengths are 1 nm,
which is much smaller than the wavelength of the light used in optical microscopes
(200–1000 nm). Therefore the theoretical resolution limit is extremely low for
electron microscopes, but in practical work resolution is limited by aberrations of
the technical components; however, still orders of magnitudes better than in an
normal optical microscope. By the use of electron lenses (either magnetic or
electrostatic) with electronically adjustable foci, no movable objective system is
needed in an electron microscope. Principally, there are two main types of electron
microscopes, scanning electron microscopes (SEM) and transmission electron
microscopes (TEM).
In the SEM a thin electron beam rasters the probed surface. In the same time
electrons that are emitted or scattered from the probed object are detected and the
registered intensity is correlated to the intensity value of the scanned pixel.
Typically, in a SEM setup secondary electrons, back scattered electrons or, less
often, cathode luminescence signals are detected. The most important strength of a
SEM is the ability in displaying surfaces in high resolution with great depth of
view.
On the other hand, the transmission electron microscope probes not only the
surface but the whole gauge of a thin (some nanometer up to several micrometer)
sample using a transmission geometry. Here, the beam is usually not scanning.
40 3 Methods
A broader beam is illuminating a part of the sample and the transmitted electron
intensity is imaged onto a fluorescent viewing screen and recorded by a CCD
camera for further computing.
Electron microscope pictures shown in this thesis are recorded by Markus
Retsch, Gabrielle Schaefer, Gunnar Glasser, or Maren Mueller at different setups
at MPIP.
3.3.3 Differential Scanning Calorimetry
The differential scanning calorimetry (DSC) is a variation of caloric thermo-

analysis techniques, as there are also differential thermal analysis (DTA) or thermo
gravimetry [21]. DSC is an isotherm technique. The sample and a reference
substance are heated simultaneously in a way that both species have always the
same temperature, i.e., DT = 0 (heat compensation). Usually, the sample is heated
(and cooled down) with a well defined temperature rate (mostly 1–20 K/min).
Therefore, a different amount of heat DQ has to be added to the sample in com-
parison to the reference substance, which shows in ideal case a linear dQ/dT. The
measurement records now the change of the sample with time d(DQ)/dt or with
temperature d(DQ)/dT, respectively. In the resulting thermogram exothermal phase
transitions (e.g., crystallization) result in a peak with relative maximum, while
endothermal transitions (e.g., melting) result in a peak with a relative minimum in
the d(DQ)/dt or d(DQ)/dT curve. The integrals over the transition peaks give the
transition heat exchanged in the respective transition.
DSC experiments in this thesis are performed by Raeder and Droege at MPIP
and in the group of Prof. M. Ediger at University of Wisconsin, Madison, USA.
3.3.4 Density Gradient Column
In some cases the exact density of a sample is needed. While for polymers or
colloids the direct measurement of volume and weight might be sophisticated, the
use of a density gradient column is a popular method to measure a sample’s mass
density [21]. The column is a long vertically standing glass, which is filled with a
liquid whose density is increasing gradually from top to bottom. This can, e.g., be
realized by using a special double-flask setup in which two aqueous solutions of a
salt with two different concentrations (and therefore different densities) are mixed
to a solution with continuously changing concentration and density, which is
carefully poured in the column [23, 24]. The density gradient is then calibrated
using several floaters of well known density. Drawing the calibration curve of
floater density against floating hight in the column, the density of other samples
can be determined by flotation.
The density gradient column technique is often used to measure the crystallinity
of a polymer. The degree of crystallization wc is defined as
.c . .a
wc ¼ ð3:5Þ
. .c .a
with ., .c , and .a being the density of the probed sample, the purely crystalline
polymer, and the purely amorphous polymer, respectively. .c is accessible from
the crystal structure, . and .a can be obtained by flotation in the density gradient
column. To obtain a completely amorphous sample, the polymer has to be molten
and quenched to low temperatures with effectual celerity.
3.3.5 Wide Angle X-ray Scattering
Another method to determine the degree of crystallinity is the wide angle X-ray
scattering technique. X-ray structure analysis is in general the most common
method to measure crystal structures. When we took the atoms in the crystal as
points in a real space lattice and transform the lattice into reciprocal space, the
distance between two atoms in real space is now the distance it d between lattice
planes. When a monodisperse, coherent X-ray beam is scattered on parallel lattice
planes, the scattered beams interfere constructively when
2d sin / ¼ nk ðn ¼ 1; 2; 3; . . .Þ: ð3:6Þ
/ is the glancing angle between the incident beam and the scattering plane and
k the wave length of the X-ray beam. Equation 3.6 is the Bragg condition for
X-ray scattering. When scanning a range of scattering angles at some distinct
angles, where the Bragg condition is fulfilled, the constructive interferences occur
as sharp crystalline peaks in the scattering spectrum. In partially crystalline
samples the X-ray spectrum at wide angles will consist of some distinct sharp
peaks but also of an underlaying broad halo, which originates from the amorphous
parts of the sample. Integrating these two areas in the spectrum gives access to the
degree of crystallization, which is
Ac
wc ¼ : ð3:7Þ
Ac þ Aa
With Ac and Aa representing the areas under the curve of the crystalline signals and
the amorphous halo. The discrimination of both parts is highly subjective, which
makes the results less accurate [21].
3.3.6 UV/VIS Spectroscopy
Ultraviolet–visible spectroscopy (UV/VIS) is the spectroscopy of photons in the

UV and visible region, usually the probing light has wavelengths between about
42 3 Methods
200 and 1000 nm. In this region of the electromagnetic spectrum, molecules
undergo electronic transitions. The use of UV/VIS spectroscopy in polymer
science lies mostly in the measurement of the concentration of a polymer in
solution. If the solution is diluted, the Lambert–Beer law is valid:
A ¼ logðI0 =IÞ ¼ cL ð3:8Þ
A is the measured absorbance, I0 is the intensity of the incident and I the intensity
of the transmitted light. is the extinction coefficient, a material constant, L is the
pathlength through the sample and c the concentration.
Most UV/VIS spectrometers work with two cuvettes, one with the sample, the
other one as reference containing only the pure solvent, which are compared for
every wavelength. All spectrometers consist of a light source, a dispersive element
(lattice, prism, monochromator filters), the sample, and a detector. Mostly ava-
lanche photo diodes are used. To obtain the spectrum the wavelength range of
interest is scanned and absorbance (or transmittance) is plotted against wavelength.
For structures with periodic pattern of the dielectric constant, e.g., colloidal
crystals, light is diffracted on these pattern with subsequent constructive and
destructive interference. That is electromagnetic waves of distinct wavelengths
cannot pass such structures in distinct crystallographic directions, dependent from
the structure’s periodicity length. This effect is referred to as photonic Bragg gap.
The corresponding wavelength can be calculated as
kBragg ¼ 2sneff sin H; ð3:9Þ
where s is the periodicity in the regarded crystallographic direction, neff is the

effective refractive index, and H is the angle between the incident light and the
scattering plane of the crystal. For a fcc colloidal crystal with sphere diameter d,
Eq. 3.9 becomes for the (111) plane of the crystal
kBragg ¼ 1:63dneff sin H; ð3:10Þ
with
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
neff ¼ /s n2s þ /m n2m ; ð3:11Þ
where /s and ns, or /m and nm are the volume fraction and the refractive index of
the spheres or the surrounding matrix, respectively [24].
In this thesis UV/VIS is used to characterize colloidal mixtures. The mea-
surements have been performed by Retsch on the setup at MPIP.
3.3.7 Gel Permeation Chromatography
The gel permeation chromatography (GPC) is a size exclusion technique to

determine the distribution of the molar mass in polymers. The principle idea is that
a diluted polymer solution travels a column containing a macroporous gel. If the

single molecules are large they cannot penetrate the holes in the gel and therefore
are not retarded, i.e., all molecules that are larger as a certain value, dependent
from the nature of the gel, eluate together with the eluent. Smaller particles are
retarded by entering the holes. With decreasing particle size the retardation time
increases up to a certain point from which the eluation time is constant for all
polymer chains, i.e., only polymers within a certain range of dimension are sep-
arated. Behind the column the amount of polymer in the eluted liquid as a function
of time is recorded using different techniques as for example refractometry or
UV/VIS spectroscopy. The intensity of the signal is proportional to the amount of
eluted polymer. The easiest and most usual way to obtain quantitative data from
the elution time versus intensity diagram is the comparison with monodisperse
polymer standards, if available. In principle it is also possible to use an universal
calibration for all polymers, because the size exclusion depends not really on the
molecular weight but on the hydrodynamic radius Rh of the polymer molecule in
solution. Rh correlates with the molecular weight M:
Rh ¼ U½gM ð3:12Þ
with U being a constant and [g] being the intrinsic viscosity given by the Mark–
Houwink equation
ag
M
½g ¼ Kg ; ð3:13Þ
g=mol
where Kg and ag are parameters depending on the polymer, its shape, and the
solvent. The details of the sophisticated universal calibration is given in standard
textbooks of macromolecular chemistry and physics [21].
3.4 Theoretical Calculations
The vibrations of individual spheres in different matrices as well as the phononic

band diagram in mesostructured media is widely discussed in terms of experi-
mental results mainly by BLS in this thesis. However, in order to come to valueble
conclusions, the most powerful approach is to compare the experimental data with
data calculated within the appropriate theories leading to either scattering cross
sections when talking about the vibrations of individual particles or phononic band
structures when talking about the phononic dispersion relation of the investigated
systems. Taking well-known parameters for the calculations allows certainty about
the nature of the experimentally observed effects, while on the other hand the use
of theoretical important parameters as free fit parameters can be utilized to extract
(e.g., nanomechanical) information from the measured spectra.
In this section I want to briefly describe the calculation methods used in this
thesis. Starting from the considerations in Sect. 2.3.3 in Chap. 2, I will introduce
the scattering cross section obtained from the scattering of a plane wave on a
44 3 Methods
single spherical scatterer. From that I will describe the related plane-wave method
to calculate the band structure of periodic composites. Anyhow, the more powerful
method to do the latter and more important for this thesis is the (layer) multiple-
scattering approach, which will be presented concludingly. Finally, the finite
element approach is discussed very briefly.
3.4.1 Single-sphere Scattering Cross-section Calculations
To calculate the scattering cross-section of a single sphere [25], one can calculate
the scattering of an incident plane longitudinal wave of the form
X 1
inc L
u ðrÞ ¼ clm r½Rl ðkl rÞYlm ð^rÞ ð3:14Þ
lm
ql
or of a transverse wave of the form

X
inc M N i
u ðrÞ ¼ clm Rl ðkt rÞYlm ð^rÞ þ clm r Rl ðkt rÞYlm ð^rÞ ð3:15Þ
lm
qt
in analogy to Eq. 2.111 after decomposition into longitudinal and transverse

components. This plane wave is propagating in a homogeneous medium. When the
wave is incident on a solid homogeneous sphere, it is scattered by it, and the wave
field in the matrix um now consists of the incident and the scattered field, uinc and
usc : That is
um ¼ uinc þ usc : ð3:16Þ
The boundary conditions at the surface of the sphere, given above in Eqs. 2.107
and 2.108, must still be fulfilled and can be written now as
ui jr¼rs ¼ uinc jr¼rs þ usc jr¼rs ð3:17Þ
and
si jr¼rs ¼ sinc jr¼rs þ ssc jr¼rs ; ð3:18Þ
with
X
usc ðrÞ ¼ L
dlm llm ðR; kli Þ þ dlm
M
mlm ðR; kti Þ þ dlm
N
nlm ðR; kti Þ: ð3:19Þ
lm
It is shown that the coefficients for the incident wave cxlm (with x = L, M, N) are
known a priori [25]. As discussed in Sect. 2.3.3 in Chap. 2, it is possible to
decompose this system into six scalar equations that can be separated into an
infinite number of equations that are functions of lm. When the coefficients cxlm are
known, there remain only the coefficients axlm, given in Eq. 2.109, and dxlm, given in
3.4 Theoretical Calculations 45
Fig. 3.4 Total dimensionless

scattering cross-section for a
polystyrene sphere in liquid
PDMS matrix (cf. Fig. 5.11 in
Chap. 5, cf. Fig. 3 in Ref.
[14]) for longitudinal incident
wave. The peaks are labeled
as spherical harmonics with
l = 2, 3, 4, 1, 5, n = 1 for
all resonances. The elastic
values used for the calcula-
tion are given in Table 5.4
in Chap. 5
Eq. 3.19 for the field inside the spheres or the scattered field, respectively.
Therefore exists for each lm a system of six equations with six unknowns that can
be solved. The detailed algebraic analysis is done for example in Ref. [25].
The scattering cross-section r is a measure for the likelihood of the physical
interaction between an incident particle and another particle—in this case, the
scattering of a wave by the single scatterer. It is defined as the ratio of the scattered
energy flux to the incident energy flux per unit surface [26].
For the discussed case of an incident longitudinal wave (subscript l) the
dimensionless scattering cross-section for a sphere with radius r is shown to
become [27]:
" m N 2#
rl X1 L 2
jdlm j k r jdlm j
¼ 4ð2l þ 1Þ m 2 þ lðl þ 1Þ lm : ð3:20Þ
pr 2
l¼0 jkl rj kt r jktm rj2
Figure 3.4 shows an example of a total dimensionless scattering cross-section,

calculated for a polystyrene (PS) sphere in a poly(dimethyl siloxane) (PDMS)
matrix. The normalization over the frequency is carried out by comparison with
the longitudinal phase velocity in the matrix cm
l .
3.4.2 The Plane Wave Method
As pointed out before (and more widely discussed in Chap. 5), in phononic sys-
tems consisting of many elastic scatterers in a matrix with different elastic prop-
erties, band gaps may occur in the band diagram, originating from either the
periodicity of scattering layers (Bragg gap, BG) or from localized states in indi-
vidual scatterers. The localized modes in the individual scatterers are given by the
maxima in the scattering cross-section diagram discussed in Sect. 3.4.1, and it
seems plausible to extend the method presented for the scattering cross-section
to calculate the band diagram of phononic systems by assuming a plane wave
46 3 Methods
scattered now by many scatterers. In order to obtain analytical solutions, the plane
wave (PW) method is restricted to periodic systems, i.e., phononic crystals [28].
The initial point of the PW approach is the general wave equation for a medium
locally isotropic
i
o2 u i 1 o oul o ou oul
¼ k þ l þ i; l ¼ 1; 2; 3: ð3:21Þ
ot2 . oxi oxl oxl oxl oxi
Here, ui,l and xi,l are the Cartesian components of the displacement vector uðrÞ and
the position vector r; respectively; kðrÞ; lðrÞ; and .ðrÞ are the Lamé coefficients
and the local mass density, respectively, which are periodic functions of r with
periodicity lattice vector R:
f ðr þ RÞ ¼ f ðrÞ: ð3:22Þ
As a result of the common periodicity of the three coefficients in Eq. 3.21, its
solutions can be chosen to satisfy Bloch’s relation
uðrÞ ¼ eikr uk ðrÞ; ð3:23Þ
where k is a vector in the reciprocal lattice restricted within the first Brillouin zone
(BZ) and uk ðrÞ is a periodic function. Hence, it is possible to expand f ðrÞ in a
three-dimensional Fourier series:
X
f ðrÞ ¼ fG eiGr ; ð3:24Þ
G
where the summation is over all reciprocal vectors G; which can be written as sum
with integer coefficients of the orthonormalized vectors that span the three-
dimensional vector space in which R is the sum of the basis vectors.
With that one can rewrite Eq. 3.23 after expanding uk in Fourier series as
X iðkþGÞr
uðrÞ ¼ uk þ Ge : ð3:25Þ
G
The substitution of Eq. 3.24 (with f ¼ k; l; .1 ) and Eq. 3.25 into Eq. 3.21
delivers finally the expression:
"
X X
2 i
x cukþG ¼ .1 0 00
GG00 kG G ðk þ G Þl ðk þ G Þi
00 0
G0 l;G

þlG00 G0 ðk þ G0 Þi ðk þ G00 Þl ulkþG0 ð3:26Þ
! #
X Xh i
1 0 00 i
þ .GG00 lG00 G0 ðk þ G Þj ðk þ G Þj ukþG0 :
G00 j
If the Fourier series in Eq. 3.25 is performed over M reciprocal vectors (i.e., for
M scatterers in the periodic medium), Eq. 3.26 is reduced to a 3M 9 3M matrix
eigenvalue equation for the 3M unknown coefficients uikþG ði ¼ 1; 2; 3Þ: Usually,

M [ 400 to achieve the desired convergence [28].
Note that for fluid systems, Eq. 3.26 can be simplified [27]. In liquids l = 0;
by introducing the pressure p ¼ kru; Eq. 3.21 can be written in the form
h i
kðrÞr .ðrÞ1 rpðrÞ þ x2 pðrÞ ¼ 0: ð3:27Þ
Following the same considerations as for the (general) solid case one obtains for
M terms in the Fourier sum a M 9 M system.
PW is a fast and easy-to-apply method to calculate the band diagram in fluid/
fluid or solid/solid systems, i.e., fluid or solid scatterers in matrices of the same
aggregate state. However, it fails when dealing with solid/fluid systems, which are
mostly discussed in this thesis.
When a solid scatterer is embedded in a liquid host, transverse waves cannot
propagate in the matrix and Eq. 3.27 becomes the appropriate elastic wave
equation. In that case the M eigenmodes for M scatterers correspond to purely
longitudinal waves. However, it is known that even for a longitudinal incident
wave, the field inside the scatterer will be both longitudinal and transverse,
i.e., there are localized transverse modes inside the scatterer that cannot propagate
[27, 29].
The approach that uses the full (including l 6¼ 0) wave equation (Eq. 3.21) fails
because of the non-propagating character of these modes and leads to no con-
vergence at all. Using Eq. 3.27 would ignore completely the transverse component
of the wave within the scatterer as it would de facto replace the solid scatterer by
the fluid scatterers of the same k and l. Although it leads to mathematically
reasonable results, it was shown that these results are not suited to describe the
reality [27].
Another limitation of the PW method is that it can be only applied on infinite
periodic samples. This also means that it is unable to calculate the transmission
properties. An approach to overcome these limitation is the multiple-scattering
method that will be discussed in the next section.
3.4.3 The Multiple Scattering Method
The main idea behind the multiple scattering method is described schematically in
Fig. 3.5. A plane wave in a homogeneous medium impinges on a system of N non-
overlapping scatterers (n = 1, 2, N). The wave that impinges on each scatterer
consists of N contributions, the incident wave and the outgoing waves from any of
the N - 1 other scatterers.
With the coefficient notation used in the sections above, for a single scatterer
the coefficients for the scattered wave field, dxlm, are related to the given coeffi-
cients (i.e., amplitudes) of the incident plane wave, cxlm:
48 3 Methods
x2
x dlm
c lm x3
dlm
t,x1
clm
x4
x6 dlm
dlm
incident
x5
plane wave dlm (x=L,M,N)
Fig. 3.5 Schematic draw of the multiple scattering mechanism. A plane incident wave (with
coefficients cxlm) is impinging an ensemble of scatterers, which can have arbitrary shape. The total
wave field acting on the n-th individual scatterer (with coefficients ct,xn
lm ), e.g., the scatterer 1 in the
middle, is a summation over the impinging plane wave and the scattered waves from the other
scatterers (with coefficients dxn
lm; in our example for n = 1)
X
x
dlm ¼ Txlm;x0 l0 m0 cxlm : ð3:28Þ
x0 ;l0 m0
The T-matrix is a 3 9 3-Tensor that can principally be calculated for scatterers

of every arbitrary shape with no need of homogeneity. The calculation for
homogeneous spheres can be found in literature [25] as well as the solutions for
inhomogeneous core–shell particles [30].
Coming back to our system with N scatterers, the total wave field incident on
the n-th scatterer is the sum of the incident plane wave and the wave field scattered
from all scatterers n0 6¼ n:
X
uinc;n ðr rn Þ ¼ uinc ðr rn Þ þ usc ðrn rn0 Þ: ð3:29Þ
n0 6¼n
That is
X
uinc;n ðr rn Þ ¼ ct;Ln t;Mn t;Nn
lm llm ðR; kl Þ þ clm mlm ðR; kt Þ þ clm nlm ðR; kt Þ
lm
X
¼ cLlm llm ðR; kl Þ þ cM N
lm mlm ðR; kt Þ þ clm nlm ðR; kt Þ ð3:30Þ
lm
X L;n0
M;n 0
N;n 0
þ dlm llm ðR; kl Þ þ dlm mlm ðR; kt Þ þ dlm nlm ðR; kt Þ
n0 6¼n;lm
or
X 1
uinc;n ðr rn Þ ¼ cLlm r½Rl ðkl rÞYlm ðr rn Þ
lm
kl
þ cM lm r ½ðr rn ÞRl ðkt rÞYlm ðr rn Þ

N 1
þ clm r r ½ðr rn ÞRl ðkt rÞYlm ðr rn Þ
kt
X L;n0 1 ð3:31Þ
þ dlm r½Rl ðkl rÞYlm ðrn rn0 Þ
lm;n0
kl
0
M;n
þ dlm r ½ðrn rn0 ÞRl ðkt rÞYlm ðrn rn0 Þ

0 1
N;n
þ dlm r r ½ðrn rn0 ÞRl ðkt rÞYlm ðrn rn0 Þ :
kt
In the last two equations the two sums on the right side belong to the incident plane
wave (with coefficients cxlm, x = L, M, N) and to the scattered waves from the
x;n0
other scatterers (with coefficients dlm ).
For the resulting total incident field the coefficients ct,xn
lm are introduced in
Eq. 3.30. They are related to the scattering coefficients of the n-th scatterer dx,n
lm by
the T-matrix (Eq. 3.28). hP
lmax
Doing so for all N scatterers, one obtains a system of N l¼0 ð2l þ 1Þ þ
P max i
2 ll¼1 ð2l þ 1Þ algebraic equations, in which the only known amplitudes are
those for of the incident plane wave. With the assumption that terms with l [ lmax
do not contribute significantly to the spherical wave expansion, this system can be
x;n
solved numerically to determine all dlm :
So far, the method is not restricted to an equal shape of all scatterers (however,
for different shapes the T-matrix of all individual scatterers would have to be
known) or to any periodicity, as the relative position of each pair of scatterers is
considered in Eq. 3.31.
However, as the aim is to calculate the band structure in phononic crystals as
function of the reduced wave vector k in the first Brillouin zone, the introduction
of the crystal periodicity is needed. It was shown that in this case the phononic
band structure can be calculated in analogy to the theory of Korringa [31], Kohn
and Rostoker [32] (KKR theory) developed to solve the Schrödinger equation for
electromagnetic waves in periodic lattices [33].
After introducing the pressure field
pðrÞ ¼ kr uðrÞ; ð3:32Þ
it is possible to write the wave equation for a periodic medium of solid scatterers in
a fluid matrix as follows [33]:

2 x2 2 1 1 1
r pðrÞ þ pðrÞ þ x 2 pðrÞ þ .ðrÞ r rpðrÞ ¼ 0: ð3:33Þ
cm c ðrÞ c2m .ðrÞ
50 3 Methods
This can be written in analogy to the Schrödinger equation in the form

Hm ðrÞpðrÞ þ UðrÞpðrÞ ¼ 0; ð3:34Þ
where Hm ðrÞpðrÞ ¼ 0 represents the wave equation for the matrix without scatterer
(Hm ðrÞ ¼ r2 þ x2 =c2m ). In a periodic system a Green’s function approach can be
chosen to reformulate Eq. 3.33.
Z
pðrÞ ¼ Gðr r0 ÞVr0 pr0 dr0 ð3:35Þ
V
with Green’s functions

X
Gðr r0 Þ ¼ eikRn G0 ðr r0 Rn Þ ð3:36Þ
n
and
0
1 eijrr jx=cm
G0 ðr r0 Þ ¼ : ð3:37Þ
4p jr r0 j
Rn is a Bloch’s vector, so that
pðr þ Rn Þ ¼ eikRn pðrÞ; ð3:38Þ

and the relation between the amplitudes of the scattered field at different lattice
sites n and n0 becomes [26]
0
x;n x;n
dlm ¼ eikRn dlm : ð3:39Þ
The next step is the expansion of Gðr r0 Þ and pðr0 Þ into spherical functions of
r and r0 : These calculations are performed in detail in Ref. [33]. In the end the
final multiple-scattering equation appears to be
X x

¼ Alml0 m0 =ðdl10 Þd ll mm al0 m0 ¼ 0:
0 d 0 ð3:40Þ
l0 m0
cm
The coefficients Alml0 m0 are the so-called structure constants as they depend on k; x,
and the periodic lattice structure, the coefficients dl0 relate the incident to the
scattered field.
Equation 3.40 represents a linear homogeneous algebraic system. Its nontrivial
solutions give the eigenfrequencies of the periodic system and hence the dispersion
relation.
An enhanced variant of the multiple-scattering method calculates the dispersion
relation for samples consisting of different composites with two-dimensional
periodicity and is called layer-multiple-scattering approach (LMS) [31, 35, 36].
For a slab parallel to a distinct crystallographic plane, the reduced vector parallel to
this plane, kk ; is usually a conserved quantity. Therefore, LMS searches in each
individual slab propagating Bloch waves for given x and kk ; which are the
eigenmodes of the elastic field in that slab. It is an on-shell method since it operates
at a given frequency. It was shown that LMS is a powerful method to determine the
dispersion relation as well as the transmittance of three dimensionally structured
systems. Example 3D phononic crystals can be regarded as a succession of planes of
scatterers parallel to a chosen crystallographic plane. As already mentioned for the
general MS approach, the LMS technique takes the full vector nature of the acoustic
field into account and is therefore not limited to certain fluid/solid combinations, as
for example PW is.
Note that nearly all calculations in this thesis using the single-sphere scattering
cross-section and the multiple scattering algorithms have been performed by my
co-worker Revekka Sainidou (University of Le Havre).
3.4.4 The Finite Element Method
The finite element method (FEM) is a numerical method to approximately solve

partial differential equations. It is widely used in engineering and many other fields
to simulate a wide range of different physical problems.
The main principle of the technique is that the treated body is initially dis-
cretized into a finite number of elements of finite size, which can be described by a
finite number of parameters. Within these finite elements, basis functions are
defined. Substitution of the linear combination of all utilized basis functions into
the differential equation to solve for each individual element (e.g., for mechanical
problems the equation of motion) in combination with the boundary conditions
leads to a quadratic matrix, whose dimension is given as the product of the number
of elements and the number of basis functions. This dimension is also the number
of degrees of freedom of the system. The boundary conditions at the knots of
neighbouring elements are often evident, e.g., in the case of a mechanical defor-
mation, the displacement (and even the first derivative of the displacement) must
be continuous at the boundary between two elements. Finally, a system of ordinary
differential equations is obtained that can be solved directly, e.g., using Gaussian
elemination. It is obvious that the quality of the approximation increases with the
number of elements, as well as the effort for the calculation does.
In this thesis, FEM is utilized only marginally to display some eigenvibrations.
These simulations have been performed by myself using the COMSOL Multi-
physics software package.
References
1. Davis KE, Russel WB, Glantschnig WJ (1989) Science 245:507

2. Trau M, Saville DA, Aksay IA (1996) Science 272:706
52 3 Methods
3. Jiang P, Bertone JF, Hwang KS, Colvin VL (1999) Chem Mater 11:2132
4. Gu ZZ, Fujishima A, Sato O (2002) Chem Mater 14:760
5. Fustin CA, Glasser G, Spiess HW, Jonas U (2004) Langmuir 20:9114
6. Denkov N et al (1992) Langmuir 8:3183
7. Wang L, Baowen L (2008) Phys World 21:27
8. Tommaseo G et al (2007) J Chem Phys 126:014707
9. Wang J, Li Q, Knoll W, Jonas U (2006) J Am Chem Soc 128:15606
10. Fustin C-A, Glasser G, Spiess HW, Jonas U (2003) Adv Mater 15:1025
11. Pusey PN, van Megen W (1987) Phys Rev Lett 59:2083
12. García P, Sapienza R, Blanco A, López C (2007) Adv Mater 19:2597
13. Li Q et al (2008) Porous networks through colloidal templates topics in current chemistry.
Springer, Heidelberg
15. Zakhidov AA et al (1998) Science 282:897
16. Li Z-Y, Zhang Z-Q (2000) Phys Rev B 62:1516
17. Li ZY, Zhang ZQ (2001) Adv Mater 13:433
18. Haginoya C, Ishibashi M, Koike K (1997) Appl Phys Lett 71:2934
20. Spahn P (2008) PhD thesis
21. Pursiainen OLJ et al (2008) Adv Mater 20:1484
22. Lechner MD, Gehrke K, Nordmeier EH (2003) Makromolekulare Chemie, 3rd edn.
Birkhäuser, Basel
23. Linderstrom-Lang K (1937) Nature 139:713
24. Toth H, Fehlauer H (2004) Phys Unserer Zeit 35:76
25. Gaillot DP, Graugnard E, King JS, Summers CJ (2007) J Opt Soc Am B 24:990
26. Psarobas IE, Stefanou N, Modinos A (2000) Phys Rev B 62:278
27. Cheng W (2007) Hypersonic elastic excitations in soft mesoscopic structures. PhD thesis
28. Kafesaki M, Economou EN (1995) Phys Rev B 52:13317
29. Economou EN, Sigalas M (1994) J Acoust Soc Am 95:1734
30. Einspruch N, Truell R (1960) J Appl Phys 31:806
31. Sainidou R, Stefanou N, Psarobas IE, Modinos A (2005) Comp Phys Commun 166:197
32. Korringa J (1947) Physica 13:392
33. Kohn W, Rostoker N (1954) Phys Rev 94:1111
36. Sainidou R, Stefanou N, Psarobas IE, Modinos A (2005) Z Kristallogr 220:848
Chapter 4
The Vibrations of Individual Colloids
4.1 Introduction
Nowadays, colloidal nano- and mesoscale particles with dimensions from few
nanometers up to 1 lm are used in a growing number of applications, e.g., as
fillers in polymer thin films to enhance thermo-mechanical properties [1], to
improve coatings performance and as components in nanocomposites operating as
photonic [2], plasmonic [3], and phononic structures [4]. For a wide range of such
applications, information on the mechanical properties and the stability of these
colloidal composite materials are of paramount importance. Conventional rheo-
logical measurements on the macroscopic system are often not sufficient to elu-
cidate the specific contributions of the nanostructured components. At the
nanoscale, forces negligible in macroscopic systems, such as depletion, interfacial,
and confinement effects often become significant, and the behavior of the same
materials in nanoscopic systems can considerably deviate from the bulk. A fun-
damental understanding of transport and thermomechanical properties of nano-
structured materials is a precondition to address a specific need by structural
engineering. In the case of colloidal composite materials, the vibrational modes
confined to the individual particles result from the elastic motion at the nanoscale
and should sensitively depend on the geometrical, architectural, interfacial, and
mechanical characteristics of the particles. However, there is a paucity of non-
destructive experimental techniques to probe this ‘music’ of particle vibrations
since both high frequency resolution and sensitivity are required to detect the
numerous eigenmodes. Raman scattering [5, 6] has been utilized to measure few
eigenfrequencies of nanoparticles with dimension below 10 nm, whereas Brillouin
light scattering (BLS) [7–9] and optical pulse-probe techniques [10–12] can probe,
respectively, the spontaneous and stimulated vibrations confined in sub-microm-
eter particles. In the latter technique, the excited acoustic oscillations are observed
in the form of modulations of the transient reflectivity of the probe laser, and hence
the particles must possess good reflectance, e.g., by introduction of gold shells. In
BLS, light is scattered inelastically by the density fluctuations (phonons)

DOI: 10.1007/978-3-642-13483-8_4, Springer-Verlag Berlin Heidelberg 2010
54 4 The Vibrations of Individual Colloids
associated with these particle localized modes at thermal equilibrium and there are
no further stringent conditions.
Self-assembly of colloidal particles in periodic structures [13, 14] (cf. Sect. 3.1
in Chap. 3) has received special attention as the resulting photonic and phononic
crystals have revealed the potential of manipulating the propagation of electro-
magnetic and elastic waves. Their propagation is forbidden at ‘Bragg’ frequencies
or wavelengths commensurate with the lattice constant, which for sub-micrometer
particles is comparable with the wavelength of the visible light. Synthetic opals
from these particles can exhibit dual, i.e., hypersonic phononic and photonic
[14, 15] band gaps allowing for acousto-optical interactions [16]. Moreover, the
design of sub-micron particles that can act as strong localized resonant elements in
an appropriate matrix provides the possibility for additional gaps well below the
Bragg frequency, termed hybridization gaps (cf. Chap. 5) [17, 18]. The opening of
band gaps perturbs the phononic density of states which impacts physical quan-
tities such as group velocity, heat capacity and heat conductivity in dielectrics
being potentially useful for thermoelectric devices [19, 20]. One of the pivotal
concerns for these systems is the phonon dispersion, which is essentially defined
by the elastic parameters of the constituent components and the spatial architecture
of the composite system. Colloid science can create novel materials that possess
spatial variation of density and elastic constants at the nanoscale but their
mechanical characterization remains difficult.
If vibrations are excited in a finite elastic body, it can act as an elastic resonator.
The elastic standing waves in stress-free boundary are referred to vibration
eigenmodes. In the case of elastic spheres, the analytical solutions were first
derived by Lamb [21]. The eigenmodes of spheres can be classified as torsional
and spheroidal modes, both labeled by three indices (nlm), which describe the
angular (lm) and radial (n) dependence of the displacement. Spheroidal resonance
modes are fully characterized by the angular momentum l, imposed by the
spherical symmetry of the particle, and n, where n denotes the n-th order solution
for a given l (cf. Sect. 2.3.3 in Chap. 2) [21, 22]. The use of inelastic light
scattering to measure vibration eigenmodes of small spherical particles was first
performed experimentally by low-frequency Raman scattering (RS) [22]. Due to
selection rules only two distinct vibration eigenmodes contribute to the RS of
spherical particles with diameter (d) much smaller than the wavelength of the
probing light (d k) [23]. For bigger spheres with d k, Brillouin light
scattering (BLS) in the GHz-range becomes the technique of choice [7, 8, 24]. Due
to the consideration of higher-order terms in the electric multipole expansion and
of retardation effects [25, 26], BLS can resolve a multitude of eigenmodes.
For homogeneous transparent systems, BLS measures the spectrum of light
inelastically scattered by the acoustic phonons with a selected polarization
[longitudinal (l) or transverse (t)] and wave vector q, leading to spectra consisting
of doublets at xB = ±cq around the central elastic Rayleigh line. Since c in soft
materials like polymers is of the order of 103 ms-1 and q is in the range
1-30 lm-1, the frequencies f = xB/2p fall into the GHz range. This yields the
4.1 Introduction 55
two elastic constants C11 ¼ .c2l and C44 ¼ .c2t with cl and ct being the two phase
sound velocities and . being the mass density. For inhomogeneous turbid systems,
e.g., powder of mesoscopic (d k) particles, q is ill-defined in the BLS experiment
due to strong multiple scattering. BLS can measure only localized in space (and
hence q-independent) vibrational modes. Each resonance mode appearing at
frequency f(n, l) is characterized by the angular momentum l of the n-th order. The
frequencies of the individual vibrational modes depend on their rigidity, mass
density, and size dimensions of the particles. For the case of homogeneous
spherical particles the frequencies are given by Lamb in Eq. 2.106.
The frequencies can be theoretically obtained from the calculated density of
states (DOS) spectra of a single sphere as a function of the two elastic constants
and the inverse diameter. For polystyrene spheres, e.g., the constant in Eq. 2.106
becomes x(1, 2) & 0.85 and R(cl, ct) = C44 with no adjustable parameter [8].
In this chapter the state-of-the-art for BLS measurements on homogeneous
mesoscopic spherical particles is briefly summarized. Novel studies dealing with
mixtures (‘hybrids’) of different kind of spheres are presented as well as a study on
spheres prepared as copolymers with different compositions to elucidate the
influence of the next neighbors and of the rigidity on the mechanic vibrations in
the mesoscale. The second part of the chapter extends the scope on nanostructured
colloids. Hybrid material spherical core–shell particles are investigated as model
systems. Therein, especially the influence of heat on polymer cores contained in a
hard silica-shell and the influence of the composition in the vice versa case of
silica–PMMA core–shell particles with different ratio of core size to total diameter
on the vibrational eigenmodes is of interest.
Due to some theoretical support, these studies give an unprecedentedly com-
prehensive picture of the elastic properties of individual colloids, although still
numerous questions have to stay open—e.g., when dealing with structures going
beyond the investigated ‘easy’ spherical model systems.
4.2 Elastic Vibrations in Homogeneous Polymer Colloids
4.2.1 The ‘music’ of the spheres
As pointed out in Sect. 2.3.3 in Chap. 2 and in the Introduction of this chapter,
BLS can be utilized to measure the resonance modes of dry non-transparent col-
loidal crystals. Due to the strong elastic form factor of the individual spheres and
the large elastical contrast with the surrounding air, the opals show strong multiple
scattering. In such samples the inelastic scattering from localized modes leads to
incoherent BLS in analogy to the Raman scattering.
Thus BLS can be utilized to analyze the particle eigenfrequencies, describing
the spheroidal (n, l)-modes, with n as the n-th mode of the l-th spherical harmonic.
The first demonstration of the feasibility of the BLS experiment was shown by
Penciu et al. in the case of dilute suspensions of giant core–shell micelles [9]. Few
years later Kuok et al. have extended this application to closely packed mono-
disperse silica nanospheres in air [7, 27]. Up to six localized particle eigenmodes
have been resolved out of the numerous possible modes, probably due to the weak
scattering of moderately compressible silica. In a subsequent study, artificial soft
colloidal crystals, composed of monodisperse submicrometer polystyrene (PS)
spheres with diameter d between 170 and 856 nm have been investigated [8]. Up
to 21 q-independent eigenmodes have been resolved.
Figure 4.1 shows exemplary three BLS spectra taken from colloidal PS opals
with diameters 180, 360 and 550 nm [24]. The resonance frequencies scale with
1/d (Fig. 4.2), which is in perfect agreement with Lamb’s theory [21] and the the-
oretical predictions based on single-phonon scattering cross-section calculations
(cf. Sect. 3.4.1 in Chap. 3) [28, 29]. In the computations, a plane sound wave
propagating in air and impinging upon a single PS sphere was considered and after
subtracting the scattering amplitude for a rigid sphere of equal size, the sphere
eigenmodes appear as resonance peaks in the plot of scattering cross-section
versus frequency. Thereby the resonance frequencies f(n, l) can be identified as
mode with angular momentum quantum number l of n-th order.
Using the experimental values for the longitudinal sound velocity
cl = 2350 ms-1, the transverse sound velocity ct = 1210 ms-1 and mass density
. ¼ 1050 kg m3 of bulk PS, all resolved frequencies are quantitatively captured
within 3% with no adjustable parameter. The product of frequency and diameter is
Fig. 4.1 Left SEM—images of colloidal PS crystals with d = 180, 360 and 550 nm (top to
bottom); right corresponding q-independent BLS eigenmode spectra of the PS opals. To capture
all possible vibrations, two spectra recorded at two different free spectral ranges are superim-
posed. Spectra are recorded at 20 (q = 0.0041 nm-1). For the thickest particles (d = 550 nm,
bottom) there is a clear cut-off bump around 15 GHz, which is the frequency of the acoustic
phonon in PS at backscattering geometry [24]
4.2 Elastic Vibrations in Homogeneous Polymer Colloids 57
Fig. 4.2 Experimental reso-

nance frequencies of the PS
opals from Fig. 4.16 (solid
symbols) versus the recipro-
cal particle diameter. The
solid lines denote the com-
puted resonance frequencies
[8, 24]. The open squares
denote the frequencies for the
first seven modes simulated
by the finite element methode
for PS spheres with
d = 540 nm
obtained to be a constant for every mode, as is indicated by the solid lines in

Fig. 4.2, which represent the computed modes (n, l). It is shown that the first
modes are the same for all samples with the (1, 2)-mode being the lowest and most
intense one.
From the longitudinal and transversal sound velocities the elastic constants are
directly accessible. These are the Poisson ratio r, the Young modulus E, and the
shear modulus G (Sect. 2.1.1 in Chap. 2) [30]. Additionally to the elastic constants,
even more information can be extracted from these spectra. The peculiar line shape
can be regarded as a sensitive index of the particle size distribution. An application
with technological relevance was the determination of the mechanical properties
of spherical glassy CaCO3-particles [30]. The particles with mono-modal size
distribution and diameters in the range between 400 and 1500 nm were made from
amorphous CaCO3, with unknown elastic properties as yet. Eigenmode acoustic
spectra of the spheres were recorded and six or seven resonance modes could be
resolved. By calculating cl from the cut-off frequencies and taking ct and the
density . as floating fit parameters, all signals could be assigned to the vibrational
eigenmodes and a very good agreement between theory and experiment was
achieved. From the obtained values for cl, ct and . the Poisson ratio r, Young
modulus E and shear modulus G were then reported. In conformity with the
analysis of the SEM-pictures all investigated colloidal samples in this thesis are
found to be quite monodisperse.
For the largest particle size in Fig. 4.1 (or Fig. 2 in [8]) there is clearly a kind of
cut-off frequency at about 14–15 GHz after which the scattering intensity
decreases rapidly towards zero. Because of strong multiple scattering, q is not
defined. The highest frequency contribution corresponds to the acoustic phonon in
bulk PS under backscattering conditions, i.e., qBS = 4pn/k0. Using n = 1.59 for
the refractive index of PS and for the laser wavelength k0 = 532 nm,
qBS = 0.0376 nm-1, and hence a cut-off frequency for the longitudinal acoustic
phonon (cl = 2350 ms-1) is indeed expected at 14.05 GHz.
A more detailed theoretical treatment of the Brillouin and Raman scattering

from the acoustic vibrations of spherical particles with diameters in the order of
magnitude of the wavelength of light was presented very recently by Montagna
[25]. While for very small particles (qR 1, with R = d/2) only the ‘Raman-
term’ that originates either from local field changes due to dipole induced dipoles
or from electronic polarizability changes with the change of the atomic distances is
important, for particles with diameters d k0 also the ‘Brillouin-term’ becomes
important. The latter comes from the polarization fluctuations caused by spacial
displacement of scatterers by acoustic vibrations. Detailed calculations are per-
formed, which show that for a given l the intensity of a resonance mode with a
certain n strongly depends on qR, hence on the particle size. However, a direct
comparison with the experiment is not straightforward due to the ill-defined q.
Nevertheless, it is pointed out that for larger particle sizes the range of qR also
increases (0 B qR B qBSR). In a given range of qR only the modes resonance
frequencies that are close to that of the acoustic phonon in the bulk, i.e., x & qcl,
contribute considerably to the Brillouin scattering. The total number of modes with
higher n and l that have their intensity maximum in the larger qR-range also
increases. This rationalizes the observation of the highest number of modes for the
better-formed cut-off for the largest particles in Fig. 4.1. It should be noted that
when calling the resonance modes ‘q-independent’, in principle the frequency of
the mode is indeed q-independent but the intensity is theoretically not. The reason
why in the BLS measurements in this section the spectra are found to be inde-
pendent from the angle is that obviously all qs 0 B q B qBS contribute equally to
the multiple scattering. In Sect. 4.2.4, an experiment is presented that shows the
q-dependence of the scattering intensity of individual modes.
However, it was not settled whether selection rules apply for the BLS from
particles with diameters comparable to the wavelength of the incident laser light
(qd & 1). While Li et al. [26], based on group theory considerations, claim that
only particle resonances with even ls contribute to BLS eigenmode spectra,
according to Montagna [25] resonances with even and odd ls should have com-
parable activities. In fact, the latter is supported by a BLS study [8] of model PS
spheres revealing up to 21 resonance modes theoretically captured with no
adjustable parameter.
Moreover, utilizing a finite element method (FEM, cf. Sect. 3.4.4 in Chap. 3),
I could validate the assignment for the first six modes with n = 1 and l = 2–7,
which are shown as open squares in Fig. 4.2 in perfect agreement with the
scattering cross-section calculations and the experimental frequencies. Figure 4.3
shows the FEM result for the lowest (1,2)-eigenmode of a PS sphere with
d = 540 nm. Using the same parameters as in the cross-section calculations
(cl = 2350 ms-1, cl =1200 ms-1, . ¼ 1050 kg m3 ) [8], the (1,2)-mode is found
at 1.87 GHz, while the experimental mode was measured at 1.94 GHz.
Figure 4.3a shows the deformated surface of the sphere, while subfigures b–d
show the deformated intersecting planes in the planes of a Cartesian coordinate
system. The color scale corresponds to the total local displacement. In the
Fig. 4.3 FEM simultation for the (1,2)-eigenmode of a PS sphere with d = 540 nm at
1.87 GHz; a total displacement (color scale) and deformation of the surface, b–d displacement
and deformation of intersecting planes in the three planes of a Cartesian coordinate system.
Deformations are amplified by a factor of 20
intersecting planes, a displacement symmetric to the center of the spheres is

visible, which is also the only knot of the displacement function, i.e., n = 1. The
shape of the deformation, which is amplified by a factor of 20 in order to clarify
the direction of the deformation, assigns the angular momentum quantum-
number l; in each intersecting plane the deformated sphere has two maxima and
two minima, i.e., l = 2. (Note: The thin black lines denote the undeformated
sphere.) Analog to the identification of the (1,2)-mode, also the modes with
higher l can be found from the FEM simulation. It is noteworthy that there are a
higher number of modes calculated following the FEM approach than seen in the
BLS experiment. However, most of these modes are torsional modes, i.e., the
displacement is fully tangential. Such modes do not cause inelastic light
scattering, only spheroidal modes, where the displacement is also (or fully)
radial, contribute to the light scattering spectrum [23, 25, 26]. By choosing only
the spheroidal modes out of the simulated eigenmodes, those with higher l are
easily obtained from the FEM results. Checking the displacement and the
deformation in the intersecting planes as in Fig. 4.3 allows the correct deter-
mination of n and l. The shape of the surface for the modes with n = 1 and
l = 3–7 is shown in Fig. 4.4 together with the intersecting planes for l = 3; for
the higher l the intersecting planes look analog to those for l = 2 and l = 3 with l
deformation maxima in each plane. Of course even higher order modes can be
Fig. 4.4 FEM simulation for the higher l eigenmodes (n = 1). Deformations are amplified by a
factor of 20. Top total displacement (color scale) and deformation of a the surface and b–d the
intersecting planes of the (1,3) spheroidal mode. Bottom displacement and deformation of the
surface for modes with l = 4–7. For a PS sphere with d = 540 nm the frequencies are found to be
2.79 (1,3), 3.58 (1,4), 4.32 (1,5), 5.03 (1,6), and 5.74 GHz (1,7) in good agreement with the
experimental data (Fig. 4.2)
found by FEM, however, with increasing order the assignment becomes more
and more complicated, since in the calculation mostly asymmetric mixed modes
appear. Theoretically, with much finer mesh, all modes can be assigned clearly.
Anyhow, even the six modes shown here (in perfect accordance with another
calculation method) proof that there is no limitation on even ls as claimed by Li
et al. [26], since the assignment of the (1,3) and the (1,5)-mode with no fitting
parameter is clear and without any alternative. On the other hand, the FEM
simulation of the eigenvibrations of silica spheres with d = 360 nm, whose
eigenmode spectrum is shown in Ref. [26], allows other assignments for the first
four modes than that chosen by the authors, who restricted themselves on even
ls, utilizing the elastic parameters given by the same authors experimentally
(ct = 2520 ms-1, cl = 3960 ms-1, . = 1960 kg m-3) [31]. The mode at
8.76 GHz [assigned as (1,0)] could be the (1,3)-mode (8.60 GHz by FEM), the
mode claimed to be the (1,6)-mode at 14.4 GHz could be a double signal
containing the modes (1,5) and (1,6), found by FEM to appear at 13.14 and
15.29 GHz, respectively. For higher orders, the assignment is more difficult,
however, the mode at 17.65 GHz [claimed to be (2,4) or (3,2)] could be
identified as the (1,7)-mode, calculated to appear at 17.4 GHz.
It should be noted that deviating from Ref. [24] in Fig. 4.2 also the breathing
mode (1,0) is shown as a theoretical fit based on FEM calculations, again in perfect
agreement with the scattering cross-section method [32]. It is found that for the
case of PS spheres the (1,0)-mode is not far away from the (1,4). In fact, the third
signal (from low to high frequency) for the PS spheres in Fig. 4.1 seems slightly
broadened, which might be related to a weaker signal from the (1,0)-mode.
4.2.2 The Influence of the Neighbors: Mixtures
The assignment of the two phononic gaps is also corroborated by their sensitivity on
the disorder. In this context, non-crystalline colloidal films were prepared by vertical
lifting deposition of ‘hybrids’, binary mixtures consisting of an equal number of two
PS spheres with different diameter (d = 300 and 360 nm). The size polydispersity is
then artificially increased and no crystallization takes place, which is affirmed by
SEM-pictures. The eigenmode acoustic spectra of these dry hybrid films and the
dispersion relations in their infiltrated counterparts were measured by BLS. The
eigenmode spectrum of the 1:1 300:360 nm hybrid is shown in Fig. 4.5 along with
the eigenmode spectra of the individual one component opals. Interestingly, the
spectrum of the hybrid is a superposition of the individual opals as is indicated by the
vertical lines denoting some exemplary resonance frequencies either from the small
or the big spheres. Moreover, hybrids of different relative ratios of the constituent
spheres were prepared. The intensity of the signals originating from a particular
particle size relates to its composition in the mixture. In Fig. 4.5, this linear
composition dependence is demonstrated by superimposing the spectrum of the 1:1
300:360 nm hybrid with the spectrum of a 3:1 300:360 nm hybrid, normalized to the
peak intensity in the 360 nm spheres. For clarity a baseline correction was
introduced into the Stokes-side of the hybrids (inset to Fig. 4.5).
Fig. 4.5 Left eigenmode acoustic spectra of the PS opals with d = 300 nm (blue), d = 360 nm
(red) and the symmetric PS hybrids 1:1/3:1 300:360 nm (black/green)(bottom to top). The ver-
tical lines (dotted for d = 300 nm, dashed for d = 360 nm) denote some exemplary eigenmodes
appearing in the opals and in the hybrids. In the top right corner the Stokes-sides of the hybrids’
spectra are compared after baseline correction to better visualize the influence of the composition
on the relative intensity of the individual signals. Right corresponding SEM-images of opals and
hybrids shown on the left side [24]
Apparently, but somewhat counterintuitive, the contact between the spheres

does not influence their eigenmodes—at least for non-infiltrated dry films. We also
examined only partially ordered samples of monodisperse nanospheres prepared
by dropping a few drops of the suspension of the spheres onto a glass substrate. In
spite of the lack of any order, the eigenmode spectra of these samples are indis-
tinguishable from those of the colloidal crystals. To undermine the contact argu-
ment, the consistency of the spectra’s features would be based upon the small
differences between the opals and the hybrids for 300 and 360 nm PS spheres, also
very asymmetric hybrids with a nine to one number ratio of 180 and 550 nm PS
beads were investigated. While for every dense packing of spheres, the number of
neighboring spheres and hence the contact points as indicated by the SEM-images
is 12, this mean number is different for the random packed hybrids depending on
their size disparity and their number ratio. For example, in the SEM image for a
9:1 180:550 nm hybrid in Fig. 4.6, the number of next neighbors of the large
spheres is increased dramatically by small beads that accumulated around the
bigger spheres. However, even in this extreme case, the resonances from the
550 nm spheres are virtually unaltered as demonstrated in the spectra of Fig. 4.6.
A careful inspection, however, of the two lowest frequencies reveals a very small
shift of the hybrid resonance modes towards higher frequencies as is shown in the
inset of Fig. 4.6 for the Stokes-sides of the spectra. Though the observed effect is
subtle ( 4% frequency shift) but significant. In this context it is noteworthy to
mention that in Fig. 4.3 of Ref. [33], Li et al. compare the BLS spectra from a
320 nm silica artificial opal and from a single sphere of the same kind, measured
by a micro-Brillouin light scattering setup. In this figure it seems as if there is a
shift between single spheres and opal. Interestingly, the authors describe only the
broadening in the opal due to the polydispersity.
Fig. 4.6 Left comparison between the eigenmode acoustic spectra of the 550 nm PS single opal
and the highly asymmetric 9:1 180:550 nm PS hybrid. Right SEM-image of the 9:1 180:550 nm
PS hybrid [24]
In summary, the number of next neighbors does not influence significantly the
eigenvibrations of the individual spheres. The effect of the disorder on the
dispersion relation of our PS hybrids after infiltration is discussed in Sect. 5.1.3
(in Chap. 5).
4.2.3 The Influence of the Rigidity: Copolymers
While there are several BLS studies dealing with the size-dependence of the
eigenmodes of homogeneous spheres from different materials (Sect. 4.2.1) [7, 8, 24],
where the elastic properties are calculated from the experimental results, as well as
two more elaborate studies on core–shell particles, presented also in this thesis
(Sect. 4.3) [34, 35], a systematic experimental investigation of the influence of the
particle rigidity on the eigenvibrations of mesoscopic copolymers is still missing.
After Lamb (Eq. 2.106)

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
Rðcl ; ct Þ
f ðn; lÞ ¼ xðn; lÞ d;
.
with x(n, l) being a temperature-independent constant for each individual mode

and R(cl, ct) being the rigidity, which is a function of longitudinal and transverse
sound velocity cl and ct [21]. The frequencies can be theoretically obtained by
single-phonon scattering cross section calculations (Sect. 3.4.1 in Chap. 3) or finite
element modeling (Sect. 3.4.4 in Chap. 3). Here, BLS is utilized in a set of nearly
uniformly sized (d & 200 nm) polymeric colloids consisting of poly(methyl
methacrylate) (PMMA) copolymerized with poly(n-butyl acrylate) (PnBA) in
different compositions in order to investigate the development of the elastic
constants as a function of the copolymer composition. Therefore, bulk and indi-
vidual colloidal properties are measured by BLS and consulted in combination to
obtain insight into the effect of copolymerization on the mechanical characteristics
in such small spheres. Due to the clearly differing rigidity of the homopolymers at
room temperature, which is confirmed by the large difference in the glass transi-
tion temperature Tg [Tg(PMMA) & 110 C, Tg(PnBA) & -43 C], the softness of
the copolymer increases rapidly with the amount of n-butyl acrylate used in the
synthesis.
Copolymerization of two distinct monomers permits the combination of their
individual properties and, hence, allows control over various physical quantities,
such as Tg. Random radical copolymerizations of MMA and nBA was conducted
by my colleague Retsch in order to tune the Tg of the resulting colloid. The
composition of nBA in the copolymer was targeted to 10, 20, and 30 wt% through
selection of the weight ratio of the two monomers in the initial emulsion.
A random copolymerization is indicated by the Q and e values of both monomers
in terms of the Q-e-schema of Alfrey and Price (nBA: Q = 0.38, e = 0.85; MMA:
Q = 0.78, e = 0.40) [36, 37]. A homogeneous copolymerization is indicated by the
Fig. 4.7 DSC data for the

four compositions at the sec-
ond heating run, i.e., bulk
properties. Tgs are denoted by
vertical lines. The inset shows
Tg as a function of the
composition together with a
Fox–Flory fit and a Gordon–
Taylor fit from Ref. [38]
very similar r values (nBA: r = 0.94; MMA: r = 0.91) not very different from
r = 1.
Differential scanning calorimetry (DSC) was used to determine the glass
transition temperature Tg. The results of the heating period of the second run, i.e.,
after heating initially to 200 C (and therefore bulk properties), are shown in
Fig. 4.7. Tg decreases rapidly with increasing amount of nBA. Using the value
measured for pure PMMA and the literature value for nBA (-43 C) [38], allows
an excellent representation of the glass transition temperature as a function of the
composition by the Fox–Flory equation [39]
1 / /
¼ PMMA þ PnBA ð4:1Þ
Tg Tg;PMMA Tg;PnBA
with no fit parameter (inset). The Gordon–Taylor fit also shown in the inset is
borrowed from Ref. [38], including the empirical parameter K = 0.82, however,
the small deviations originate partially in different values measured for pure
PMMA. Our Tgs could be perfectly represented with Tg(PMMA) = 113 C and
K = 0.65. In addition to the single Tg, dielectric spectroscopy has been performed
by my colleague Mpoukouvalas, which showed a single a-relaxation for each
sample, i.e., both kind of segments feel the same energy landscape.
The density . of the copolymers was determined by a density gradient column
using aqueous calcium nitrate solutions [40]. The experimental results shown in
Fig. 4.8 conform to a weight average density, using for the density of the
homopolymers .PnBA ¼ 1:035 g cm3 and .PMMA ¼ 1:195 g cm3 .
The size of the spheres is ascertained by scanning electron microscopy (SEM).
The average molecular weight Mn and the polydispersity index PDI of the
polymers were measured relative to PMMA standards by gel permeation
chromatography (GPC). Crystallinity was excluded by small angle X-ray
scattering measurements. The particle properties are itemized in Table 4.1.
Fig. 4.8 Determination of

mass density as a function of
the composition by a density
gradient column for the two
lightest samples. The squares
are well-known calibration
standards. The inset shows a
fit for all four compositions
Table 4.1 Particle and material properties of the investigated spheres obtained by SEM (d), GPC
(Mn, PDI), DSC (Tg) and density gradient column(.)
nBA (wt%) d (nm) Mn (g mol-1) PDI Tg (C) . (g cm-3)
0 232 92,500 4.2 113 1.195
10 214 89,500 4.2 92 1.179
20 200 84,200 3.2 68 1.163
30 204 110,000 3.5 50 1.147
By vertical lifting deposition [41] non-transparent films of the colloids with

thickness of about 30–50 lm were prepared by Retsch. The BLS eigenmode
spectra were recorded in transmission geometry at (non-arbitrative, due to strong
multiple scattering) scattering angle of 50. Figure 4.9 shows the experimental
spectra. Because of the slightly differing particle diameters given in Table 4.1 and
the d-1-dependence of the modes given in Eq. 2.106 the frequency axis is nor-
malized to the product of frequency and diameter. The eigenmodes correspond to
the maxima of the spectra, all four spectra can be represented by five Lorentzian
signals, denoted by the dotted vertical lines on the anti-Stokes side of the spectra.
However, the fourth spectral line is much broader than the others.
Obviously, with increasing amount of nBA (bottom to top) the modes are
shifted towards lower frequencies. This correlates with the decreasing rigidity in
Eq. 2.106 caused by the softening in the copolymer. Notably, the change in the
density alone (ca. 4%) would lead to a frequency shift in the opposite direction.
According to Eq. 2.106, the frequencies of the eigenmodes are functions of
three parameters: cl, ct, and . or, alternatively, Young’s modulus E, Poisson’s ratio
r, and .. Since the mass density is experimentally known, the adjustable param-
eters are the two sound velocities.
For the bulk copolymers, BLS can, in principle, deliver cl and ct by recording the
dispersion relations for polarized and depolarized light scattering, respectively.
Transparent samples were prepared by heating some of the vertical lifted opals by a
heat gun (T & 100–200 C) for a few seconds, which results in a melting of the
Fig. 4.9 Experimental BLS

eigenmode spectra with fre-
quency f normalized with the
particle diameter d. The
dotted vertical lines on
the anti-Stokes side denote
the positions of the five
resolved maxima for the pure
PMMA spheres [42]
mesospheres. The polarized spectra delivered cl, but the intensity of the theoretically
accessible ct in the depolarized spectra was not strong enough to determine a single
signal. The measured cls are presented in Table 4.2, together with the value of a bulk
PnBA sample.
Previous studies on homogeneous mesoscopic polymer and silica beads show
that there should be no significant change in the longitudinal sound velocity that
has to be taken into account in the colloids compared to that in the bulk material
[8, 24, 31]. Since the rigidity R for the lower frequency eigenmodes is found to be
a much stronger function of ct, i.e., shear modulus G ¼ .c2t , than of cl, i.e.,
longitudinal modulus M ¼ .c2l (cf. Sect. 2.1.2 in Chap. 2), it is meaningful to take
the bulk longitudinal sound velocities also as given in the description of the
vibrational modes.
Table 4.2 shows the decrease of cl, i.e., softening, in the copolymers with
increasing amount of nBA. However, this decrease is small compared to the
experimental value of cl in the two homopolymers. In Sects. 5.1.2 and 5.2 in
Chap. 5 of this thesis, several approaches to describe the sound propagation in an
effective medium composed of (at least) two mechanically different materials are
discussed. It turns out that the simple expression
Table 4.2 Experimental and theoretical bulk cl for homopolymers and copolymers at room
temperature
nBA (wt%) cl(BLS) (ms-1) cl(Wood) (ms-1) cl(Wood) (ms-1)
cl,PnBA = 2200 ms-1
0 2,755 2,755 2,755
10 2,678 2,569 2,663
20 2,615 2,421 2,584
30 2,498 2,300 2,515
100 1,835 1,835 2,200
1 / 1 /1
¼ 1þ ; ð4:2Þ
Meff M1 M2
which is known as Wood’s law [43], is in many cases a good approximation to
give an effective modulus Meff as a function of the components individual moduli
and the composition. If one assumes Wood’s law to be, in principle, a good
approximation also for copolymers, the theoretically expected values for cl can be
easily calculated, as shown in the second column of Table 4.2. Obviously, the
presented approximation strongly overestimates the softening due to the nBA in
the copolymers, leading to a deviation of up to 8% for 30 wt% nBA. It will be
shown in Sect. 4.3.1 that the mechanical moduli are temperature dependent,
strongly decreasing at temperatures above Tg. At room temperature, all sound
velocities except for bulk PnBA are measured in the glassy state, i.e., for purely
elastic response. In contrast, bulk PnBA at room temperature is in the rubbery
regime, i.e., it behaves viscoelastically and hence its cl is lower than for a glassy
PnBA. In other words, the nBA segments in the three glassy PMMA–r-PnBA
copolymers assume a dense packing and therefore should display elastic response
due to the very slow (essentially frozen) dynamics. The success of Eq. 4.2 in
representing the experimental cl values of the copolymers is optimized if a fictive
cl = 2200 ms-1 is used for the bulk PnBA, i.e., by introducing an ‘effective glassy
longitudinal sound velocity’. The third column in Table 4.2 shows the good
representation by Wood’s law, using this artificially increased value.
Returning to the eigenmode spectra in Fig. 4.9, the fixation of d, ., and cl to
their experimental values reduces the number of adjustable parameters in the
single phonon scattering calculations (Sect. 3.4.1 in Chap. 3) or FEM (Sect. 3.4.4
in Chap. 3) into one (ct).
Scattering cross-section calculations were performed by my coworker Sainidou
[29, 44]. Utilizing reasonable values for ct, the first three modes could be clearly
assigned as the (1,2), (1,3), and (1,4)-mode, in analogy to the results found in
polystyrene and silica samples in Sect. 4.2.1. However, the fourth signal, which is
much broader than the others, cannot be assigned to the expected (1,5)-mode, since
it appears at higher frequencies. Indeed, its frequency is found between those of
the (1,5) and the (1,6)-mode, indicating that the signal is a superposition of both
modes. This explains also the broad appearance of the signal, although a fit with
two Lorentzians does not converge. Therefore, instead of pretending a meaningful
fit of two individual modes, one must be content to show the signal in between
these two modes. The ‘fifth’ mode, on the other hand, can be assigned as the
(1,7)-mode with acceptable accuracy, by choosing ct given by the mean square fit
of only the first three modes. In fact, for this last mode, the deviation from the
theory decreases with increasing amount of nBA in the copolymer.
All experimental modes are shown in Fig. 4.10 together with the fitted values
for fd, using ct as the only floating parameter. The linear fits have no theoretical
meaning, however, they follow the trend of most signals quite well, although the
change between 0 and 10 wt% does not follow the linear trend for the first two
modes. For the (1,7)-mode (‘mode 5’), however, the change between 20 and
Fig. 4.10 fd against the

amount of nBA in the
copolymer individually for all
four samples for each of the
five observed signals and in
the bottom right panel toge-
ther for all signals together.
The number of the mode
corresponds to the number
given in Fig. 4.9 next to the
dotted vertical lines. The
open blue circles belong to
the theoretically obtained
frequencies for the (1,l)-
modes with l = 2–7 [full
circles for the (1,6)-mode],
denoted in the top right cor-
ners of each panel. The linear
fits in red are guides for the
eyes. Error bars are 2% [42]
30 wt% is quite large regarding the virtually unaltered fd-values between pure
PMMA and 20 wt% nBA. Correspondingly, the relative deviation between
experiment and theory is largest (&4%) for the (1,7)-mode and pure PMMA. On
the other hand, a linear fit through only the 20 and 30 wt% points captures the
calculated points with a deviation \2%.
From the experimentally obtained material parameters, in combination with the
fitted values for ct, it is possible to calculate the elastic moduli of the mesospheres
and by that to quantify their softening due to the addition of nBA. The calculated
values for the elastic moduli as well as for Poisson’s ratio can be found in
Table 4.3.
As expected, all moduli go down with increasing amount of nBA, indicating the
strong softening effect (by &20% between pure PMMA and the softest copolymer).
The more striking feature is the constance of Poisson’s ratio r. In a copolymer with
increasing amount of a rubber-forming monomer (and consequently strongly
decreasing Tg), one could expect a more rubber-like behavior of the copolymer, i.e.,
an increasing r. At room temperature, i.e., about 30 C below the lowest Tg,
however, this is not the case. This finding means that the compressibility
Table 4.3 Fitted ct, elastic moduli (Young’s, longitudinal, and shear modulus), and Poisson’s
ratio
nBA (wt%) ct (ms-1) E (GPa) M (GPa) G (GPa) r
0 1,530 7.14 9.07 2.80 0.277
10 1,520 6.88 8.46 2.72 0.262
20 1,466 6.35 7.95 2.50 0.271
30 1,394 5.68 7.15 2.23 0.274
characteristics at room temperature are virtually unaffected by the softening, there is

no shift towards rubber-like behavior below Tg. In turn, this result corroborates the
‘trick’ applied in the beginning of this section, when introducing the effective glassy
sound velocity for PnBA at room temperature to fit Wood’s law.
In summary, this section contains a systematic study on the softening of
mesoscopic PMMA spheres by copolymerization with nBA with the help of BLS
bulk and eigenmode measurements. The elastic moduli are quantified. The Poisson
effect is virtually unaffected by the softening. This is described by an ‘effective
glassy sound velocity’ of the homopolymer PnBA at a temperature at which pure
PnBA is a rubber.
4.2.4 The Influence of the Wave Vector: Suspensions
In Sect. 4.2.1, I pointed out that, theoretically, the scattering intensity of the
individual modes should be strongly q-dependent, however, due to the strong
multiple scattering all qs corresponding to values between zero and backscattering
contribute equally to the spectrum and hence the intensity distribution is the same
independent from the scattering angle h. Based on symmetry arguments, Montagna
showed that for a given n the BLS intensity of modes with increasing l are only
active for increasing values of qd [25]. In fact, for a given system, the intensity
maximum of each mode (n, l) should appear at a different q, i.e., for a transparent
system that allows the observation of eigenmodes at different scattering angles h,
while the angle of maximum intensity hmax increases with l.
It was shown first in 2003 by Penciu that BLS can be used to measure the
eigenmodes of silica colloids suspensions in a refractive index matching liquid
(cyclohexane/decalin) [29, 45]. By measuring the suspension in an NMR-tube the
sample can be regarded as quite thick, which allows to see also eigenmodes that
are not resolved anymore in the systems presented in Chap. 5. It is shown that the
appearance of the eigenmodes depends on the filling fraction of the silica spheres
in the suspension. In fact, quite high filling fraction are needed to resolve the
eigenmodes at all. Those have been achieved by centrifugation of the suspension at
the bottom of the NMR-tube. The graphs in Fig. 7 of Ref. [29] show the resolved
modes as a function of qd for two different silica particles with diameter
d = 328 nm and 250 nm, respectively, at three different filling fractions (/ = 0.3,
0.62, and 0.67). For all systems, an effective medium acoustic branch is observed
that will not be further regarded in this section. Besides that, there are up to four
modes that appear at the same frequency for each qd, resonance modes of the silica
spheres. While at / = 0.3 only one such mode is found, in the polycrystalline or
glassy samples with significantly higher / more modes can be distinguished.
Although in Ref. [29] some exemplary spectra at different qd give evidence that
the q-dependence of the BLS intensity can be shown by these samples, a sys-
tematic investigation is still missing. Therefore, in this section the results of BLS
measurements on silica particles with d = 375 nm in an index matching matrix of
liquid ethoxy-ethoxyethyl acrylate (SR256) are presented in order to bridge this

gap of knowledge. The sample was synthesized by my coworker Kiefer at DKI,
Darmstadt, with an initial filling fraction / = 0.35. The dispersion relation for this
filling fraction is presented in Sect. 5.4.1 (in Chap. 5). Here, I will focus on the
polycrystalline sample that was obtained from the dispersion by centrifugation for
90 min at 4000 rpm (/ & 0.70–0.74).
Figure 4.11 shows BLS spectra taken from the polycrystalline sample at seven
different angles (3.3 B qd B 11.6) for two different free spectral ranges. The
eigenmodes are marked by vertical lines, the dispersion relation is shown in
Fig. 4.12. While for the two lowest eigenmodes the deviation from the mean varies
by about ±1 GHz, the distribution of frequencies for the other resonance modes is
very narrow.
Fig. 4.11 Exemplary BLS (a)

spectra of polycrystalline
silica colloids (d = 375 nm)
in SR256 for seven different
scattering angles, corre-
sponding to 3.3 B qd B 11.6,
at two different free spectral
ranges (a 40 GHz,
b 15 GHz). The vertical lines
denote the approximate
frequencies of the
eigenmodes
(b)
Fig. 4.12 Eigenmode fre-

quencies as a function of qd
extracted from the spectra in
Fig. 4.11. The dashed lines
represent the mean frequency
for each eigenmode. The
solid line shows a linear fit
of the longitudinal acoustic
branch (open circles)
(cf. Sect. 5.4.1 in Chap. 5)
Though, when carrying out a more detailed investigation of the spectra in

Fig. 4.11, it becomes clear that the relative intensities of the individual resonances
vary systematically with the scattering angle h. While the lowest frequency
eigenmode around 5 GHz is strongest at low h, the highest frequency modes
appear strongest at high angles, qualitatively conforming the theoretical predic-
tions [25]. The lowest mode is strong at h = 30 (qd = 3.3), weak at 90
(qd = 9.0) and nearly disappears at 110 (qd = 10.4). The second mode around
9 GHz only appears between 50 and 90, while it is invisible at lower and higher
qd. Also the higher frequency modes could be only foreboded in the spectrum at
30. At 50 and 70 the third and fourth mode start to appear weakly, the fifth
mode quite feeblish at 70. Especially for the fourth and fifth mode, intensity
increases clearly up to 110, while the difference between 110 and 130 is only
small.
Although a full theoretical description of the q-dependence is missing for the
present system, the qualitative agreement of the experiment with the predictions is
a striking hint of their correctness.
4.3 Elastic Vibrations in Nanostructured Colloids
4.3.1 The Influence of the Temperature: PS–SiO2 Core–Shell

Particles
Core–shell particles exhibit properties that may be substantially different from

those of the templated core. These include their density, mechanical stability, and
optical characteristics. Proper design of core and shell material provides the
possibility to optimize the properties of core–shell particles according to their

applications, thus making them attractive from both a scientific and a technological
point of view. Silica shells are desirable because of excellent biocompatibility and
easily tunable surface functionality [46, 47]. Furthermore, silica shells are much
stiffer compared to most organic shells [48, 49]. Here the rigidity of thin glassy
silica shells is utilized for the realization of shape persistent polymer cores at
temperatures well above the polymer glass transition temperature Tg.
In this section BLS is employed to investigate the shape-persistence of spherical
core–shell colloids consisting of a soft polystyrene (PS) core with d = 400 ±
12 nm and a hard silica (SiO2) nanoshell with thickness L = 37 ± 3 nm against
heating; this size was determined by scanning electron microscopy, averaging over
about 100 spheres. The synthesis of the core–shell particles by soap-free emulsion
polymerization of the PS core (slightly copolymerized with divinylbenzene) and
subsequent coating by silica is described elsewhere [49]. Here, it is shown that a
relatively thin SiO2 shell acts as a ‘nano-armor’ of the polymer core even in the
molten state above its glass transition temperature while the particles retaining
remarkably high mechanical strength.
Herein, the strong localization of the vibration modes in core–shell particles
with high elastic contrast is utilized to demonstrate the shape robustness of
polymer based colloidal particles upon heating well within the fluid phase of the
core. Both sound velocities (cl, ct) are sensitive index of Tg [50–53], due to the
rapid increase of compressibility (C-111 ) and decrease of shear rigidity (C44). Hence,
tracking the frequency of a distinct vibration mode (Eq. 2.106) with increasing
temperature senses the decrease of the mechanical strength and marks the glass
transition in the f-T diagram if the particle shape is preserved. Thus, BLS
measures single sphere mechanical properties.
For the BLS experiment, an opaque dry film of randomly packed PS–SiO2 with
a total thickness of several tens of microns was used [24]. Figure 4.13 shows high
resolution BLS eigenmode spectra of the PS–SiO2 and the parent plain PS spheres
(d = 400 nm) recorded at room temperature. Both spectra exhibit several clearly
distinguishable modes, with the lowest frequency mode being the strongest
spectral line. The comparison of the line shapes of the two spectra shows that the
spectrum of the core–shell particle is very akin to that of the bare PS spheres. The
low frequency eigenmodes are therefore core-like, i.e., the signals are related to
vibrations localized in the PS core. In the inverted case of core–shell particle with
hard silica core and soft polymer shell only shell-like modes were identified [34].
Both cases can be explained by the fact that the BLS signal increases with the
compressibility of the medium in which the elastic energy of the mode is localized,
i.e., the scattering from a hybrid particle structure is dominated by the contribution
of the soft component.
To verify that the softening of the core is monitored while the shape of the
particle remains unaltered, I followed the T-dependent eigenmode spectra,
focusing on the first (and strongest) signal of the core–shell particles (arrow in
Fig. 4.13). This is assigned to the (1,2)-mode by comparison with the resonance
modes of the computed DOS spectra as reported for soft (PS, PMMA) and hard
4.3 Elastic Vibrations in Nanostructured Colloids 73
Fig. 4.13 Eigenmode

spectra of the uncoated
d = 400 nm PS spheres
(black, top) and of the
PS–SiO2 core–shell particles
(grey, bottom) at room
temperature. The (1,2)-mode
(arrow) is the strongest [35]
(silica) spheres [7, 8, 24]. Due to the small expansion coefficient of fused silica
(5.5 9 10-7 K-1), the diameter d in Eq. 2.106 is virtually temperature
independent.
The heating experiment was performed at a non-arbitrative scattering angle of
50 while the sample was encased in a glass windowed oven (cf. Fig. 2(b) in
Ref. [51]). The sample was heated from room temperature to 165 C in steps
between 30 (low T) and 10 K (high T) within a few minutes for each step. After
equilibrating the sample for 15 min at each temperature, the eigenmode spectrum
was recorded for about 15 min probing always the same spot ( 50 lm). This
ensures that even the intensities of the main modes can be traced as a function of
temperature. Figure 4.14a shows eigenmode spectra at different temperatures upon
heating and after cooling over a narrower frequency range than in Fig. 4.13, in
order to further boost the resolution. All recorded spectra have been represented in
the range of the (1,2)-signal by a Lorentzian using its frequency, line width, and
amplitude as adjustable parameters. The (1,2)-mode is red shifted by &0.2 GHz at
165 C relative to its frequency at room temperature either before (bottom) or after
(top) heating, indicating a small decrease of shear modulus (C44). Note that the
same mode undergoes a much smaller shift at 70 C.
The experimental frequencies f(1,2) shown as a function of temperature in
Fig. 4.15 are well represented by two distinct straight lines, expectedly both lines
with negative slopes due to decreasing rigidity. The characteristic kink in the f-T
diagram signifying the transformation of a glassy into a rubbery state occurs at
107 ± 3 C, which is slightly higher than the Tg of bulk amorphous PS (100 C)
due to the crosslinking of the PS core with divinylbenzene (DVB) [54]. The
presence of a single Tg in Fig. 4.15 indicates a homogeneous, over few hundreds
nanometers, core environment and hence this value of Tg is further confirmed by
differential scanning calorimetry (DSC) measurements on the bare PS cores and
the PS–SiO2 particles. In both systems Tg was found to amount to 107.5 ± 3 C,
whereas in the absence of crosslinking Tg lowers to &100 C (see Fig. 4.15b), in
(a) (b)
Fig. 4.14 a Exemplary eigenmode spectra during heating and after cooling (side arrows). The
solid lines in the Stokes side of the spectra indicate the representation of the two low frequency
modes by Lorentzian spectral lines (individual modes and sum). The dashed line indicates the
position of the (1,2)-mode at room temperature. b SEM pictures of the core–shell particles before
and after heating to 165 C and cooling back to room temperature [35]
agreement with DSC. The clear detection of the glass transition affirms the validity
of the obtained findings.
The relative drop of f(1,2) between room temperature and Tg is less than 2% in
the PS–SiO2 particles and more than 6% in the bare PS cores (cf. Fig. 4.16). In the
rubbery state of the core, f(1,2) at the highest examined temperature (165 C) is
decreased relatively to its value at Tg only by about 6%. Assuming a constant
density, this f(1,2) value corresponds to a shear modulus c44 & 1.45 GPa in the
confined polymer melt at about Tg + 60 K. For comparison, a corresponding
decrease in the transverse sound velocity of PS films is about 40% (i.e.,
c44 \ 1 GPa) [51]. This remarkable small temperature effect on the rigidity of the
present three-dimensionally confined PS–SiO2 particles as a result of the increased
pressure in the core is unprecedented.
Even at temperatures far above Tg, the spectra (Fig. 4.14a at 165 C) still
possess the typical shape for T \ Tg with minor changes in the intensity. This
Fig. 4.15 a The variation of (a)

the frequency of the (1,2)
vibration eigenmode (arrow
in Fig. 4.13) of the PS–SiO2
core–shell particle with
temperature indicating two
linear regimes (solid lines).
The shaded area embraces the
glass transition temperature
of the PS core. b Experiment
analogue to a, but with no
crosslinker (DVB) in the PS
core (b: dPS = 480 + 60 nm
SiO2-shell) [35]
(b)
proofs that even in the rubbery regime the PS core keeps its spherical shape. The
thermal expansion of the PS core is expected to increase by more than a factor two
above Tg and hence the pressure on the silica shell [55]. Especially at temperatures
well above Tg, it is this pressure that could, in principle, break the shell. However,
when heating up to 165 C, the intensity of the observed mode changes only very
little within the experimental error. After cooling back to room temperature, not
only the frequency (dashed line in Fig. 4.14) but also the intensity and the line-
width (&0.39 GHz full width half maximum) assume their initial values before
heating. Thus the thin silica shell can stand the higher pressure under these con-
ditions and prevents the PS core to change its shape. This notion is corroborated by
the scanning electron microscope (SEM) photographs of the core–shell particles
before and after the heating cycle to 165 C shown in Fig. 4.14b. The lack of
visible defects after the heating procedure confirms the results obtained by the
BLS. Actually, the silica shell acts as an armor for the PS core.
In Fig. 4.16, the corresponding BLS experiment on bare PS particles is pre-
sented. These eigenmode spectra are very sensitive to the particle shape. For the
bare PS particles f(1,2) displays stronger decrease with temperature than for the
Fig. 4.16 Eight exemplary

spectra for BLS eigenmode
measurements on the bare
d = 400 nm PS particles as a
function of temperature: the
two lowest resonances are
fitted by Lorentzians. The
dotted line denotes the center
of the lowest eigenmode at
room temperature. Measure-
ments above 105 C are per-
formed with a new sample
(ns) rapidly heated to these
temperatures. The inset in the
top left shows the f–T dia-
gram for T B Tg in analogy
to Fig. 4.15 [35]
PS–SiO2 core–shell particles (inset to Fig. 4.16) due to the absence of the
confinementand partially to the now unsuppressed expansion of the PS spheres,
i.e., decrease of c44 and increase of d in Eq. 2.106. Already below Tg the line shape
of the (1,2)-mode changes and above Tg the signal vanishes completely within a
few minutes. Starting at room temperature, for increasing temperature a shift to
lower frequencies and a broadening/splitting of the signal is observed. The spectral
shape of the (1,2)-mode (around 2.5 GHz) in the BLS spectra of the bare PS core
in Fig. 4.16 severely changes as the temperature increases towards Tg. At
temperatures above 85 C, the single Lorentzian shape initially splits into two
through the appearance of a second peak at the high frequency side. Finally, above
103 C the intensity decreases rapidly and eventually vanishes at 108 C within a
few minutes. Apparently, shape alterations start already below Tg because of
surface melting and breaking the particles spherical symmetry. With the onward
and patchy destruction of the spherical shape, the resonance modes of the unarmed
PS particles finally disappear, and the destruction of the spheres is, of course,
irreversible. SEM pictures show only an undefined surface. Probing the shape of
the PS mesospheres on the average during the melting process is another inter-
esting but demanding task. In the context of the present study, it would require
finite element modelling of all eigenfrequencies supported by a temperature
dependent SEM or atomic force microscopy.
In summary, the morphological and thermo-mechanical response to heating of

PS–SiO2 core–shell particles at submicrometer length scales is probed by BLS that
delivers quasi in situ insight into the individual colloids. By monitoring the
eigenmode spectra, it was shown that silica shells of a few tens of nanometers
protect a polymer core from changing its spherical shape even at temperatures well
above Tg, which can be determined from the same experiment. The shape per-
sistence of these hybrid systems is accompanied by a remarkable enhanced rigidity
of the confined core above its Tg. The inevitable change of the amorphous packing
inside the rigid nanocontainers will likely prove useful for fundamental studies.
4.3.2 The Influence of the Components: SiO2–PMMA

Core–Shell Particles
Colloid science can create novel materials that possess spatial variation of density
and elastic constants at the nanoscale but their mechanical characterization
remains difficult. Similarly, the influence of the particle architecture (core–shell
spheres or hollow capsules), size, and shape on the eigenfrequencies, and the
localization of elastic energy in specific regions of an individual particle is not
known. In addition to the characterization on the nanomechanical properties, tai-
lored acoustic confinement will be important for precise phonon management by
structural engineering.
In this section, BLS is employed for the first measurement of the resonant
modes (the ‘music’) in sub-micron core–shell spheres [silica-poly(methyl meth-
acrylate)—SiO2–PMMA] having constant core radius and varying shell thickness
and in the corresponding spherical PMMA hollow nanoshells after dissolving the
silica core. We observed up to nine particle vibrational frequencies with increasing
size of the core–shell spheres and revealed the strong impact of the empty core of
the hollow capsules on the size dependence of the resonance frequencies. The
observed vibration eigenfrequencies are identified by detailed and thorough
numerical calculations as the resonance eigenmodes of the individual spheres. The
good overall agreement with the experiment allowed to determine the core density
and the two elastic constants of both constituents in the hybrid particles. We found
a significant deviation of these material properties in the nanostructured hybrid
spheres from their values in the macroscopic bulk systems, which underlines the
importance of such measurements at the nanoscale. The simulation of the
displacement fields of the different elastic modes allowed to visualize their
localization in different regions of the hybrid particles and provides a deeper
insight into their origin. These first findings illustrate qualitatively general
features of the localization of the elastic energy in nanostructured colloids beyond
core–shell particles.
The SiO2–PMMA particles with rigid silica cores and softer PMMA shells were
prepared by my coworker Dr. Spahn (DKI, Darmstad) in a two-step process
starting with the Stöber synthesis of the core followed by emulsion polymerization
of the shell [56]. The silica core, with a diameter of 181 ± 3 nm, was coated with
PMMA spherical shells of three different thicknesses (in average 25, 57, and
112 nm), leading to core–shell particles of final (outer) diameter ranging from
d = 232–405 nm. All samples were characterized by scanning electron micros-
copy (SEM), using a 1530 Gemini SEM by LEO with acceleration voltages setup
between 0.2 and 1.0 kV. Figure 4.17a shows exemplary details for the uncoated
silica cores and the three core–shell particles. Although there is some degeneration
of the PMMA by the electron beam at 1 kV or some blurring using lower voltage
(200 V), the particles’ diameters d can be determined by averaging the software-
aided gauged diameters over about 100 spheres for each sample. The size poly-
dispersity is about 5%, which is confirmed by the formation of crystalline films
(Fig. 4.17a) after vertical lifting deposition from the particle suspensions (cf. Sect.
3.1 in Chap. 3) [57]. Figure 4.17b shows SEM images of the spherical hollow
capsules with d = 232 and 294 nm. After dissolution of the silica core with
aqueous hydrofluoric acid, the diameters are found to be unchanged within the
experimental error. For the smaller hollow capsules (bottom left) a considerable
fraction (approximately 50%) of the particles possess holes, while the 294 (top
right) and the 405 nm (not shown) hollow shells are nearly defect free due to their
thicker shells. Notably these defects in the 232 nm hollow capsules lead to a
broadening of the spectral lines, but do not significantly change their peak posi-
tions in the BLS spectra (see Fig. 4.20a below). The sizes obtained from the SEM
were confirmed by the hydrodynamic radii, Rh, of the core–shell particles in dilute
suspension measured by photon correlation spectroscopy (Sect. 3.3.1 in Chap. 3).
Within 3%, Rh amounts to 214 nm for bare silica and 262, 352 and 504 nm,
respectively, for the three core–shell particles. These values are expectedly higher
than the geometric radii R measured by SEM since for homogeneous spheres the
ratio R/Rh assumes the value of 0.78 [58]. Experimentally, R/Rh is found to be
slightly higher, varying between 0.80 and 0.85. For the BLS experiment, films of
all seven particles were prepared on a thin glass substrate using the vertical lifting
technique [4].
Fig. 4.17 SEM-images of a the bare silica (left, d = 181 nm) and silica–PMMA core–shell
particles (from right: d = 405, 294 and 232 nm), and b the hollow PMMA capsules (bottom left:
d = 232 nm, top right: d = 294 nm). For the smaller particles in b, holes in the spheres are
observed, whereas there are nearly no defects in the thicker shells [34]
In analogy to the experiments in the previous sections, Fig. 4.18a shows the
q-independent BLS spectra of the bare silica particles and the three core–shell
particles. The spectra can be well represented by up to nine Lorentzian line shapes
as shown by the solid lines on the Stokes side of the BLS spectra. The peak
position of the spectral lines yields the resonance frequencies of the eigenmodes
indicated by solid circles in Fig. 4.18b–e for the four particles. Advantageously,
BLS can, in principle, record all thermally excited modes within one measurement,
which is not possible, e.g., in the pump-probe technique. For the bare silica
spheres, only two resonance frequencies at about 13 and 19 GHz can be resolved.
However, with increasing the PMMA-shell thickness of the core–shell particles
(d = 232, 294 and 405 nm), the BLS spectra become richer as it was observed for
pure polystyrene colloidal particles [8]. For the particle with the thinnest shell a
third weak peak is discernible in the spectrum in the inset of Fig. 4.18c, whereas
five modes are observed in medium-thickness shell (Fig. 4.18d), and even nine
modes are resolved in the BLS spectrum of the thickest shell (Fig. 4.18e).
The increased number of the resolved modes in the BLS with increasing particle
size relates to the intensity of the resonance signals which depends on qd. In the
present case of strong multiple scattering q B 2ki (the backscattering vector),
(a) (b) (c)
(d) (e)
(f)
Fig. 4.18 a BLS eigenmode spectra of bare silica (top) and the three core–shell particle films. The
representation of the BLS spectra with up to nine Lorentzians (solid lines) is shown on the Stokes-
side. (Note the different frequency scale for the spectrum in the top.) b–e Enlarged Stokes-sides of
the spectra from a (b pure silica, c–e core–shell). The experimental frequencies are accented by
orange spheres, the small vertical lines denote the corresponding calculated resonance frequencies,
each of them characterized by its angular momentum, l, shown at the top of the lines. All experi-
mental and theoretical (solid bars) values are summarized in f, with the dotted lines connecting
modes of the same angular momentum l; n = 1 for all observed modes [34]
and the number of resolved modes increases with the 2kid as discussed in
Sect. 4.2.1 [25].
The elastic parameters (longitudinal and transverse velocities, cl and ct) of the
two constituents (core- and shell-materials) are not a priori known for such
nanostructured systems. An access to these material properties at these length scales
and high frequencies is important since they can considerably differ from their
values in macroscopic systems. The elastic constants are frequency dependent, and
BLS specifically yields their limiting high frequency values, which relate to local
packing and interactions, as well as, the glass transition temperature [51]. The
detection of more than two particle elastic excitations in the experimental (BLS)
spectra of Fig. 4.18 allows for an unambiguous determination of the elastic moduli,
shear modulus G ¼ .c2t and Young modulus E ¼ .c2l ð1 þ rÞð1 2rÞ=ð1 rÞ, with
r ¼ ðc2l 2c2t Þ=½2ðc2l c2t Þ being the Poisson ratio and . being the mass density.
The experimental values of the resonance frequencies are compared with the
resonance frequencies obtained from the calculated density of states (DOS) spectra
of a single constituent sphere of the experimental systems. The theoretical
computations were performed by Dr. Sainidou (University of le Havre) using a
formalism, appropriately developed for this case and presented elsewhere [28, 59].
Each resonance mode appearing at frequency f(n, l) in these DOS spectra is
characterized by the angular momentum l, imposed by the spherical symmetry of the
particle, where n denotes the n-th order solution for a given l. All the shell-localized
modes reported in this section have n = 1. The materials elastic parameters (cl, ct)
and densities are used as adjustable parameters in order to achieve the least deviation
between theoretical and experimental eigenfrequencies. Obviously, in the theoret-
ical calculations the constituent spheres are considered as homogeneous and
isotropic, and their elastic coefficients are frequency-independent.
First, the bare silica particles are considered (Fig. 4.18b). The two sound
velocities treated as adjustable parameters are obtained from representation of the
two experimental frequencies (solid circles) by the calculated resonance frequencies
(small vertical lines in Fig. 4.18b). The obtained sound velocities for the bare-silica
particles, cl = 4420 ms-1 and ct = 2780 ms-1, are significantly lower (&25%)
than the values of dense bulk amorphous silica (cl = 5970 ms-1, ct = 3760 ms-1,
. ¼ 2200 kg m3 ), indicating the presence of porosity in these particles [31]. It also
underlines the necessity of the BLS experiment to determine these values avoiding
erroneous assumptions. The mass density does not sensitively affect the DOS
spectra, due to the huge impedance difference between silica and air. Nevertheless,
these porous silica spheres should be less dense than bulk silica (see below).
Next, the PMMA coated silica particles consisting of same silica cores are
considered. For a first description of the DOS spectra, representing the experi-
mental frequencies seen in the BLS spectra of Fig. 4.18c–e (solid circles), the
elastic constants measured for the bare silica particles were used and the values for
the sound velocity and density of the PMMA shell were fixed on the values for
bulk PMMA (cl = 2800 ms-1, ct = 1400 ms-1, . ¼ 1190 kg m3 ) [17]. However,
this choice for the set of the elastic parameters of both the silica core and the
PMMA does not quantitatively represent the experimental resonances in the BLS
spectra of Fig. 4.18c–e.
A systematic theoretical analysis based on the DOS spectra of the three core–
shell spheres has shown that changes in the elastic parameters of both materials of
the hybrid particles are required. Notably, it turned out that for the silica core one
must assume sound velocities, which are about 3% higher than in the bare silica
particle, i.e., cl,c = 4540 ms-1 and ct,c = 2860 ms-1. This hardening of the silica
core is probably due to a partial infiltration of methyl methacrylate in the pores and
subsequent polymerization to PMMA during the formation of the shell. Due to the
reduced impedance contrast between the silica core and the PMMA shell relative
to the bare silica spheres vs. air, the core mass density .c has now a substantial
influence on the DOS calculations. For a given density .c ¼ 1900 kg m3 , the
sound velocities in the PMMA shell are cl,s = 3080 ms-1 and ct,s = 1540 ms-1,
i.e., about 10% higher than in bulk PMMA; the PMMA mass density was kept at
the bulk value. Therewith, nearly all measured signals can be identified as
spherical eigenmodes with angular momentum l and can be captured quantitatively
by the theory within about 3%. The calculated resonance frequencies are shown by
small vertical lines in Fig. 4.18c–e and summarized in Fig. 4.18f along with the
corresponding experimental values. The Young modulus E and the shear modulus
G are directly accessible for both the core and the shell components.
In order to obtain an insight into the nature of the experimentally observed
modes (Fig. 4.18f), the elastic field at the resonance frequencies in the region of
the sphere was calculated, assuming a longitudinal acoustic plane wave of the
same frequency, impinging on the sphere (cf. Sect. 3.4.2 in Chap. 3). An example
of elastic-field intensity plots is given in Fig. 4.19 for the case of the thickest
PMMA shell sample (d = 405 nm). Their topology is that of a field-intensity
having 2l maxima on the internal interface of the outer circumference of the core–
shell particle: the two of them are strong (global) maxima along the direction of
the incident field, while the rest 2(l - 1) are weaker and distributed equidistantly
along the circumference. The vibration of the observed modes is radial at the
maxima changing alternatively direction (outwards or inwards) from maximum to
maximum. In the regions of minima the field is directed tangentially. Based on
Fig. 4.19, the experimentally observed modes (Fig. 2f) are shell-localized with
l = 2-4. Core-localized modes either do not exist within the considered fre-
quency range (films with d = 232 and 294 nm), or they are not observable in the
experimental BLS spectra (thickest-shell case, Fig. 4.18e), probably due to the
very strong localization (virtually delta-functions in the DOS spectra). It is worth
mentioning, however, that the elastic parameters affect the frequencies of both
core- and shell-like eigenmodes.
Figure 4.20a shows the BLS spectra of two hollow capsules (d = 232 and
294 nm), akin to those of double-shelled hollow carbon microspheres published
recently [60]. By comparison of the two spectra, it is apparent that the thicker
hollow capsules display significantly sharper peaks than the thinner hollow
capsules. For solid spheres, the line width of the spectra is associated with the size
Low
(a) (b) (c)
High
Low
(d) (e) (f)
High
Fig. 4.19 a–c Elastic-field distribution for the first three resonances (n = 1) of the 405 nm core–
shell particle (Fig. 4.18e) in a cross section through the center of the sphere. Plane wave inci-
dence is assumed along the horizontal axis, from the left. d An example of core-localized mode
with l = 1 (n = 2) at 9.3 GHz, which is not observed experimentally. Plots e and f give,
respectively, the real part of the radial and polar component of the elastic field for the case shown
in c, with the arrows visualizing the vibrational mode of the shell [34]
polydispersity [8]. For the thin hollow spheres (d = 232 nm), however, an
additional source for the line broadening is the significant fraction of particles with
holes as seen in Fig. 4.17b. These defects are neither uniform in size nor in shape
and hence cause a distribution of the elastic properties that further broadens the
experimental spectra (Fig. 4.20a). In order to identify the nature of the experi-
mental eigenmodes in the two hollow capsules, the resonance frequencies of shell-
localized modes were computed, adopting for the PMMA shell the values of the
elastic moduli obtained from the representation of the eigenfrequencies of the
core–shell particles (Fig. 4.18f). Figure 4.20b shows the calculated (solid lines)
eigenfrequencies for the modes characterized by the angular momentum l = 2–4
for hollow PMMA nanoshells with constant inner diameter (181 nm) as a function
of the (outer) particle diameter d. The comparison with the experimental fre-
quencies (orange spheres) identifies the observed frequencies in the three hollow
PMMA spheres with the two lowest modes with l = 2 and l = 3. Interestingly, the
frequencies of these two modes vary very little with d due to two competing
effects. At a constant particle diameter, the resonance frequencies of hollow
particles increase with shell thickness [61], whereas the resonance frequencies of
filled spheres decrease with d since f ðn; lÞ 1=d [7–9]. Since in our case both shell
thickness and total diameter increase simultaneously, the net effect is essentially
the apparent insensitivity of the eigenfrequencies to the d variation of the branches
with l = 2 and 3 seen in Fig. 4.20b. Notably, the elastic parameters of the PMMA
were not affected by the core etching.
Fig. 4.20 a BLS eigenmode (a)

spectra of spherical hollow
PMMA capsules with
d = 232 and 294 nm. b
Comparison between calcu-
lated (solid lines) and
experimental (spheres)
resonance frequencies for the
hollow capsules (n = 1) [34]
(b)
In summary, this section reports the first study of localized vibrational exci-
tations in silica-PMMA core–shell particles and PMMA hollow capsules using the
powerful optical technique of BLS, which is applicable for turbid films. The BLS
spectra show up to nine eigenfrequencies of the core–shell particles, which sen-
sitively depend on the particle architecture and the mechanical moduli of the
constituent parts. The values of the Young moduli E and shear moduli G of the
constituent components are computed from the elastic parameters cl and ct and
densities, obtained for each component from the identification of the experimental
modes with the resonance modes appearing at f(n, l) for the n-th order of the l-th
harmonic in the calculated density of states (DOS) spectra. The anticipated
reduction of the moduli of the neat core compared to the bulk material due to its
porosity and subsequent subtle increase above the bulk values upon grafting with
PMMA chains are revealed. The anchoring and confinement of the PMMA
nanoscopic layer impact both types of moduli which were found to exceed the bulk
PMMA values. The observed eigenfrequencies in the hollow capsules exhibit a
peculiar but apparent insensitivity to the variation of the diameter as a result of
antagonistic trends inferred by the DOS calculations, while the PMMA elastic
constants are not affected by removal of the silica core. In addition to the fun-
damental understanding of localized elastic modes in hybrid particles and the
concurrent determination of the mechanical moduli, the findings of this study will
contribute to a rational design of nanostructured colloids with strong resonances
that can be selectively excited [10].
4.4 Materials
4.4.1 Homogeneous Polymer Colloids
Homogeneous polystyrene, poly(methyl methacrylate), and poly(methyl methac-

rylate–co–n-butyl acrylate) colloids have been prepared by my coworker Dr.
Retsch at MPIP by emulsifier-free emulsion polymerization [62], adapting recipes
from the literature [63–66]. A general recipe for the used particles is given by the
following [62]:
The polymerization reactions were carried out either in a 300 ml reactor with a thermostat
jacket or a 500 ml three necked flask. At first water was heated to the reaction temperature.
Then all additives, monomers, and comonomers were added; solid chemicals were dis-
solved in water prior to addition. Strong magnet stirring was used and the chemicals were
allowed to equilibrate for 5 min. Finally, the initiator was added as aqueous solution. The
reactor was flushed with N2 or Ar during charging with the chemicals and a slight inert gas
flow was maintained throughout the reaction. Reaction times were typically between 12
and 20 h. […] Latex and silica nanoparticles were either purified by several cycles of
centrifugation and redispersion, by ultrafiltration, or dialysis. In all cases the purification
protocol was adjusted to the particles size and composition by means of centrifugation
speed and duration, or filtration and dialysis membrane molecular weight cut-off.
PS Styrene ([99%, Aldrich) was washed three times with a 10% KOH solution
and three times with MilliQ water and then distilled under reduced pressure.
Acrylic acid (AA, 99%, Aldrich) was distilled without washing with KOH.
Sodium 4-vinylbenzenesulfonic acid (NaPSS, 90%, Aldrich) and potassium per-
sulfate (KPS, 99%, Acros) were used as received. By playing around with
monomer, comonomer (KPS, AA), and initiator concentration, M. Retsch could
prepare monodisperse PS spheres with diameter between 180 nm and 1 lm. All
reactions were carried out with 0.1–0.2 g KPS, 0.005–0.06 g NaPSS, and 0–0.3 ml
AA in 250 ml H2O at 70 or 80 C.
PMMA Methyl methacrylate (MMA, 99%, Acros) was washed three times with a
10% KOH solution and three times with MilliQ water and then distilled under
reduced pressure. Sodium 4-vinylbenzenesulfonic acid (NaPSS, 90%, Aldrich),
potassium persulfate (KPS, 99%, Acros), and 2,20 -azobis(2-methyl propionami-
dine) dihydrochloride (ABA, 97%, Aldrich) were used as received.
Cationic initiator (ABA) as well as an anionic system (KPS, NaPSS) have
been used. The particle diameter is found to depend mainly on the initial MMA
concentration.
4.4 Materials 85
PMMA-nBA Copolymers Methyl methacrylate (MMA, 99%, Acros) was washed

three times with a 10% KOH solution and three times with MilliQ water and then
distilled under reduced pressure. n-Butyl acrylate (nBA, [99%, Aldrich) was run
over a basic alumina column. Sodium 4-vinylbenzenesulfonic acid (NaPSS, 90%,
Aldrich) and potassium persulfate (KPS, 99%, Acros) were used as received. All
reactions were carried out with 0.2 g KPS and 0.02 g NaPSS in 250 ml H2O at
780 C.
4.4.2 Homogeneous Silica Colloids
There are principally two synthetic strategies to obtain monodisperse silica

colloids: a one-step batch process, which is also termed ‘Stöber’ process [67, 68],
and a seeded growth process that can be performed step-by-step or continuously
[69–71]. While the Stöber process is the simplest and fastest method, the mono-
dispersity obtained by that is less good then that achieved using the seeded growth
processes.
All processes avail themselves of the hydrolysis of tetraethyl orthosilicate
(TEOS) to orthosilicic acid and subsequent condensation of the silicic acid to
poly(silicon dioxide) (silica) under alcalic catalysis [ammoniac, NH3(aq)] and
strong magnetic stirring. As solvent a water/ethanol mixture is utilized due to the
poor solubility of TEOS in water. The temperature is chosen between room
temperature and 60 C.
In order to control the particle size over a wide range and achieve high
monodispersity by the same time, the typical approach is to start with a sample of
monodisperse particles from the Stöber process (smaller diameter than the wished
size) and then TEOS in EtOH and NH3(aq) either in steps or in a continuous flow
over several hours to days. It is found that the diameter increases monotonically
with the amount of added monomer following a cubic root dependency and that
the polydispersity decreases with increasing amount of added monomer, i.e., with
the size. Both, step-by-step and continuous process, deliver high quality particles
with comparable monodispersity [62]. The silica particles used in this thesis have
been synthesized by my coworkers Retsch at MPIP and Kiefer at DKI, Darmstadt.
4.4.3 PS–SiO2 Core–Shell Particles
PS–SiO2 core–shell particles have been provided by my coworkers D’Acunzi and

Schäfer at MPIP. Therefore they started with the soap-free emulsion polymeri-
zation of the polymer core and used them as templates for the synthesis of the
silica shell employing the Stöber method [49, 72]. As the details are clearly and
completely given in the mentioned literature, here I just want to introduce the main
principles of the synthesis.
The synthesis of the PS core by soap-free emulsion polymerization was carried

out using ammonium persulfate as initiator. Divinylbenzene (DVB) and acrylic
acid were facultatively added as crosslinking (DVB) comonomers. Monomers and
comonomers were dropwise added to a stirred aqueous solution of the initiator
ammonium persulfate. The cores were cleaned by centrifugation.
To ease the coating with silica, the PS spheres were treated with polyelectro-
lytes [poly(allylamine hydrochloride)] and then functionalized with poly(vinyl-
pyrrolidone) (PVP). The functionalized PS beads were transferred into ethanol. To
form the silica shell, ammonia and tetraetoxysilane (TES) were added under
stirring. The completed core–shell particles were washed three times in ethanol.
4.4.4 SiO2–PMMA Core–Shell Particles
The SiO2–PMMA particles with rigid silica cores and softer PMMA shells were
prepared by my coworker Spahn at DKI Darmstadt in a two-step process starting
with the Stöber synthesis of the core followed by emulsion polymerization of the
shell [56]. The silica core, with a diameter of 181 ± 3 nm, was coated with
PMMA spherical shells of three different thicknesses (in average 25, 57, and
112 nm), leading to core–shell particles of final (outer) diameter ranging from
d = 232 to 405 nm.
References
1. Liff SM, Kumar N, McKinley GH (2007) Nat Mater 6:76

flow of light. 2nd edition, Princeton University Press
3. Ozbay E (2006) Science 311:189
5. Duval E, Boukenter A, Champagnon B (1986) Phys Rev Lett 56:2052
6. Courty A et al (2005) Nat Mater 4:395
9. Penciu RS et al (2000) Phys Rev Lett 85:4622
11. Akimov AV et al (2008) Phys Rev Lett 101:033902
12. Hodak JH, Henglein A, Hartland GVJ (1999) Chem Phys 111:8613
14. Xia YN, Gates B, Yin YD, Lu Y (2000) Adv Mater 12:693
15. Hynninen A-P et al (2007) Nat Mater 6:202
16. Maldovan M, Thomas EL (2006) Appl Phys Lett 88:251907
20. Gorishnyy T et al (2005) Phys Rev Lett 94
References 87
21. Lamb H (1882) Proc Lond Math Soc s1-13:189

22. Duval E, Boukenter A, Champagnon B (1986) Phys Rev Lett 56:2052
23. Duval E (1992) Phys Rev B 46:5795
24. Still T et al (2008) J Phys: Condens Matter 20:404203
26. Li Y et al (2008) Chem Phys Lett 461:111
29. Penciu RS et al (2003) J Chem Phys 118:5224
30. Faatz M et al (2005) Langmuir 21:6666
31. Lim HS, Kuok MH, Ng SC, Wang ZK (2004) Appl Phys Lett 84:4182
32. Sainidou R (personal communication)
33. Li Y et al (2006) Appl Phys Lett 88:023112
35. Still T, D’Acunzi M, Vollmer D, Fytas G (2009) J Coll Int Sci 340:42
36. Alfrey T, Price CCJ (1947) Polym Sci 2:101
37. Greenley RZ (1999) Polymer handbook, vol. 1, 4th edition. Wiley, Hoboken
38. Penzel E, Rieger J, Schneider H (1997) A Polym 38:325
39. Strobl G (1997) The physics of polymers, 2nd edition. Springer, Berlin
40. Linderstrom-Lang K (1937) Nature 139:713
42. Still T et al (2010) Macromolecules 43:3422
43. Wood AB (1941) Textbook of sound. Macmillan, New York
46. Gill I, Ballesteros A (1998) J Am Chem Soc 120:8587
47. Van Helden AK, Jansen JW, Vrij A (1981) J Coll Int Sci 81:354
48. Donath E et al (1998) Angew Chem Int Ed 37:2201
49. Zhang L et al (2009) Langmuir 25:2711
50. Cheng W et al (2008) Nano Lett 8:1423
52. Forrest JA, Dalnoki-Veress K, Stevens JR, Dutcher JR (1996) Phys Rev Lett 77:2002
53. Keddie JL, Jones RAL, Cory RA (1994) Faraday Discuss 98:219
54. Fox TG, Loshaek SJ (1955) Polym Sci 15:371
55. Mark JE (1999). Polymer data handbook. Oxford University Press, Oxford
56. Spahn P (2008) Phd thesis
57. Vlasov YA, Bo X-Z, Sturm JC, Norris DJ (2001) Nature 414:289
58. Pecora R (1985) Dynamic light scattering: applications of photon correlation spectroscopy.
Plenum Press, New York
59. Sainidou R, Stefanou N, Modinos A (2004) Phys Rev B 69:064301
60. Li Y, Lim HS, Ng SC, Kuok MH (2007) Chem Phys Lett 440:321
61. Sainidou R, Stefanou N, Psarobas IE, Modinos A (2005) Z Kristallogr 220:848
62. Retsch M (2009) Complex materials via colloid crystallization. PhD thesis
63. Goodwin JW, Hearn J, Ho CC, Ottewill RH (1974) Colloid Polym Sci 252:464
64. Chung-Li Y, Goodwin JW, Ottewill RH (1976) Prog Coll Polym Sci 60:163
65. Shouldice GTD, Vandezande GA, Rudin A (1994) Eur Polym J 30:179
66. Egen M, Zentel R (2004) Macromolecular chemistry and physics 205:1479
67. Stöber W, Fink A, Bohn E (1968) J Coll Int Sci 26:62
68. Giesche H (1994) J Eur Ceram Soc 14:189
69. Bogush GH, Tracy MA, Zukoski Iv CF (1988) J Non-Cryst Solids 104:95
70. Zhang JH et al (2003) J Mater Res 18:649
71. Giesche H (1994) J Eur Ceram Soc 14:205
72. Graf C, Vossen DLJ, Imhof A, van Blaaderen A (2003) Langmuir 19:6693
Chapter 5
Phononic Behavior of Colloidal Systems
5.1 Introduction
The first papers on photonic effects by Yablonovitch [1] and John [2] in 1987
stimulated over the years much theoretical and experimental work on the
propagation of electromagnetic waves through appropriately structured materials
and subsequently led to the birth of the new research field ‘photonic crystals’ [3].
The tremendous interest in photonic crystals with specially designed periodic
variations in dielectric constant largely originates from their display of propagation
band gaps for light. The appearance of band gaps makes an advanced control over
light propagation possible and permits as well a series of novel optical phenomena
such as slowing and localization of light or negative refraction [3]. Soon after the
discovery of photonic crystals, it was found that in analogy to the electromagnetic
waves, band gaps also exist for the propagation of acoustic waves, and the so-called
phononic crystals [4–10] are the elastic analogue of photonic crystals replacing the
role of the dielectric constant by the elastic parameters and density. Such an analogy
exists as a consequence of the common origin of the band gaps in both cases, i.e., the
destructive interference of Bragg diffracted waves in periodic structures [11, 12],
and hence these gaps are also termed as Bragg gaps. However, the different nature of
electromagnetic and acoustic waves also guarantees the existence of some important
differences between photonic and phononic phenomena. Unlike electromagnetic
radiation that is characterized as transverse waves, acoustic waves in general are full
vector waves with both longitudinal and transverse polarizations, and their
propagation depends additionally on the material density. Even for a homogeneous
and isotropic medium, the acoustic wave propagation is governed by three
parameters, the two Lamé coefficients and density, in contrast to the single
parameter, the dielectric constant, that determines the propagation of light.
Evidently, the phononic phenomena are anticipated to be more complex and rich.
For phononic crystals, the band diagram depends on several parameters such as
the elastic constants and density of the component materials, symmetry of the
lattice, shape of the inclusions, and the filling fraction. The width of the band gap

90 5 Phononic Behavior of Colloidal Systems
generally increases with the contrast between the densities and sound phase
velocities of the component materials, and the center of the gap can be tuned by
changing the lattice parameter [9, 10]. The search for phononic structures started
with two theoretical works in 1993, which predicted the existence of phononic
band gaps in periodic two-dimensional (2D) elastic composites of parallel cylin-
ders embedded in a host matrix [9, 10]. The experimental verification of phononic
band gaps followed few years later, realized in metallic macrostructures with gaps
at sonic or ultrasonic frequencies [4, 5, 8, 13]. Further explorations also revealed a
number of peculiar phenomena with potential applications associated with acoustic
wave propagation including tunneling effect [14], negative refraction and focusing
[15], double refraction, etc. [16].
Theoretical calculation of phononic band diagram requires no specification of
the lattice constant of the structure or the corresponding wave frequencies as long
as the crystal is defined by the same set of frequency-independent elastic
parameters. In other words, a fundamental length scale does not exist for phononic
phenomena, which is a direct consequence of the invariance of the wave equation
of elasticity under the simultaneous transformation of space coordinates and
frequency. However, acoustic waves of different frequencies do bear distinct
characteristics, particularly when their applications are concerned, and therefore in
practice the frequency range of the waves of interest constitutes an important
consideration. Recently, growing attention has been paid to hypersonic (GHz)
phononic crystals and the first experimental observations of the hypersonic Bragg
gaps have been lately reported in three-dimensional (3D) colloidal crystals made
by self-assembly [17] and in 2D polymeric porous structures with hexagonal
symmetry created by laser interference lithography [18]. Hypersonic waves, owing
to their high frequencies, display certain unique features that are not possessed by
ordinary acoustic waves such as being thermally excited, acting as the main heat
carrier in dielectrics, and interacting with electrons and photons in a rich manner.
Consequently, hypersonic crystals with the potential to mold the flow of
hypersound may be utilized to achieve high-level control over many important
physical processes involving heat transport and complex phonon–photon or
phonon–electron couplings. A detailed understanding of phonon propagation in
hypersonic crystals thus becomes important.
The fabrication of hypersonic crystals, compared with their sonic and ultrasonic
counterparts, is much more demanding as the dimension typifying the structure has
to be scaled from macroscopic down to sub-micron scale. On the other hand,
advance in relevant nanofabrication, partially driven by the desire of creating
various photonic structures of similar dimensions, has offered some available
means to achieve such a purpose including, for example, holographic interference
lithography [19], direct laser-writing [20], two-photon polymerization [21], or
self-assembly [22, 23]. Colloidal superstructures, self-assembled from colloidal
particles, represent a promising material class for phononic applications. The
maturation of colloidal science enables the preparation of colloidal particles with
well-defined size and shape for a great many of materials ranging from organic to
inorganic, thus providing abundant building blocks with varied elastic properties.
5.1 Introduction 91
The progress in colloidal particle self-assembly allows easy and cheap fabrication
of large area high quality single crystalline 2D and 3D crystals as well as
non-crystalline structures. The appearance of binary [24] and ternary [25] crystals
further enriches the available structure types. By filling the interstitials between the
colloidal particles with different materials, additional freedom in elastic parame-
ters of the system is provided. Moreover, in most cases the colloidal particles
possess a spherical shape due to surface tension effects, thus represent a strong
scattering unit when the elastic contrast between the particle and surrounding
becomes large, which may cause additional gap formation with origin different
from Bragg diffraction as will be encountered later [26].
The experimental exploration of hypersonic crystals, however, faces additional
challenge in monitoring the phonons in such small structures. Apparently, the
commonly used sonic and ultrasonic transmission techniques for macroscopic
sized structures cease to work. It has been demonstrated that Brillouin light
scattering (BLS), which takes advantage of the inelastic scattering of photons by
thermally excited high frequency phonons, represents a powerful tool to record the
phonon dispersion relation in hypersonic crystals.
This chapter deals with the phononic band diagrams of transparent colloidal
systems that can be measured by BLS. A short introduction into the nature of the
bands in the dispersion relation is given as well as into the properties of the
effective medium. Experimental results are discussed treating the effective
medium velocity of infiltrated defect doped colloidal crystals and the nature of
acoustic band gaps in phononic systems, including the first realization of a
hybridization gap.
5.1.1 The Phononic Band Diagram
The information about the propagation of elastic waves in periodic media is

described by the band diagram. It contains the dispersion relation frequency as a
function of the wave vector, xðkÞ: Therefore the vertices of the irreducible
Brillouin zone (BZ) are connected with a line for each acoustic mode appearing.
Figure 5.1a shows for example the band diagram for an fcc lattice (see also
Fig. 5.5) of lead spheres in a beryllium matrix (filling fraction 8.23%), calculated
by Economou and Sigalas using the plane wave method (cf. Sect. 3.4.2 in Chap. 3)
[27]. The lowest two branches (indeed, the lowest three as the two very lowest
ones are degenerated) are the so-called acoustic ones. At the central point of the
BZ, C, their frequency is zero, and around C the frequencies increase linearly
(dispersionless) in all crystallographic directions, i.e., in these regions phase and
group velocities of the corresponding acoustic waves are equal.
Away from this long wavelength (small k) limit, when the branches approach
the edge of the (first) BZ, the linear behavior vanishes and a bending occurs. That
bending can be best explained for the simplest case of a periodic elastic system—
an one-dimensional lattice of vibrating identical spheres with mass m connected
(a) (b)
Fig. 5.1 a From (Ref. [27]) calculated acoustic band diagram for lead spheres in a beryllium
matrix forming an fcc lattice with an occupancy of 8.32 vol% the lowest (transverse) branch is
degenerated. The three flat branches around 2.8 xa/c are degenerated, too. There are two
complete band gaps. b Dispersion of the longitudinal waves in a 1D simple crystal (black line)
over several Brillouin zones. The dashed line denotes the dispersionless ‘free particle’ case
with massless elastic springs with spring constant D and equilibrium distance a as
shown in Fig. 5.2 [28, 29]. Let us assume a longitudinal wave, which creates
displacement of the spheres in the direction of the chain. When assuming (in
reasonable approximation) that the elastic forces that determine a vibration have
their origin in the relative displacement only of neighboring particles, symbolized
by the compressed and stretched springs in Fig. 5.2 leading to forces (black
arrows) in the same direction, while the inertia originates from the absolute
displacement of the spheres in space. With u(na) being the absolute displacement
of the nth sphere, the harmonic potential between all neighboring spheres becomes
1 X
U harm ¼ D ðuðnaÞ uð½n þ 1aÞÞ2 : ð5:1Þ
2 n
Since elastic forces and inertia must compensate each other, the equation of
motion becomes
o2 uðnaÞ oU harm
m 2
¼ ¼ D½2uðnaÞ uð½n 1aÞ uð½n þ 1aÞ: ð5:2Þ
ot ouðnaÞ
Assuming a plane longitudinal wave of the form
uðna; tÞ / eiðknaxtÞ ð5:3Þ
Fig. 5.2 Illustration of a

longitudinal mechanical wave
in an 1D elastic lattice
5.1 Introduction 93
and inserting into Eq. 5.2 delivers

mx2 eiðknaxtÞ ¼ D½2 eika eika eiðknaxtÞ : ð5:4Þ
After algebraic transformation the dispersion relation is obtained:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffi
2Dð1 cosðkaÞÞ D ka
xðkÞ ¼ ¼2 sin : ð5:5Þ
m m 2
The dispersion relation for this simple case is shown in Fig. 5.1b for several
BZs. The periodicity of the lattice is recovered also in the dispersion relation in
reciprocal space. In the long wavelength limit, i.e., k pd ; Eq. 5.5 can be
simplified to
rffiffiffiffi
D
xðkÞ kd; ð5:6Þ
m
using the approximation cosðkdÞ 1 ðkdÞ2 =2: This corresponds to the
mentioned linear behavior at small ks. In Fig. 5.1b, dashed lines are drawn for
comparison that show purely linear slope. This behavior corresponds to the ‘free
particle’ case, i.e., a phonon in a continuum. Of course, in this case the axis caption
including the periodicity d is meaningless and k is simply given by k = 2p/K with
phonon wavelength K. The slopes of the linear regions of the acoustic branches
give the effective longitudinal and transverse velocities cl and ct.
At the edge of the first BZ (k = p/d), Eq. 5.5 becomes
rffiffiffiffiffiffi
4D
x¼ ; ð5:7Þ
m
pffiffiffiffiffiffiffiffiffiffi
with corresponding longitudinal phase velocity cl ¼ D=m 2d=m and group
velocity zero. Waves that fulfill that condition are reflected into themselves
(for normal incidence) and form a standing wave.
Above the acoustic branches, in Fig. 5.1a some flat bands appear around 2.8
reduced wavelength (indeed, there are three of them). Since these branches are
apparently almost k-independent and because of their degeneracy, they are
identified as eigenwaves basically trapped within the lead spheres with little
leaking out [27]. These are resonance modes similar to the eigenvibrations
discussed in Chap. 4, however, this time no free boundary conditions exist, but
there is an exchange of energy between the spheres and the matrix. Only if the
elastic mismatch (i.e., densities and moduli) is large enough between spheres and
matrix, such a trapping can happen. With decreasing mismatch the amount of
dissipation increases and finally, for no elastic mismatch, the systems behaves like
a continuum, i.e., only acoustic modes appear.
At higher frequencies also mixed modes with transverse as well as longitudinal
components appear. These modes may become quite numerous even for relatively
simple systems. For systems with liquid matrix that do not support shear waves,
the calculation might become a bit easier, however, it is still more complex than
the analogue investigations for electromagnetic waves. In systems where one
component supports shear waves and the other component does not, one must not
ignore the conversion between waves of different nature into each other that is
described mathematically by the T-matrix (Eq. 3.28).
Regions in the band diagram where no band exists for all ks are called phononic
band gaps. These are frequency regions where no mechanic wave can propagate.
Since the existence of such gaps is one of the striking features of the band diagram
and promises also some technical importance, the search for and investigation or
tailoring of phononic band gaps is a demanding question in the young field of
phononics.
In the following parts of the introduction, the properties of the effective
medium as well as a short overview over the phononic band gaps is given. The rest
of this chapter is concerned with experimental studies on colloidal phononic
systems, including the investigation of the effective medium velocities in defect
doped opals, the first experimental realization of a hybridization gap and the band
diagram of nanostructured systems.
5.1.2 The Effective Medium
The slope at the long wavelength limit in the band diagram gives the longitudinal
and transverse sound velocities, cl and ct, respectively. While these velocities are
well known for the most pure materials, their prediction for composite materials of
varying composition and geometric structure is not straight forward and the topic
of the effective medium theory (EMT).
A very general approach to predict the effective longitudinal velocity of a
two components system is given by Wood’s law as the harmonic mean of the
moduli [30]:
1 / ð1 /1 Þ
¼ 1þ ; ð5:8Þ
Meff M1 M2
where M ¼ c2l . is the bulk longitudinal modulus of each component. In principle
this approach can be generalized for n components to
1 / ð/ Þ ð/ Þ
¼ 1 þ 2 þ þ n ð5:9Þ
Meff M1 M2 Mn
P
with n /n ¼ 1: For equal mass densities ð.1 ¼ .2 Þ; the effective longitudinal
velocity would be given as the harmonic mean of the squared sound velocities of
the components (cl ½h1=c2l i1=2 ). For transverse sound velocities, it should
be sufficient to replace the bulk longitudinal moduli in Eq. 5.8 by the share
moduli.
Anyhow, the simple assumption of Wood’s law ignores for example the
interaction of longitudinal and transverse waves between different components as
it (in the longitudinal version) compares only the bulk moduli. However, in certain
5.1 Introduction 95
cases, these interactions may lead to conversion of energy between bulk and shear
waves (T-matrix) and must not be ignored anymore. In the theoretical calculations
mentioned above [27], the authors compare the calculated effective velocities for
several different cases with the arithmetic mean and the harmonic mean of the
sound velocities as well as with the square root of the harmonic mean of the
squared velocities of the components—all weighted with the filling fraction, but
not with the mass densities. This deviation from Wood’s law does not lead to a
reproducible result for all cases, since in some cases the one mean is nearer to the
calculated result while in other cases another mean is nearer to the theoretical
findings. However, even applying Wood’s law, which is not done in the paper,
does not deliver perfect agreements for all cases, although a relatively large
deviation is achieved only in the case of lead inclusions in a beryllium matrix with
a filling fraction of 8.23%. Here, using the parameters given by the authors,
Wood’s law delivers an effective longitudinal sound velocity of 9,250 ms-1, while
the calculated velocities are about 10% higher, between 10,226 and 10,619 ms-1,
depending on the crystallographic direction.
For the special case of spherical inclusions (and the practical application for
approximately spherical inclusions), a more sophisticated but still easy-to-apply
EMT method was developed mainly by Gaunaurd and Wertman, reviewed by the
same in Ref. [31]. In their calculations they take the ratios of longitudinal and
transverse sound velocities for both components into account as well as the ratio of
the mass densities. Furthermore, the effective density is calculated differently for the
two general subcases of matrices that can only support insignificant amounts of shear
(‘fluid matrices’) and ‘elastic matrices’ supporting non-negligible amounts of shear.
When calculating the effective longitudinal sound velocities for the different
cases in Ref. [27] following this approach (utilizing a small C++ program written
by myself), reasonable results are obtained for all subcases (Au/Si, Pb/Si, Pb/Be,
and Au/SiO2), although for some subcases from the theoretical band diagram there
are different effective velocities in different crystallographic directions, which are
described unequally well by this EMT. E.g., for the Pb/Be subcase mentioned
above, the program gives an effective sound velocity of 10,820 ms-1, which is
only about 2% more than predicted for the C–L direction.
In Table 5.1 three general cases of phononic systems with filling fraction / are
compared for Wood’s law and the EMT of Gaunaurd and Wertman: polystyrene
spheres in silicon oil (solid/liquid), alumina spheres in PS (solid/solid), and air
bubbles in silicon oil (gas/liquid).1 For all subcases the values following the
Gaunaurd and Wertman approach are given for fluid (GWf) and elastic matrix
(GWe), although principally the latter should be meaningful only in the alumina/
PS subcase since silicon oil that is matrix in the other two subcases does not
support shear waves.
1
Material parameters: PS: ct: 1,200 ms-1, cl: 2,350 ms-1, .: 1,050 kg m-3; silicon oil: ct: 0 ms-1,
cl: 1,400 ms-1, .: 1,000 kg m-3; Al2O3: ct: 6,345 ms-1, cl: 10,850 ms-1, .: 3,970 kg m-3; air:
ct: 0 ms-1, cl: 343 ms-1, .: 1.2 kg m-3.
Table 5.1 Comparison between calculated effective longitudinal sound velocities (in ms-1)
using the EMT of Gaunaurd and Wertman (GW) and Wood, respectively, for three cases and five
filling fractions /
/ PS in silicon oil Al2O3 in PS Air in silicon oil
GWf GWe Wood GWf GWe Wood GWf GWe Wood

0.74 1,714 1,714 1,925 2,854 2,522 2,590 42 27 27
0.50 1,588 1,587 1,691 2,479 2,104 2,136 34 24 24
0.20 1,466 1,466 1,496 2,362 2,095 2,103 35 30 30
0.05 1,416 1,415 1,422 2,351 2,257 2,258 57 54 54
0.001 1,400 1,400 1,400 2,350 2,348 2,348 364 364 364
GW calculations for fluid and elastic matrix, GWf and GWe, respectively
For the first case of PS in silicon oil, it does not make any significant difference
which kind of matrix is taken into account. Both calculations come to practically
the same results, leading to sound velocities lower than those predicted by Wood’s
law. With increasing filling fraction the difference increases up to 12% for the
dense packing case / = 0.74.
In the subcase of a harder solid embedded in a softer one (alumina in PS), GWe
and Wood’s law deliver quite similar results, differing by less than 3% at
/ = 0.74, while the theoretically meaningless GWf delivers significantly higher
values.
The last subcase of air bubbles in silicon oil shows that in this case, where both
components’ transverse sound velocities are set to zero, GWe delivers exactly the
same results as Wood’s law, while the GWf calculations give higher values.
However, even for very low filling fractions all theories give extremely low values.
This means that even a low amount of air bubbles should slow down the effective
sound velocity in a liquid.
Another EMT method is described by Waterman and Truell [32]. Their
approach is in principle a full multiple scattering theory including numerical
calculation with n equations for n scatterers. For every scatterer the exact form of
the multipole coefficient must be known, and so this method is too complicated to
be used for simple calculations of the effective sound velocity. Taking the
effective sound velocity from the theoretical band diagrams calculated by the
multiple scattering method touched on in Sect. 3 in Chap. 3 is very akin to this
approach.
5.1.3 Phononic Band Gaps
To measure the phononic dispersion relation of colloidal crystals by BLS, the

strong multiple scattering that leads to the eigenmode spectra discussed in Chap. 4
should be overcome. This is achieved by infiltration of the dry opals by a fluid
with refractive index close to that of the colloidal spheres (Fig. 5.3). For
5.1 Introduction 97
Fig. 5.3 Infiltration of a turbid colloidal crystal with an index matching liquid resulting in a
transparent ‘wet’ crystal (according to [17])
poly(methyl methacrylate) (PMMA, n = 1.49) or PS (n = 1.59) colloids for

example silicon oil (n = 1.45) or liquid poly(dimethylsiloxane) (PDMS, n = 1.41)
are suitable infiltration liquids. Using such ‘wet’ colloidal crystals, the first
experimental access to the phononic band diagram along high symmetry directions
in submicrometer artificial crystals was recently reported [17], leading to the
observation of phononic band gaps in the GHz-range. Till then, only sonic and
ultrasonic crystals had been investigated by other techniques, mostly by sonic- and
ultrasonic transmission experiments or by a special video spectroscopy methods
applied to 2D [33] and 3D [34] experiments. The observation of a hypersonic band
gap in a 2D hypersonic phononic crystal by BLS was subsequently reported [36].
The nature of the observed band gaps is clearly that of a Bragg gap (BG). A BG
exists due to annihilation of the incident mechanical wave and the waves scattered
on the individual spheres in the crystal. To create a band gap, the periodic structure
length in the crystal must be of the order of the relevant phonon wavelength. If the
phonon wavelength and the distance between the crystal planes in the direction of
the phonon propagation become equal, the scattered mechanical waves interfere
constructively and the energy of the incident wave is reflected back opening a
phononic gap.
Fcc crystals of sub-micron PS spheres are prepared by vertical lifting
deposition. After infiltration with PDMS, the dispersion relation, f(q), was
recorded by angle dependent BLS measurements along high-symmetry directions
in the reciprocal space. As shown in Fig. 5.4 at low qs, the dispersion relation is
that of mechanical waves traveling through a homogeneous medium with an
effective sound velocity, i.e.,
x ¼ 2pf ¼ ceff q: ð5:10Þ
pffiffiffi
As q reaches the edge of the first BZ at qBZ ¼ ð3=2Þ3=2 p=ð 2dÞ in the probed
C–M direction (cf. Sect. 5.1.3), the fulfillment of the Bragg condition leads to a
gap opening. Accordingly, the frequency of the observed gap is [5, 35]
Fig. 5.4 The dispersion relations of a colloidal PS (d = 307 nm) opal infiltrated with silicon oil
(left see also Ref. [17]) and with PDMS (right, see also Fig. 5.11 and Ref. [26]) in the C–M
direction of the fcc crystals are compared. In both spectra there is a clear BG (diagonal pattern).
The dashed lines indicate the C–M distance in the reciprocal space, hence the edge of the first BZ.
The dotted lines denote the frequencies that correspond to qBZ (cf. Sect. 5.1.3). For the opal
infiltrated with PDMS a HG (vertical pattern) appears, too [35]
f ¼ qBZ ceff =2p: ð5:11Þ

The position of the gap can be widely tuned by changing the lattice parameters of
the hypersonic phononic crystal due to the linear dependence of the gap frequency
from the inverse diameter.
The width of the BG increases with elastic mismatch between the spheres and the
infiltrated liquid. In fact, the width of the gap in the wet PS opals systematically
increases when the infiltrated liquid changes from glycerin ðcl 2; 200 ms1 Þ to
silicon oil (cl = 1,400 ms-1) and to PDMS (cl = 1,050 ms-1). According to the
results of effective medium theory [31, 37], the value of ceff lays in between the
components’ individual sound velocities. It is noteworthy that the calculation of
effective sound velocities following the approach of Gaunaurd and Wertman [31]
tends to result in moderately but certainly too small values compared with the
experiment, actually the deviations are smaller than 10%.
In Sect. 5.3.1 the first experimental realization of an additional gap besides the
BG, the so-called hybridization gap (HG) is demonstrated [35]. This HG originates
from the interaction of the band of quadrupole particle eigenmodes with the band of
the effective medium. This kind of gap opens up from level repulsion when two
bands of the same symmetry cross each other (in analogy to the linear combination
of atomic orbitals) [38]. In this case, the involved bands are the acoustic field of the
extended states in the effective medium and the bands from the multipole modes of
the interacting spherical particles, which can be treated as local resonant elements.
The first hint on this gap was found in BLS experiments in colloidal suspensions
[39, 40], but a clear demonstration, assignment and the concurrent observation along
with the BG was just reported in the case of PS and PMMA wet opals infiltrated with
PDMS.
5.1 Introduction 99
5.1.3.1 Normalization
In general, the condition that has to be fulfilled to create a Bragg gap is that the
distance between two scattering planes in the direction of the incident wave
matches the wave length of the wave, or, in reciprocal space, that the wave vector
q is half the diameter of the first Brillouin zone in a certain crystallographic
direction. By the method of vertical lifting deposition (Sect. 3.1 in Chap. 3) the
colloidal fcc crystals show a preferred orientation relative to the substrate. When
inserting the sample in the same orientation as prepared into the BLS setup and
utilizing the special transmission geometry, where the wave vector is parallel to
the substrate, the probed direction in the reciprocal lattice is usually the C–M
direction. The first BZ of an fcc lattice is shown in Fig. 5.5. In the C–M direction,
the distance in reciprocal space between C and M is
33=2
p
jCMj ¼ 2pffiffiffi ð5:12Þ
2d
with sphere diameter d. Therefore, the wave vector qBZ in this crystallographic
direction must have the same value to fulfill the Bragg conditions, because |C–M|
is half the diameter of the Brillouin zone DBZ, i.e.,
33=2
DBZ p
qBZ ¼ ¼ 2pffiffiffi : ð5:13Þ
2 2d
If one assumes pure acoustic behavior with an effective sound velocity ceff the
frequency at qBZ would be
Fig. 5.5 First Brillouin zone

(BZ) of an fcc lattice. The
black line denotes the equator
for the body standing on a
hexagon, on which the point
M is located
33=2
ceff qBZ ceff DBZ ceff
fBZ ¼ ¼ ¼ 2 pffiffiffi ð5:14Þ
2p 4p 2 2d
with f = x/2p = cq/(2p). Now one can normalize the measured values for the
frequency and the wave vector to the values at the edge of the Brillouin zone:
pffiffiffi
f 2f 2d fd
fnorm ¼ ¼ 3=2 1:54 ð5:15Þ
fBZ 3
ceff c eff
2
pffiffiffi
q q 2d
qnorm ¼ ¼ 3=2 0:245qd ð5:16Þ
qBZ 3
p
2
5.2 Effective Medium Velocity in Defect Doped Opals
Brillouin light scattering can be used to measure the effective sound velocity in
homogeneous materials as well as in structured materials if the structure size is
smaller than the traveling phonon, or in other words, if 2p/q [ d with structure
spacing d. E.g., for q = 0.01 nm-1 (that corresponds to an angle of 50 in
transmission geometry for the setup described in Sect. 2.3.2 in Chap. 2) the
structure must be clearly smaller than 600 nm.
While in bulk, not too thin films of a single polymer of course the polymer’s
sound velocities (if it support shear waves) are found, in films of thin multilayers
of different components, e.g., for two different polymers, the individual sound
velocities of both components are found when measuring parallel to the interfaces
[41]. If the wave vector is chosen perpendicular to the layers, at least for thin
(\30 nm) layers, only one signal corresponding to the effective sound velocity is
found. Following the effective medium theories, ceff lays in between the velocities
of the individual components [42].
Effective sound velocities are also found in nano- and mesoscopic 3D structures
like the infiltrated colloidal crystals presented in Sect. 5.1.3. In Fig. 5.4 ceff is shown
for PS colloidal crystals (/ = 0.74) infiltrated with silicon oil (left) or PDMS (right).
The effective sound velocities are 1,990 and 1,670 ms-1, respectively. These are
between those of PS (cl = 2,350 m/s, . ¼ 1; 050 kg m-3) and PDMS (cl =
1,050 m/s, . ¼ 965 kg m-3) or silicon oil (cl = 1,400 m/s, . ¼ 1; 000 kg m-3)
[26]. The EMT of Gaunaurd and Wertman [31] delivers effective velocities of 1,714
and 1,488 m/s, for silicon oil and PDMS as infiltration liquid, respectively, using the
numbers given above. Wood’s law, on the other hand, delivers 1,925 and 1,617 m/s,
respectively, i.e., both EMTs seem to underestimate the real effective sound
velocities, however for Wood’s simple law, the error is significantly smaller.
A more systematic study of the influence of the filling fraction on the effective
sound velocity in PS colloidal crystals infiltrated with PDMS and the comparison
with EMT was performed within the framework of this thesis. Therefore, a set of
colloidal crystals consisting of PS colloids (d = 260 nm) and a certain amount of
silica particles with practically the same diameter (d = 255 nm) were synthesized
5.2 Effective Medium Velocity in Defect Doped Opals 101
by my coworker Markus Retsch (who also performed the characterization in

Fig. 5.6). Since the size difference is clearly below 5% [44], the particles form
random co-crystals during vertical lifting deposition under stirring. Figure 5.6a–d
shows the SEM pictures of four different co-crystals with composition of the initial
particle dispersion PS:SiO2 between 2:1 (a) and 12:1 (d); the brighter spheres
correspond to the silica particles. It can be seen that the samples are nicely
crystalline and nearly crack- and defect-free over many micrometer, i.e., the filling
fraction / is in good approximation that of an fcc lattice, / = 0.74.
It must be expected that the real composition in the composite crystals differs
from that in the initial dispersions due to the different mass density of both types of
particles and other possible surface effects that cannot be fully discussed in this
context. In order to determine the real ratios, TGA (Fig. 5.6i) and UV/VIS
(a) (b) (c) (d)
(e) (f) (g) (h)
(i) (j)
Fig. 5.6 SEM pictures: a–d PS:SiO2 colloidal crystals with composition 2:1, 4:1, 8:1, and 12:1;
e–h The corresponding defect doped crystals after etching away the silica beads. Analytics:
i TGA measurement on the composite PS:SiO2 colloidal crystals (decomposition of PS). The
inset shows the loss of water (legend like in j); j UV/VIS spectra of the PS:SiO2 colloidal crystals
shown in a–d. For comparison also the spectra of the pure PS colloids, the pure silica colloids and
of two mixtures with higher amount of silica are shown (according to [43])
measurements (Fig. 5.6j) were performed and the results were compared with the
counting of the SEM pictures. From the TGA measurements it is easy to calculate
the PS:silica ratio at high temperatures the PS is decomposed and only the silica
spheres remain. By comparing the relative mass loss and taking into account the
different mass densities and the exact radius of the spheres the relative amount of
the polymer is obtained [43]. It should be noted that it is meaningful to compare
the change for the dry samples i.e., to set the starting point of the comparison at
T [ 100 °C (inset in Fig. 5.6i).
The counting of the SEM pictures shows that systematically less silica particles
are built in the crystals than should be expected from the initial dispersion
concentrations. However, the SEM pictures show only the distribution on the
imaged surface and, by what reason ever, the ratio inside the opal could be different.
Anyhow, the comparison between the data obtained by the TGA measurements and
the SEM counting lead to a systematic and consistent result [43]. The amount of
silica is always given by approximately 0.8 times the expected fraction, i.e., the real
ratio of the ‘12:1’ PS:SiO2 sample is &12:0.8, etc., however, for the rest of this
section the names originating from the initial ratio are kept to avoid confusion.
The so found real compositions are verified by the UV/VIS measurements
shown in Fig. 5.6j. From the graph that contains also two additional samples with
initial ratio 1:1 and 0.3:1 it can be seen that there is a clear blue shift of the Bragg
peak for increasing amount of silica, resulting from the change in the effective
refractive index (cf. Sect. 3.3.6 in Chap. 3). The utilization of the modified Bragg
equation (Eq. 3.10)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kBragg; theo ¼ 1:63d 0:74vPS n2PS þ 0:74vSiO2 n2SiO2 þ 0:26n2air ð5:17Þ
delivers excellent agreement between the theoretically expected Bragg wavelengths

kBragg, theo and the experimentally measured kBragg as far as refractive indices a bit
lower than for the bulk materials are assumed (nPS = 1.55, nSiO2 ¼ 1:37 [43]).
Since the number ratio of PS and SiO2 spheres in the composite is clear now,
the next step is to remove the silica spheres by etching with hydrofluoric acid.
Figure 5.6e–h shows the SEM pictures of the composite crystals after etching
away the silica beads. The removed spheres appear now as randomly distributed
hole defects in the colloidal crystal. From the SEM pictures it is obvious that for
the chosen compositions the crystalline structure is not disturbed (e.g., by a
collapse). Therefore, the filling fraction of the whole crystal is now simply given
by 0.74 times the experimental ratio PS:(PS + SiO2) before the etching.
After infiltration with PDMS, the polarized Brillouin spectra are measured by
BLS in the long wavelength limit. The results are displayed in Fig. 5.7. All signals
increase linearly with q with increasing slope for increasing ratio PS:SiO2 before
the etching, i.e., with increasing filling fraction, which is expected due to the
higher cl of bulk PS compared to bulk PDMS. For all samples, ceff is calculated
from the linear fits as ceff = 2pdf/dq.
The experimental ceff, cexp, is compared to theoretical predictions of the EMT in
Table 5.3, utilizing the real filling fractions / from the experimental ratio in
5.2 Effective Medium Velocity in Defect Doped Opals 103
Fig. 5.7 Dispersion at low

qs and ceff for defect doped
PS opals infiltrated with
PDMS
Table 5.2. First, the velocities are compared to the results of Wood’s simple law,
using the standard elastic parameters for PS and PDMS (given under the table).
Interestingly, cWood seems to capture the experimental values quite well within a
non-systematic error of 1.5% or less, which is in the range of the error of /. Using
the same parameters in the algorithm of Gaunaurd and Wertman (cGW,s) leads to a
significant deviation from the experimental values between 6% and more than 9%,
while the deviation increases systematically for increasing /.
Anyhow, it is possible to bring the results of their EMT in better accordance with
the experimental results when taking cl(PS), ct(PS), and cl(PDMS) as fit parameters
in a mean square fit for all five filling fractions. So the cGW,f are obtained, keeping
the parameters in relatively reasonable ranges. The PDMS is assumed to be about
8% softer (1,109 ms-1 instead of 1,200 ms-1), for the PS the transverse sound
velocity must be chosen smaller (1,100 ms-1 instead of 1,200 ms-1) and the
longitudinal one must be chosen higher (2,470 ms-1 instead of 2,350 ms-1). Doing
so, the error stays within about 1%, however changing systematically from a little to
high values for low / to a little to low values for the undoped crystal. Even better
results can be achieved using really free fit parameters without any fitting ranges.
Table 5.2 Characterization of the PS:SiO2 composite colloidal crystals

Initial ratio Experim. ratio Mass lossa (%) kbBragg, theo (nm) kBragg (nm)
PS:SiO2 PS:SiO2
0:1 0:1 – 533.7 530.0 ± 7.9
2:1 2:0.8 60.7 588.3 586.0 ± 1.6
4:1 4:0.8 75.5 595.3 593.0 ± 3.6
8:1 8:0.8 86.1 599.7 599.0 ± 0.8
12:1 12:0.8 90.3 601.4 601.5 ± 0.5
1:0 1:0 – 605.0 602.1 ± 3.3
The initial ratio is also the name of the samples in Figs. 5.6 and 5.7
a
.PS ¼ 1:05 g cm-3, .SiO2 ¼ 1:7 g cm-3, dPS = 260 nm, dSiO2 ¼ 255 nm
b
nPS = 1.55 g cm-3, nSiO2 ¼ 1:37 g cm3 , d = 260 nm
Table 5.3 Comparison between experimental effective longitudinal sound velocities and results
from Wood’s law and the EMT of Gaunaurd and Wertman using standard and fitted parameters
(in ms-1)
PS:SiO2 / cexp cWood D/% cGW,s D/% cGW,f D/%
2:1 0.529 1,382 1,362 -1.5 1,304 -6.0 1,395 1.0
4:1 0.617 1,464 1,451 0.9 1,372 -6.7 1,473 0.6
8:1 0.673 1,522 1,520 0.1 1,421 -7.1 1,530 0.5
12:1 0.694 1,567 1,548 -1.2 1,441 -8.7 1,553 -0.9
1:0 0.740 1,627 1,617 -0.6 1,488 -9.3 1,609 -1.1
Constant: .(PS) = 1,050 kg m-3, .(PDMS) = 965 kg m-3, ct(PDMS) = 0 ms-1; cWood, cGW,s:
cl(PS) = 2,350 ms-1, ct(PS) = 1,200 ms-1, cl(PDMS) = 1,050 ms-1; cGW,f: cl(PS) = 2,470 ms-1,
ct(PS) = 1,100 ms-1, cl(PDMS) = 1,109 ms-1
However, the best fit is achieved with unreasonable parameters, especially for PS
(cl(PDMS) = 1,069 ms-1, cl(PS) = 2,622 ms-1, and ct(PS) = 1,082 ms-1).
In summary, it can be said that for different filling fractions the investigated
colloidal system shows the expected trend in the effective medium sound velocity.
The increase in ceff with increasing amount of the harder component was predicted
qualitatively by Wood’s law as well as by the more elaborated EMT of Gaunaurd
and Wertman, developed for the special present case of spheres embedded in an
elastic or fluid matrix. However, while the predictions utilizing Wood’s law agree
also quantitatively well with the experimental results, there is a systematic deviation
to the velocities calculated by the other EMT. The reason of the deviation is unclear.
It might originate in a kind of size or frequency effect that leads to deviations for
small spheres (possibly originating from elastic confinement inside the spheres) or
for high frequencies. Anyhow, these assumptions are somehow speculative.
5.3 Band Gaps in Polymer Opals and Disordered Systems
5.3.1 The Influence of the Order in Colloidal Systems
Structured materials with a periodic modulation in the density and elastic coeffi-
cients, so-called phononic crystals [11], can exhibit phonon band gaps at Bragg
frequencies or wavelengths commensurate to their lattice constant. In addition to
Bragg gaps (BG), theory predicts gaps evoked by resonance modes of the
constituent components interacting with the extended acoustic branch of the
composite structure [45]. These gaps prevent elastic waves with certain frequen-
cies to propagate through the crystal at least in certain crystallographic directions.
The width and the position of the BG in general depends on the contrast between
the densities (.), longitudinal and transverse sound velocities (cl and ct) of the
component materials, and on the lattice parameter [9, 10, 33]. Yet, structures with
strong localized resonant elements can shift the gap well below the Bragg
frequency associated to the lattice constant [8, 46]. Soon after the first
5.3 Band Gaps in Polymer Opals and Disordered Systems 105
experimental observation of ultrasonic band gaps [4–6, 8, 13], peculiar phenomena

with potential applications dealing with the propagation of elastic waves in
periodic composite materials such as tunneling [47], negative refraction, focusing
[48, 49], and enhanced transmission through 1D gratings [15] have been discov-
ered. Earlier realized systems with periodicity in the millimeter range deal with
sonic and ultrasonic frequencies [5–8, 13, 46–49], while periodic patterns at the
submicrometer scale becomes necessary to shift the gap to hypersonic frequencies
[18, 50]. The phononic dispersion relation of such a single crystalline pattern
created by vertical lifting deposition of polymer colloids and measured by BLS led
to the first observation of a hypersonic BG along high-symmetry crystallographic
directions [17]. By varying the particle size and the infiltrated fluid, the hypersonic
frequency and the width of the gap could be tuned. More recently, a hypersonic
BG was also observed in 2D structures fabricated by optical lithography [36].
In soft opals, the spherical particles represent local resonant elements and hence
the bands, which originate from the multipole modes of these interacting particles,
can overlap with the acoustic field of the extended states in the effective medium.
As a result of this hybridization, a gap opens up in the vicinity of the eigenfrequency
of the quadrupole particle modes [45], which is referred to as hybridization gap
(HG). Albeit there was a hint of this HG in colloidal suspensions [25, 39, 40, 51], its
demonstration, assignment and the concurrent observation along with the BG has
been missing. In this section, the first realization of a double phononic gap in fcc
colloidal crystals formed by colloidal self-assembly during vertical lifting deposi-
tion with subsequent fluid infiltration is presented [52]. The effect of the elastic
contrast between the fluid matrix and solid inclusions as well as the crystalline order
of the structure on the two gaps is distinct. Furthermore, in amorphous colloidal
glasses, only the HG persists.
The fabrication of the colloidal films starts with the deposition of the particles by
vertically lifting a glass substrate from the aqueous colloid suspension [53]. For
different experiments, two kinds of monodisperse polystyrene (PS) spheres with
diameters of d = 307 and 360 nm and poly(methylmethacrylate) (PMMA) particles
with a diameter of d = 327 nm were used for the fabrication of the opals.
Additionally, a noncrystalline hybrid film consisting of PS sphere mixture with two
different diameters (300 and 360 nm) was prepared in the same way. After complete
drying, the films were infiltrated with liquid poly(dimethylsiloxane) (PDMS) and
any liquid excess was blown off (Fig. 5.3). By this method uniform wet colloidal
films with a thickness of about 10 lm were obtained. The phononic dispersion
relation was measured by BLS for the dry and liquid infiltrated state of these films.
BLS probes the thermal phonon propagation in the sample along a selected direction
determined by the transmission scattering geometry adumbrated by Fig. 5.8
(cf. appendix) [17]. For the given experimental conditions, the scattering wave
vector of the photon lies in the (111) plane of the fcc lattice and its amplitude
q ¼ ð4p=kÞ sinðh=2Þ depends on the scattering angle h and the wavelength of the
incident laser beam k. At low q values for which the system appears homogeneous,
the wave vector of the phonon k and that of the inelastically scattered light q are
equal. Under these conditions, the BLS spectrum at a given q consists of a doublet
Fig. 5.8 BLS parallel to the

substrate in a colloidal crystal
with a Doppler frequency shift x = ±ck, where c is the speed of sound with
longitudinal (or transverse) polarization in the effective medium (e.g. lowest
spectrum in Fig. 5.10, cf. Sect. 2.3 in Chap. 2).
The precursor dry PS and PMMA opals have a highly ordered crystalline
morphology as indicated by the SEM image in Fig. 5.9. The single crystalline order
extends over a few hundred micrometers. The dry films exhibit strong multiple light
scattering due to the large optical contrast between the particles and the surrounding
air and due to the particles elastic form factor. The wave vector q is, therefore,
ill-defined which precludes the measurement of the dispersion relation x(q).
Nonetheless, the BLS spectrum of the dry opals reveals several localized
(q-independent) modes [54, 55]. As discussed in Chap. 4, these are identified as
Fig. 5.9 SEM top-view images of the PS-307 crystal (left) and the PS-hybrid (right)—both
‘‘dry’’ before infiltration (scale bar is 1 lm). The insets display the Fourier-transform images
computed from the SEM pictures over an area of about 4 by 4 lm [26]
Table 5.4 Sound velocities, particle sizes and lowest eigenfrequencies in colloid-based
phononics
Material ct/ms-1 cl/ms-1 d/nm f(1, 2)/GHz c/ms-1
PS 1,200 ± 30 2,350 ± 50 307 3.3a
360 2.9a
PDMS 0 1,050 ± 20a
PS/PDMS 1,670 ± 30a 307 2.2b 1,490b
1,570 ± 30a 360 1.9b 1,490b
Hybrid/PDMS 1,510 ± 20a 300/360 1,400b, c
a
PMMA 1400 ± 40 2,800 ± 50 327 4.0
PMMA/PDMS 1,720 ± 30a 327 2.3b 1,560b
Densities . (kg/m3): PS: 1050, PMMA: 1,190, PDMS: 965
a
Measured by BLS
b
Computed for the wet opals
c
For 65% filling fraction of spheres
vibration eigenmodes of the particles and each mode can be specified by a pair of
indices (n,l) defining the lth spherical harmonic of the nth radial mode. The
frequencies of these sphere eigenmodes depend on the size of the particle, the mass
density and the speed of sound in air and in the particle for both polarizations (i.e.
compression and shear moduli). Consequently, the mechanical properties of the
samples can be reliably determined from the theoretical fit of the experimental
eigenfrequencies. The obtained longitudinal and transverse sound velocities in the
two types of particle materials along with the lowest f(1, 2) eigenfrequency are
given in Table 5.4. For the particles embedded in the infiltrated PDMS, the eigen-
frequencies can be theoretically computed using the elastic parameters of Table 5.4
[51, 56]. Expectedly, f(1, 2) decreases when the spheres are embedded in PDMS.
The infiltration of the thin dry opals by PDMS with a refractive index (n = 1.45)
close to that of the PS particles (n = 1.59) diminishes multiple light scattering and
hence q is well defined. To obtain the desired dispersion relation x(q), the BLS
spectra of the wet opals were recorded along the high-symmetry directions of the
reciprocal space. In the present case of the fcc lattices, the first Brillouin zone (BZ) is
a truncated octahedron (Fig. 5.5). The experimental q is confined in the hexagon
formed by the intersection of the (111) plane with the BZ [17]. The direction of q is
selected along C–M, where M is the edge center of the hexagon and the evolution of
the BLS spectra with q near the BZ for the PS wet opal is shown in Fig. 5.10. The
simple picture of single phonon propagation in the effective medium at low q values
becomes complex as q increases towards the BZ boundary. Up to four Lorentzian
curves are required to represent the experimental BLS spectrum. In contrast to the
PS/silicon oil opal with a single peak splitting across the BZ boundary [17], the same
opal infiltrated in PDMS providing higher elastic contrast displays richer spectral
features and exhibits a second splitting at lower q values within the first BZ. The
experimental dispersion relation is depicted in Fig. 5.11. In the hypersonic PS/
PDMS crystal (d = 307 nm) only one longitudinal phonon branch is observed at
low q values with cl = 1,670 ms-1, intermediate between the longitudinal sound
Fig. 5.10 BLS spectra of the

wet opal of PS spheres with
diameter 307 nm in a PDMS
matrix at different wave vec-
tors in the (111) plane of the
fcc crystalline colloidal film.
The edge of the BZ along the
probed direction corresponds
to qBZ& 0.013 nm-1. The
deconvolution in different
spectral components is
indicated for the Stokes side
of the spectrum. Each
spectrum is normalized to the
total integrated intensity, for
comparison of the relative
contributions [57]
velocities in the pure component materials. Transverse phonons are not observed
experimentally, probably PDMS cannot support shear waves and the mechanical
contact between the particles is weakened. The most striking feature of the
dispersion diagram is the simultaneous presence of two band gaps at about 3 and
4 GHz, respectively. The latter is clearly a BG since it occurs at the edge of the BZ,
q ¼ qBZ 0:0133 nm1 ; that matches the distance C–M, i.e. (3/2)3/2p/a, where
pffiffiffi
a ¼ 2d is the lattice constant of the given fcc crystal. An analogue behavior is
observed in the second wet opal of the larger PS spheres (d = 360 nm) for which the
gap positions shift to lower wave vectors and frequencies (open circles in the upper
left diagram of Fig. 5.11).
Fig. 5.11 Phononic band

diagrams for the different PS
wet opals under consideration,
along the C–M direction (see
text). The hatched bands mark
the phononic gaps and the
vertical arrows indicate the
q values in the corresponding
spectra of Figs. 5.10 and 5.12.
The vertical lines denote the
first BZ limit (for d = 307 and
d = 360 nm, respectively).
The change in the DOS
induced by the corresponding
single PS particle in PDMS is
shown in the right-hand panel
[26]
In order to elucidate further the nature of the two gaps, the effect of the
crystalline order on the experimental band diagram was examined. The formation
of a hybrid colloidal film consisting of a mixture of an equal number of two PS
spheres (d = 300 nm and d = 360 nm) artificially broadens the size distribution
beyond the polydispersity limit of about 5% necessary for crystallization [44].
Indeed, the crystallization is prohibited in this hybrid colloidal film, as indicated
by the lack of a long-range order in the right-hand panel of Fig. 5.8, leading to an
amorphous colloidal glass. The BLS spectra of the infiltrated monodisperse opal
and hybrid films are shown in Fig. 5.12 for two wave vectors. The deletion of one
peak in the spectrum of the hybrid is due to the disappearance of the BG in the
disordered hybrid, as it is clearly shown in the dispersion plot in the upper left
diagram of Fig. 5.11 (solid squares). Apparently, the crystalline order is a
prerequisite for the BG but not for the newly observed lower frequency HG, which
is omnidirectional in the colloidal glass. An identification of the latter as a
theoretically anticipated HG [45], through density-of-states (DOS) calculations
[56], is examined in Fig. 5.11. The right-hand panel of Fig. 5.11 displays several
eigenmodes of the individual PS spheres embedded in the fluid PDMS. The lowest
f(1, 2) appears to compare well with the frequency at the crossing with the acoustic
branch and the opening of the HG. Moreover, the lowest frequency points at the
experimental dispersion at high qs compare very well with the computed f(1, 2).
This and the two higher flat bands of localized modes in Fig. 5.11 cause
sufficiently strong inelastic light scattering at high qs [51] and compare well with
the particle resonances of higher l. Their presence in the BLS spectra obscures the
resolution of the phonons in the second BZ. Interestingly, the bands originating
from particle resonant modes appear to be considerably narrower and occur at
higher frequency than theoretically predicted [45]. This can be ascribed to viscous
losses in the liquid matrix that were not taken into account in the theoretical
calculations, and which weaken the interparticle interactions as a result of the
reduced overlap between the corresponding wave fields. Qualitatively, the opening
of the two gaps is also observed in a third wet opal of PMMA with d = 327 nm
in PDMS (Fig. 5.13). The HG occurs at 2.5 GHz, very close to the f(1, 2) =
Fig. 5.12 Exemplary BLS

spectra of the PS
(d = 360 nm)/PDMS wet
opal (bottom) and the wet PS-
hybrid colloidal film (top) at
two q values indicated by the
arrows in the corresponding
diagram of Fig. 5.11. The
deconvolution into different
spectral components is shown
below each spectrum (note
the absence of the BG-
induced spectral features in
the hybrid) [26]
Fig. 5.13 Phononic band diagrams for PMMA wet opals in PDMS with d = 327 nm, along the
C–M direction. The change in the DOS induced by the corresponding single PS particle in PDMS
is shown in the right-hand panel [26]
2.35 GHz of this particle in PDMS, whereas the wet opal exhibits a sound velocity
of 1,720 ± 30 ms-1 in the long-wavelength limit (Table 5.4).
It is worth noting that the various effective-medium theories (EMT) yield sound
velocities about 10% smaller than the experiment [31, 37], even if the viscosity
in the liquid matrix is taken into account. On the other hand, if one considers the
colloidal film as a polymer (solid) matrix with fluid inclusions [37], EMT strongly
overestimates the sound velocity. The above, in view also of the fact that measured
sound velocities agree generally well with the results of EMT in non-close-packed
colloidal crystals [58], suggests the existence of consolidation, at least to some
degree that may be different for different samples, in the colloidal films [59].
This also explains why the measured effective sound velocity is different in the
two PS/PDMS crystals (see Table 5.4).
In conclusion, this work presents, for the first time, the discovery of two
phononic band gaps of different nature coexisting at hypersonic frequencies in the
same physical system and elucidated the underlying physical mechanisms. Induced
disorder did not destroy the newly demonstrated HG. This study has been possible
by taking advantage of the opportunities offered by the colloidal science to tailor
the phononic band diagram of nanostructured materials, measured directly by
BLS. Manipulating the flow of phonons may allow heat management, e.g. in
thermoelectrics. Finally, it was pointed out the need of a detailed quantitative
evaluation of the dispersion diagrams of colloid-based phononic structures, by
means of full elastodynamic calculations that take into account consolidation and
soft matter properties (e.g. structural dynamics of the component materials or
interfacial effects), which still remains an open challenging theoretical problem.
5.3.2 The Influence of the Composition in Disordered Colloidal

Systems
In Sect. 5.3.1 the realization of a hybridization gap (HG) was presented for
monodisperse colloidal crystals as well as for disordered ‘hybrids’, e.g., for the
1:1 number ratio 300:360 nm PS hybrid in Fig. 5.11. However, when discussing
the HG at that point, the mixing of two different sizes of colloids was only
mentioned in order to introduce structural disorder, i.e., to destroy the colloidal
crystal, but the influence of the composition on the position of the HG was not
elucidated. Indeed, in the top panel of Fig. 5.11 it seems that the HG for the pure
360 nm opal and the 1:1 hybrid come more or less at the same position. In this
section, an experiment is presented in which the number ratio of two kinds of PS
colloids with clearly different size (180 and 360 nm) is varied in a wide range.
The HG for these systems infiltrated with PDMS is observed and the changes in
width and position are discussed; of particular interest is the question if it is
possible to tailor this band gap by simply controlling the size composition of the
particles.
The set of samples has been realized by Retsch mixing distinct amounts of
180 nm PS beads and 360 nm PS beads in aqueous dispersion and preparing the
hybrids by vertical lifting deposition under continuous stirring in order to
guarantee the homogeneous distribution of both particles in the whole sample.
Doing so, hybrids with composition between 1:1 and 40:1 180:360 nm particles
number ratio have been prepared, i.e., the volume ratio varied between 1:8 and 5:1
for the smaller particles compared to the larger ones. After infiltration with PDMS,
BLS was performed in transmission geometry and the dispersion relation was
obtained fitting the spectra with Lorentzians in the q-range between about 0.006
and 0.020 nm-1. The result is shown in Fig. 5.14.
Figure 5.14 contains the experimental band diagrams for the mentioned
systems as well as for bare 180 nm PS particles. The position of the HG is marked
by colored rectangles. Note that for all systems only one HG between &4 and
5.5 GHz was found in the experimentally accessible range, only for the 1:1 hybrid,
a second HG was observed between &2 and 2.5 GHz. By comparison with the HG
from the bare 180 nm PS particles and taking into account the scaling of the HG
with 1/d, it is obvious that this gap originates (mostly) from the 360 nm particles.
The 1:1 hybrid is the only one whose volume fraction of the larger particles
exceeds that of the smaller spheres. Anyhow, in the rest of this section, I want to
focus on the higher frequency HG above 4 GHz.
Indeed, when comparing the HGs of the bare 180 nm particles and of the hybrids,
one finds that the HG for every hybrid is shifted to lower frequencies in a mostly
systematic way. The higher frequency HG of the 1:1 hybrid is distinctly lower and
smaller than that of the opal. With increasing amount of 180 nm particles in the
mixture, the gap widens up and is shifted to higher frequencies. This is true at least
for the 9:1 and the 20:1 hybrid, the 40:1 hybrid, however, seems to contradict the
trend, although its HG is still broader and at clearly higher frequency than that of the
Fig. 5.14 Development of

the HG in binary PS hybrids
infiltrated with PDMS as a
function of the composition.
The colored rectangles mark
the HGs for the different
systems. The empty black
rectangle corresponds to the
HG from the 180 nm PS
opals after correction for
effective sound velocity. The
light red rectangle around
2.5 GHz for the 1:1 hybrid
belongs to the HG originating
from the larger spheres
1:1 hybrid. On the other hand, when ignoring the lowest point in the upper branch of
the 40:1 sample, also this sample would follow the trend.
When comparing the opal with the hybrids, one should note that for an fcc
colloidal crystal of course the filling fraction (/ = 0.74) is different than for a
disordered system. Approximating a uniform filling fraction of / = 0.65 for all
hybrids [60], which is admittedly a bit arguable especially for the extreme ratios,
one can estimate that for the infiltrated opal the effective longitudinal sound
velocity ceff should be higher by about 6.3 or 8.5%, applying the EMT of
Gaunaurd and Wertman [31] or Wood’s law [30], respectively (cf. Sect. 5.2)—in
reasonable agreement with the experimental ceff. Since the position of the band
gaps scales with ceff, it is possible to normalize the frequencies by that. Doing so,
also the HG of the 180 nm opals is shifted towards somehow lower frequencies,
still following the trends mentioned above. In Fig. 5.14 the normalized (relative to
the amorphous systems) HG of the PS 180 nm opal is symbolized by the open
black rectangle.
To rationalize the findings in this experiment one must go back on the nature of
the HG (Sect. 5.3.1). The HG originates from the level repulsion between the
bands of the linear acoustic mode and those from localized modes, i.e., the
eigenvibrations inside the spheres. The generation of a band originating from these
eigenvibrations can be described in analogy to the tight coupling method for the
electronic states in solids, in which the overlap of atomic orbitals is assumed to be
sufficient to require corrections of the picture of the isolated atom, but not so much
to render the atomic description completely irrelevant [29]. This theory predicts
that in a sample of N atoms each electronic state consists of an N-fold degenerated
level. If the atoms come closer together, i.e., if their wave functions overlap
significantly, the levels broaden up into bands. In the analogy of the phononic
system, the individual spheres play the role of the atoms, and instead of electronic
wave functions one deals with the acoustic resonance modes of the spheres
(as ‘acoustic wave functions’), which are both described by the same mathematics
(cf. Sect. 2.3.3 in Chap. 2).
This means that in order to obtain a strong band, the spheres must be in close
contact to each other and the acoustic wave function must overlap with that of the
next neighbor. Since the spheres are close-packed (or randomly close-packed) in
all cases, bands can be assumed. The question is how do modes of the same type
but from different particles sizes interact with each other. From symmetry
considerations, there is no reason for them not to interact, i.e., to hybridize.
However, if their energy levels are too different, they should be not able to interact
anymore. Anyhow, in the present experiment the frequency of the gaps is shifted,
which is a hint on a shift of the position of the resonance band [in this case of the
(1, 2)-mode], too. The most reasonable explanation is that, although there is a
significant difference in energy for 180 and 360 nm spheres, the (1, 2) modes of
the 180 nm spheres couple with those of the 360 nm spheres, leading to shift to
lower frequencies with increasing fraction of the 360 nm spheres wave function.
On the other hand, the coupling seems to be weak enough to allow two HGs for
the 1:1 hybrid. Since the higher frequency HG is clearly shifted towards lower
frequency, it is obvious that there cannot be two fully decoupled bands, at least
there must be some influence of the bigger spheres on the localized mode of the
smaller spheres. For lower number ratios of the larger spheres the low frequency
HG completely disappears, which is another hint for the rightness of the
conclusion that this gap must originate from the 360 nm particles, which are too
separated in the other hybrids to interact effectively. The smaller width especially
for the 1:1 hybrid can be rationalized by the larger average sphere to sphere
distance of the small spheres in the sample.
In summary, one can conclude that the experiment can be principally captured
qualitatively by simple hybridization considerations. However, there are still open
questions that could only be solved with additional experiments and—even more
needed—strong theoretical support. From this section, I would like to retain the
message that the distribution of colloids of different size has an influence on the
hybridization gap. Under certain circumstances, the mixing of particles of different
sizes may be a way on which future studies to tailor the band gap could proceed.
5.4 Band Gaps in SiO2 Colloidal Systems
In the previous sections of this chapter it was shown that the elastic contrast
between matrix and scatterers is crucial for phononic phenomena. Another
important role is played by the density contrast, which is, however, usually quite
weak for polymer/liquid systems. In this section, experiments on silica colloids in
liquid or elastic matrix are shown, i.e., the elastic contast is relatively high as well
as the density contrast is. In such systems, additional effects can be found that
cannot be fully captured by theory (and the explanations in the previous chapters),
yet.
5.4.1 Phononic Behavior of Silica Suspensions
In Sect. 4.2.4 (Chap. 4) a system consisting of dense packed silica spheres in an

index-matching liquid ethoxy-ethoxyethyl acrylate (SR256) matrix was introduced
to show the q-dependence of the light scattering intensity of the particles eigen-
modes. Since the eigenmodes appear at high f compared to the longitudinal
acoustic branch, there was no need to go into the development of the acoustic
behavior on that point. However, since, in first approximation, the acoustic branch
in Fig. 4.12 in Chap. 4 appears to be linear, a more detailed investigation of the
dispersion relation of the acoustic mode shows some surprising details, which will
be presented in this section.
The investigated materials are the sample presented already in Sect. 4.2.4
(Chap. 4, d = 375 nm) as well as an analogue sample with d = 219 nm silica
spheres. The samples have been synthesized by Kiefer at DKI with an initial con-
centration (filling fraction) of / = 0.34. A dense packing (/ = 0.74) was achieved by
ultra-centrifugation on the bottom of an NMR-tube (see also Fig. 1.3b in Chap. 1).
Figure 5.15 shows ten exemplary spectra for the dense packed sample of
d = 375 nm spheres in the range 30° B h B 75°, i.e., 0.009 nm-1 B q B
0.020 nm-1, recorded at low free spectral range. In this q-range, the general shape
of the observed signal changes dramatically. Since at the lowest angles the signal
is clearly a single peak, the signal broadens rapidly with increasing angle.
Furthermore, between 30° and &50° the frequency of the signal is virtually
unshifted (dashed line), and only the linewidth C increases. At higher angles, f(q)
is not longer constant and the frequency increases again with q, although the
signals remains extremely broad.
The corresponding dispersion relation is shown as open triangles in Fig. 5.15c.
For the lowest qs, the acoustic signal behaves as an effective medium with ceff
cSR256 : For comparison the pure SR256 is shown as gray squares (with linear fit).
From about q = 0.008 nm-1 the dispersion relation shows an abrupt bending, the
effect that is seen in panel a. At q & 0.015 f(q) increases again. Although the
absolute values for f(q) at high qs, due to the large C, depend on the chosen limits
of the fitting range, the trend is clear and looks for an unchanged fitting range as
presented in Fig. 5.15c: f(q) increases monotonically and linearly again, and the
slope of =oq is essentially the same as that of the longitudinal phonon in the matrix
(second linear fit).
An akin dispersion relation is obtained for the smaller silica spheres with
d = 219 nm. However, the bending starts at higher qs and the second linear range
must be expected at qs beyond the range measurable by BLS. In fact, the position
of the bending scales with d-1, and it appears about p/d.
5.4 Band Gaps in SiO2 Colloidal Systems 115
(a) (b)
(c)
Fig. 5.15 a Development of the acoustic signal in the BLS spectrum of d = 375 nm silica
spheres in SR256 (/ = 0.74) for 30° B h B 75°. b Exemplary spectra at h = 40° for low
(/ = 0.34) and high (/ = 0.74) filling fraction. c Experimental dispersion relations for dense
packed silica spheres with d = 219 or 275 nm in SR256. For comparison also pure SR256 as well
as silica with d = 375 nm and / = 0.34 are shown
The appearance of the bending requires high filling fraction, since the suspension
with / = 0.34 (black circles) shows only the effective medium acoustic phonon,
which appears at slightly higher frequencies than for the pure matrix. This is
expected by effective medium theory (cf. Sect. 5.1.2) since the density of the silica
inclusions is much higher than that of the matrix. The high density contrast leads to a
strong scattering of the phonons on the hard spheres; i.e., the phonons travel mostly
through the liquid matrix, and hence, ceff is nearly the same in both cases.
Correspondingly, the linewidth at a given q is smaller for lower filling fraction
as is shown for one exemplary angle (h = 40°) in Fig. 5.15b. C is inversely
proportional to the phonon lifetime. If the phonon path is disturbed by many
scatterers, the phonon lifetime decreases drastically. In fact, if C f ; as it is found
at higher qs for / = 0.74, the free phonon pathlength is akin to the phonon
wavelength; i.e., the phonon cannot travel anymore within the medium (the group
velocity is zero), the system is overdamped. However, at higher qs the group
velocity comes back to its value before the damping.
If one wants to summarize the experimental findings, one should start with the
fact that there is no band gap. At a certain q-range the group velocity becomes
zero, however, there is no forbidden f-range. Since the centrifugated sample is

polycrystalline, the lack of a Bragg gap is not surprising. On the other hand, it is
noteworthy that the bending occurs at p/d, which means q equals the shortest
intersphere distance of touching spheres. Anyhow, higher qs within the bending
correspond to even shorter distances, while other crystallographic directions
correspond to q \ p/d.
The lowest particle eigenmodes are found at clearly higher frequencies (&5 GHz
for the d = 375 nm particles, Sect. 4.2.4 in Chap. 4). Therefore, it can be excluded
that the bending originates from a localization of the phonon in the individual
spheres, similar to HG (Sect. 5.3.1). On the other hand, a localization in the liquid
‘voids’ between the spheres is unlikely, since the liquid forms a continuous matrix,
hence, there are no closed cavities to apply standing-wave conditions. Also the
interpretation of the the cavities as open Helmholtz resonators (similar to those in
musical instruments and subwoofers) seems to be not adjuvant [45, 61].
Another possible explanation for strong localization that does not require a
periodic structure, but, moreover, has ‘large’ disorder as a precondition, is the
Anderson localization of sound [62, 63]. Only recently, John Page and coworkers
could show the realization of the localization of ultrasound in a three-dimen-
sional elastic network consisting of disordered, sintered, monodisperse aluminum
beads [64]. However, the theoretical treatment of Anderson localization is
extremely demanding. Furthermore, it must be challenged that there is enough
disorder in the polycrystalline samples. In addition, in Page’s experiment a band
gap for certain frequencies was found, which is not the case in the experiment
presented here.
The bending must correspond to the appearance of a flat band in the theoretical
band diagram due to the folding of bands in an fcc lattice [65]. An underlying
symmetry can, theoretically, lead to the degeneracy of bands in the theoretical
band diagram. The perturbation introduced by actual scatterers in the theoretical
lattice may remove this degeneracy, leading to non-degenerated flat bands, which
can interact with the acoustic band. In fact, layer multiple scattering calculations
performed by Stefanou and Gantzounis at University of Athens [66], can capture
the experimental band diagram quite well. Figure 5.16 shows the calculated
phononic band structure for an fcc crystal of the given materials along two distinct
crystallographic directions together with the normalized experimental points. The
theoretical band diagram describes nicely the long wavelength limit, but there is
also a flat band expected, which agrees with the flat band observed by BLS.
However, there should be clear band gaps, which are not found in the experiment,
which is most probably caused by a certain directionality in the band gaps, which
is not fulfilled in the polycrystalline sample.
Although the band diagram can be quantitatively captured by theoretical
calculations, the interesting question remains still unsolved: What is the origin
of the flat band? As mentioned above, the explanation given for soft scatterers
(Sect. 5.3.1) that the flat band corresponds to a tight-binding-like process from
localized single particle modes, i.e., eigenmodes, is not suitable for this case,
neither is the localization in single liquid cavities. Rather, the flat band must be
Fig. 5.16 Normalized theoretical band diagrams for fcc opals of SiO2 spheres in SR256 matrix
along two crystallographic directions. The experimental dispersion relations in Fig. 5.15 are
shown as circles and squares
understood as a multiple-scattering effect. Its details, however, are still unclear and
must be developed by theoreticians in the future towards a concrete, descriptive
explanation.
5.4.2 Silica–Poly(ethyl acrylate) Films (PhoXonics)
Another realization of silica based phononic materials are melt compressed films
with silica inclusions [22, 67, 68]. The technique of melt compression is presented
in Sect. 3.2 in Chap. 3 When choosing silica–poly(ethyl acrylate) (PEA) core–shell
particles as the starting materials, films can be pressed with different filling
fraction of silica particles in a rubbery PEA matrix, i.e., with different spacing of
the silica spheres in an fcc lattice.
In this section, the results of BLS measurements on several silica–PEA films is
presented, with dSiO2 ¼ 216 or 253 nm: The total diameter of the initial core–shell
particles was chosen between dCSP = 308 and 498 nm, leading to filling fractions of
/ = ðdSiO2 =dCSP Þ3 between / = 0.08 and 0.36 in the pressed films. The crystallo-
graphic orientation of the silica spheres in the pressed film is known to be an fcc
lattice with orientation of lines of next neighbors in the radial direction from the
center of the film [69]. Due to the melt compression, in the film the silica cores come
nearer then in the initial opal, as is schematically shown in Fig. 5.17b. The shortest
intercore distance becomes (0.74)1/3dCSP along the direction marked by the white
arrow, which is also the main orientation of q in the following experiments. The
pffiffiffi
corresponding lattice constant for an fcc-lattice is a ¼ 2ð0:74Þ1=3 dCSP :
Figure 5.17a shows five exemplary spectra for such a silica–PEA film
(dSiO2 =dCSP ¼ 253=387 nm) at different qs along the direction marked by the white
arrow in panel b. The spectra are represented by a fit of one (lowest q) or two
Lorentzian signals on the Anti-Stokes side. Around q 0:008 nm-1 a band gap
opens up. The corresponding band diagram is shown in Fig. 5.17c, normalized by
the effective medium sound velocity ceff, measured in the long wavelength limit,
(a) (b)
(c)
(d) (e)
Fig. 5.17 a Experimental spectra for a melt compressed film (dSiO2 =dCSP ¼ 253=387 nm) at
different qs, represented by Lorentzian lines. b Scheme of the result of melt compression
(Sect. 3.2 in Chap. 3); the white arrow indicates the orientation of q. c Normalized dispersion
relations for three different films. d Insensitivity of a melt compressed film (217/329 nm) to
stretching e and the orientation of q
and by the intercore distance in the investigated crystallographic direction, which

depends only on the initial total diameter of the core–shell particles (see above).
The diagram contains also the normalized dispersion relations for two films with
dSiO2 ¼ 217 and dCSP ¼ 308 and 329 nm: The dashed lines mark the qBZ, the
wavevector whose magnitude is half the diameter of the first Brillouin zone (BZ)
along the observed direction, and the corresponding frequency. Obviously, a band
gap is observed for all three films at the same normalized q and f, at the edge of the
first BZ, hence it is a Bragg gap (BG) as introduced in Sect. 5.1.3.
The films compared in Fig. 5.17 have / C 0.28. In fact, also two films with
significantly lower / have been investigated, but neither for / = 0.08 nor for
/ = 0.16 a BG was found. These films show only a linear acoustic branch in the
band diagram. These results demonstrate that a certain filling fraction is required to
realize a BG, however, for sufficient elastic contrast, even with filling fractions
much lower then in the crystalline case (/ = 0.74) or the dense amorphous case
(/ = 0.65) (cf. Sect. 5.1.3), a relatively large BG can be realized. The comparison
of the three band diagrams in Fig. 5.17c shows that the width of the gap increases
systematically with increasing amount of silica in the continuous PEA matrix.
Indeed, the ratio w of width of the band gap to its frequency is even higher for the
melt flow pressed films (w = 0.25 for the 217/308 nm system with / = 0.36) than
for the PS/PDMS opal presented in Fig. 5.11 (w = 0.18).
So far, a large BG at low filling fractions was presented in a handy, all solid
material, but these results are basically not astonishing. However, additional, more
elaborate experiments lead to results that elude the straightforward explanation of
BG. Figure 5.17d shows the dispersion relation of the 217/329 nm sample for four
different strains. The results were measured by BLS utilizing a stretching machine
constructed by my colleague Nikos Gomopoulos. Surprisingly, it turned out that the
dispersion relation is virtually unchanged from the unstretched (strain = 100%) film
up to strains of 250%, i.e., films stretched to the two and a half fold of their initial
length (strain = 250%), while the stretching is in the direction of q.
Theoretically, since the band gap is assumed to be a BG, one must expect that
the gap scales inversely with the spacing between the spheres, as is also
observed in Fig. 5.17c. However, when stretching the film, the distance of the
spheres should also change accordingly. From light diffraction experiments on
similar samples, also synthesized by DKI, it is experimentally proven that
stretching clearly changes the intersphere distances [69]. In the direction of the
strain, the distance between the spheres increases, while perpendicular to that the
spheres come nearer together. Accordingly, also the BG should be shifted to
lower frequencies, which is not the case within the experimental error, or, with
other words, the gap is robust against structural variation. At high strains, it is
even likely that the crystallographic order vanishes, since the spheres rearrange
in a random way.
How can such robustness be explained? As was shown in Sect. 5.1.3, a
hybridization gap (HG) originates from the individual scatterers properties. But
the present band gap is for sure not a HG, since the position of the HG would
scale with dSiO1

2
and not as observed with d-1 CSP. Furthermore, the lowest
eigenmodes of silica embedded in the PEA rubber should be not so different
from those found in the liquid SR256 matrix in Sect. 4.2.4 in Chap. 4, which
appear at clearly higher frequencies than the band gap does in the films.
Additional to that, the evidence of the gap being a BG is striking. Not only the
scaling with d-1CSP, but also the correct prediction of the gap position by theo-
retical consideration in the unstretched case, are clear proofs. Moreover, there is
no reason why there should be no BG.
So, the open question remains why the gap does not react on the stretching. So
far, there can be only speculation on this matter. A possible explanation might be
that BLS is probing a kind of ‘effective distance’ between spheres, which is a
somehow averaged value. Therefore, one must assume that the probed phonons are
not only on a well defined path like in Fig. 5.17b, which might be explained by
strong scattering by the silica spheres. Even at higher strain, the average distance
between each two neighboring spheres does not change to much (in fact, it is a
function of Poisson’s ratio). On the other hand, this explanation seems not to be
very satisfactory. If there would be very strong scattering, the phonon lifetime
would decrease and the linewidth would increase drastically. Furthermore, an
analogue band gap would have to appear also in the dense silica suspensions of
Sect. 5.4.1, which is not the case. Therefore, instead of presuming to absurd
theories, I would like to let the explanation of this effect as an open question.
Additional experiments to determine the crystallographic behavior at high strains
(e.g., AFM, TEM) and further theoretical support will be needed to solve this
demanding problem in the future.
In summary, both investigated systems with hard silica spheres in soft liquid
or elastic matrix show unprecedented phononic effects, which will challenge
experimentalists and theoreticians. A dense packing of silica spheres in an index
matching liquid shows an unseen dispersion relation with an intermediate flat band
only for certain qs, which is probably a result of the folding of bands. In melt flow
pressed films containing about 30 vol% of silica cores in a crystalline order, a band
gap is presented that can be assigned as a BG. However, stretching of the film
results in an unexpected robustness of the gap position, which cannot be explained
satisfactory so far.
5.5 Materials
The synthesis of the polymer and silica colloids is already described in Sect. 4.4 in
Chap. 4. Opals and disordered opals (hybrids) are obtained by vertical lifting
deposition described in Sect. 3.1 in Chap. 3.
For the infiltration of opals shown in Fig. 5.3 silicon oil and PDMS made in
house or purchased from Sigma Aldrich were used with no further purification.
2(2-Ethoxyethoxy) ethyl acrylate (SR256) used in the suspensions was purchased
from Sartomer with no further purification.
5.5 Materials 121
5.5.1 Melt Compressed Silica–Poly(ethyl acrylate) Films
The preparation of the compression molded films is explained in Sect. 3.2 in Chap. 3.
The synthesis of the initial core–shell particles was performed by my coworker
Diana Kiefer at DKI, Darmstadt and is briefly described here [22, 67, 70].
The synthesis starts with the preparation of the silica core following the Stöber
process and subsequent step-by-step growing as described in Sect. 4.4 in Chap. 4.
The silica core’s surface was functionalized by acryl silanes on which a thin
(*5 nm) PMMA interlayer was crafted by emulsion polymerisation of methy
methacrylate (MMA) together with some allyl methacrylate (ALMA) monomer,
using sodium dodecylsulfate (SDS) as an emulsifier and ammonium peroxodisulfate
and sodium dithionite as a redox initiator system. Finally the PEA shell was crafted
onto the allylic double bonds of the ALMA, utilizing again SDS as emulsifier.
References
1. Yablonovitch E (1987) Phys Rev Lett 58:2059

2. John S (1987) Phys Rev Lett 58:2486
flow of light, 2nd edn. Princeton University Press, Princeton
4. Martinez-Salazar R et al (1995) Nature 378:241
5. Montero de Espinosa FR, Jiménez E, Torres M (1998) Phys Rev Lett 80:1208
6. Vasseur JO et al (2001) Phys Rev Lett 86:3012
9. Sigalas M, Economou EN (1993) Solid State Commun 86:141
10. Kushwaha MS, Halevi P, Dobrzynski L, Djafari-Rouhani B (1993) Phys Rev Lett 71:2022
11. Gorishnyy T, Maldovan M, Ullal C, Thomas E (2005) Phys World 18:24
12. Sheng P, Chan CTZ (2005) Z Kristallogr 220:757
13. Sanchez-Perez JV et al (1998) Phys Rev Lett 80:5325
14. Dobrzynski L et al (1999) EPL (Europhy Lett) 46:467
15. Lu M-H et al (2007) Nat Mater 6:744
16. Lin K-H et al (2007) Nat Nanotechnol 2:704
18. Gorishnyy T et al (2005) Phys Rev Lett 94
19. Campbell M, Sharp DN, Harrison MT, Denning RG, Turberfield AJ (2000) Nature 404:53
20. Deubel M, Wegener M, Kaso A, John S (2004) Appl Phys Lett 85:1895
21. Cumpston BH et al (1999) Nature 398:51
23. Xia YN, Gates B, Yin YD, Lu Y (2000) Adv Mater 12:693
25. Wang J, Li Q, Knoll W, Jonas UJ (2006) Am Chem Soc 128:15606
27. Economou EN, Sigalas MJ (1994) Acoust Soc Am 95:1734
28. Meschede D (2006) Gerthsen Physik, 23 edn. Springer, Heidelberg
29. Ashcroft NW, Mermin ND (1976) Solid state physics. Thomson Learning, London
30. Wood AB (1941) Textbook of sound. Macmillan, New York
31. Gaunaurd GC, Wertman WJ (1989) Acoust Soc Am 85:541
32. Waterman PC, Truell RJ (1961) J Math Phys

33. Baumgartl J, Zvyagolskaya M, Bechinger C (2007) Phys Rev Lett 99:205503
34. Reinke D et al (2007) Phys Rev Lett 98:038301
36. Gorishnyy T, Jang J-H, Koh C, Thomas EL (2007) Appl Phys Lett 91:121915
37. Harker AH, Temple JAG (1988) J Phys D Appl Phys 21:1576
39. Liu J, Ye L, Weitz DA, Sheng P (1990) Phys Rev Lett 65:2602
40. Jing X, Sheng P, Zhou M (1991) Phys Rev Lett 66:1240
42. Gomopoulos N (personal communication)
43. Retsch M (2009) Complex materials via colloid crystallization. Ph.D. thesis
44. Phan S-E, Russel WB, Zhu J, Chaikin PM (1998) J Chem Phys 108:9789
45. Psarobas IE, Modinos A, Sainidou R, Stefanou N (2002) Phys Rev B 65
46. Goffaux C et al (2002) Phys Rev Lett 88
47. Yang SX et al (2002) Phys Rev Lett 88
48. Zhang XD, Liu ZY (2004) Appl Phys Lett 85:341
49. Yang SX et al (2004) Phys Rev Lett 93
50. Jang JH et al (2006) Adv Mater 18:2123
55. Penciu RS et al (2000) Phys Rev Lett 85:4622
56. Sainidou R, Stefanou N, Psarobas IE, Modinos AZ (2005) Z Kristallogr 220:848
57. Cheng W (2007) Hypersonic elastic excitations in soft mesoscopic structures. Ph.D. thesis
59. Johnson DL, Plona TJ (1982) J Acoust Soc Am 72:556
60. Yerazunis S, Cornell SW, Wintner B (1965) Nature 207:835
61. Stefanou N (personal communication)
62. Anderson PW (1958) Phys Rev 109:1492
63. Lagendijk A, van Tiggelen B, Wiersma DS (2009) Phys Today 24
64. Hu H et al (2008) Nat Phys 4:945
67. Spahn P (2008) Ph.D. thesis, Technical University Darmstadt
68. Ruhl T, Spahn P, Winkler H, Hellmann GP (2004) Macromol Chem Phys 205:1385
69. Pursiainen OLJ et al (2008) Adv Mater 20:1484
70. Ruhl T, Hellmann GP Macromol (2001) Chem Phys 202:3502
Chapter 6
Smaller than Colloids: Characterization
of Stable Organic Glass
6.1 Introduction
A wide range of packing structures are available to glasses, with more efficient
packing leading to higher moduli materials [1]. Aging a glass allows for better
packing and a higher modulus, but even long aging times increase the modulus by
only a few percent [2, 3]; preparing high modulus materials in this manner is
impractical. In this chapter it is shown that physical vapor deposition can be used
to circumvent this kinetic limitation and produce glasses whose moduli exceed
those of the ordinary glass by up to 19%. These high modulus glasses resist
thermal treatment and take at least 104 times longer than the structural relaxation
time to transform to the supercooled liquid. The ability to easily produce high
modulus glasses will prove to be useful for fundamental investigations and coating
technologies.
Unlike their crystalline counterparts, glasses have a nearly limitless array of
packing arrangements. As a supercooled liquid is cooled, molecular motions
eventually slow to such an extent that equilibrium cannot be maintained. Below
this transition temperature Tg, a mechanically stable, non-equilibrium glass is
formed. Glasses slowly evolve towards equilibrium (i.e., aging) in a process that
optimizes packing and creates higher moduli materials. The structural relaxation
time sa dictates the rate at which this process takes place, and due to the steep
temperature dependence near Tg, sa is on the order of days only a few degrees
below the glass transition temperature Tg. The aging process thus changes the
moduli so slowly that in practice changes of only a few percent are possible [2, 3].
If high modulus amorphous materials are to be utilized for science and technology,
new preparation techniques are needed which circumvent these kinetic restrictions
and allow for more optimized amorphous packing [4].

124 6 Smaller than Colloids: Characterization of Stable Organic Glass
6.2 BLS Experiments on IMC
This section shows that high modulus glass materials can be made efficiently with
physical vapor deposition. Using this preparation technique, one can avoid the
kinetic limitations of aging and prepare high modulus glasses in a matter of hours.
Enhanced dynamics at the surface of amorphous materials allows for rapid con-
figurational sampling in the top few nanometers [5–9]. Vapor deposition can build
an efficiently packed amorphous material in a layer-by-layer fashion by taking
advantage of the enhanced surface dynamics and thus is not limited by the slow
relaxation dynamics of the bulk [10]. The mechanical properties of vapor-depos-
ited films are determined using Brillouin light scattering (BLS) spectroscopy.
Because of the non-destructive nature of BLS, the moduli of the as-deposited
glass, the supercooled liquid, and ordinary glass (created by cooling the liquid) can
be determined from a single sample.
Physical vapor depositions were performed by my coworker Kenneth Kearns
in Prof. Mark Ediger’s group at University of Wisconsin, Madison, separately on
two organic glass-forming materials: indomethacin (IMC, Tg = 315 K) and tris-
naphthylbenzene (TNB, Tg = 348 K). These two molecules are well-known glass-
formers and their glasses have been previously prepared with physical vapor
deposition [10–12]. During deposition, the temperature of the substrate Tsubstrate
and the deposition rate are the important control parameters. For this work,
Tsubstrate was held near 0.85 Tg, i.e., 265 K for IMC and room temperature
(&295 K) for TNB. The rate of the deposition in all cases was 0.2 ± 0.03 nm/s. It
has previously been shown that these deposition conditions produce glasses with
low enthalpy, high density, and high kinetic stability [10–14]. Tsubstrate was con-
trolled by attaching the SiO2 substrates to a copper temperature stage (Fig. 6.1a).
The deposition rate was controlled by adjusting the temperature of the crucible.
Further details are given in Methods (Chap. 3).
In the BLS experiment, the incident laser polarization was chosen to be per-
pendicular to the scattering plane. Scattered light polarized perpendicular and
parallel to the incident polarization was measured in separate experiments, pro-
viding access to scattering from longitudinal and transverse phonons, respectively.
The scattering from these two polarizations is shown in Fig. 6.1d for vapor-
deposited IMC glass at 298 K (lower panel) and supercooled liquid IMC at 336 K
(upper panel). Stokes and anti-Stokes shifts by the longitudinal (L) and transverse
(T) phonons are observed to the left and right of the Rayleigh line region (shaded
area), respectively. The vertical lines drawn on the Stokes side of the spectrum
illustrate the temperature-independent scattering of the SiO2 substrate and the
temperature-dependent scattering of the IMC. The absence of longitudinal pho-
nons in the transverse spectrum indicates no birefringence in the vapor-deposited
film within the sensitivity.
The peaks in the BLS spectra provide access to the phase velocities cl,t for the L
and T polarizations and through this route the moduli can be obtained. Spectra
similar to those found in Fig. 6.1d were obtained at multiple temperatures for both
6.2 BLS Experiments on IMC 125
(a) (d)
(b) (c)
Fig. 6.1 a Schematic representation of the physical vapor deposition chamber with temperature
controlled substrate. b, c Chemical structures of TNB and IMC, respectively. d Representative
BLS spectra for a stable vapor-deposited IMC glass (lower panel) and supercooled liquid (upper
panel) at the wave vector q = 0.0136 nm-1. (Inset transmission scattering geometry with scat-
tering angle h of 70°.) The Rayleigh line region (shaded area) was removed for clarity. Both the
longitudinal (L, black) and transverse (T, red) spectra are shown. Longitudinal and transverse
phonon scattering for the IMC glass (LIMC and TIMC, respectively) and amorphous SiO2 substrate
(Lsubstrate and Tsubstrate, respectively) are indicated in the lower panel. The vertical lines drawn
near the Stokes scattering peaks illustrate the difference between the supercooled liquid and
as-deposited glass [15]
IMC and TNB. Each peak was fitted with a Lorentzian lineshape yielding the
linewidth CL,T and the peak frequency fl,t. The corresponding phase velocities cl,t
for the L and T polarizations were calculated using cl,t = 2pfl,t/q, and are plotted in
Fig. 6.2. Four experiments are shown in Fig. 6.2; in each case, the as-deposited
stable glass was heated to a temperature above Tg, held isothermally until equi-
librium was attained, then further heated, and finally cooled. During the initial
heating of the stable glass (SG), cl and ct have high values as compared to the
ordinary glass (OG), indicating high moduli. During isothermal annealing above
Tg (arrow 2), significant decreases in the phase velocities are observed. This
change, discussed in detail below, signifies the transformation of the stable glass
into the supercooled liquid (SCL). Once the transformation is complete, the
sample, now a supercooled liquid, is heated still higher and then cooled to 295 K
(arrows 3 and 4). A characteristic kink in the temperature-dependent phase
velocity indicates Tg and the value obtained is in reasonable agreement with the
value obtained from differential scanning calorimetry [11, 12].
From the phase velocities cl,t in Fig. 6.2, the moduli of the stable glass and the
ordinary glass can be calculated. The longitudinal bulk modulus M ¼ .c2l ; the
shear modulus G ¼ .c2t and the commonly used Young’s modulus E = 9MG/
(3M ? G) can all be determined from c. For the stable as-deposited glass of IMC,
M, G, and E moduli were determined to be 8.3, 1.7, and 4.9 GPa at Tg,
(a) (b)
Fig. 6.2 a Temperature dependence of phase velocity c for IMC. The longitudinal (squares) and
transverse (circles) c are shown for the stable as-deposited glass (SG), the supercooled liquid
(SCL), and the ordinary glass (OG). Arrows indicate the progression of the heating and cooling
cycles. Arrow 2 indicates the isothermal transformation of the SG to the SCL at Tg + 10 K. b
Temperature dependent longitudinal and transverse c of TNB for the SG, OG and SCL. The
thermal cycle is similar to the one described for a [15]
respectively. The moduli calculated for the ordinary glass of IMC are smaller than
those obtained for the stable glass samples. At 312 K, M for the stable as-deposited
IMC glass is 14% greater than the ordinary glass while G and E are 19% greater.
A similar situation holds for TNB where M, G, and E are 10, 15, and 14% greater,
respectively, for the stable glass samples as compared to the ordinary glass. To
obtain these values, the published densities of 1.31 g/cm3 for IMC [16] and 1.16 g/
cm3 for TNB [17] were used. The stable as-deposited glass was assumed to be
1.5% more dense than the ordinary glass in accordance with previous measure-
ments on TNB vapor-deposited under similar conditions [10, 14]. Notably, the
Poisson ratio r has the same value (0.36 ± 0.01) in both stable and ordinary glass
samples and increases for the supercooled liquid as expected [18].
In contrast to the results presented here, aging experiments on glass-forming
systems show only small improvements in mechanical properties during aging.
BLS experiments were performed on glycerol by quenching from above Tg to
Tg -4.2 K; in this experiment, a 0.6% increase in modulus was achieved over
20 days [3]. For silicone oil, temperature down-jumps of 1.5 K near Tg resulted in
a 2.7% increase in the modulus during the 104 s equilibration time [2]. Based on
calorimetry measurements [10, 12], it was previously estimated that IMC and TNB
samples vapor-deposited under the conditions utilized here have properties similar
to those expected after aging an ordinary glass for more than 300 years. Thus it is
not surprising that aging for hours or days does not produce the large modulus
changes reported here. Pressurization experiments have been shown to change
moduli by up to 20%, but this effect is reversible and upon pressure release, the
system reverts back to the ordinary glass value rendering it of little use for
potential applications [19].
6.2 BLS Experiments on IMC 127
Notably, the high modulus IMC and TNB glasses produced by vapor-deposition
exhibit remarkable thermal stability. Figure 6.3a and b shows the change in cl
and ct for vapor-deposited IMC during isothermal annealing above Tg. In these
experiments, the temperature of the stable as-deposited glass was increased to
325 K in about 15 min and then held for several hours. Both phase velocities (and
moduli) remain relatively unchanged for the first 120 min of the experiment. After
this induction time, the phase velocities begin to decrease from the high values of
the stable glass to the supercooled liquid values. Approximately 200 min is needed
to complete the transformation to the supercooled liquid. To put this value into
context, the structural relaxation time sa of the IMC supercooled liquid at 325 K
is about 1 s based on dielectric spectroscopy studies [20]; thus the isothermal
transformation of stable vapor-deposited IMC into the supercooled liquid requires
about 104sa. Similar results are obtained for TNB where the stable as-deposited
glasses required 104.8sa to transform into the supercooled liquid at 358.2 K [21].
Conceptually similar experiments have been performed on ordinary glasses that
have stabilized by aging. For example, Kovacs aged poly(vinyl acetate) for
2 months below Tg and then performed dilatometry during isothermal annealing
above Tg. He observed the transformation into the supercooled liquid in a period of
(a) (c)
(b)
Fig. 6.3 BLS experiments on vapor-deposited IMC glasses during isothermal annealing above
Tg (325 K). a Longitudinal, cl, and b transverse, ct, phase velocity changes as a function of time.
Sigmoidal fits are given as guides to the eye. Two different samples (squares and circles) are
shown in a to demonstrate the reproducibility of the transformation. In a and b, the full width at
half height linewidth, C, is shown as crosses. c Spectral line shapes and Lorentzian fits (solid
lines) at three different times during isothermal annealing [15]
35sa [22]. Thus, in comparison to glasses prepared by cooling a liquid and then
aging, the vapor-deposited IMC and TNB samples maintain their extraordinarily
high modulus for a remarkably long time.
Changes observed in the BLS spectral linewidth during isothermal annealing
provide insight into the mechanism by which the stable glass transforms into the
supercooled liquid. Figure 6.3a shows that a maximum in C is reached near the
midpoint of the transformation (C is read from the right axis). A similar feature is
observed in Fig. 6.3b although the smaller signal of the transverse scattering
makes this less apparent. The observation of a change in C during the transfor-
mation of a glass to a liquid is unprecedented. The increase in C indicates that the
sample is heterogeneous on the 500 nm (*2p/q) length scale during the trans-
formation. Figure 6.3 shows the BLS spectral lineshapes at three times during
isothermal annealing at 325 K. Before the transition begins (30 min) and after the
transition ends (270 min), narrow lines are observed. In contrast, near the midpoint
of the transformation (180 min), a broad line is observed. The shape of the line
suggests that the sample is a mixture of the supercooled liquid and the stable glass
at this stage. This point is illustrated by the dotted red curve in Fig. 6.3c which is a
linear combination of narrow lineshapes from the 30 min (stable glass) and
270 min (supercooled liquid) data. This conclusion is consistent with a recent
quasi-isothermal differential scanning calorimetry study of these materials [13].
While that work suggested that stable glasses transform to the supercooled liquid
via two-phase intermediate state, the spatial extent of these regions (at least
500 nm) has not been known prior to this BLS study.
Significantly increasing the modulus of amorphous solid materials using con-
ventional aging methods is a time consuming and ineffective process. The work
presented in this chapter shows that physical vapor deposition can be used to
prepare glasses with moduli that are as much as 19% greater than those made by
cooling the liquid. These glasses resist thermal treatment and retain their high
modulus values for at least 104 times longer than the structural relaxation time of
the glass at temperatures significantly above Tg. During this slow transformation, a
mixture of supercooled liquid and glass is observed on length scales greater than
500 nm. High modulus vapor-deposited glasses may prove useful for applications
because of their remarkable mechanical properties but also their extraordinary
thermal stability. Given the known correlation between amorphous packing and
modulus, new insights into this packing could be realized through fundamental
studies of these materials.
6.3 Materials and Methods
6.3.1 Stable Glass
IMC and TNB stable glass films were prepared by my coworker Kenneth Kearns at
University of Wisconsin in Madison, WI.
6.3 Materials and Methods 129
IMC was purchased from Sigma Aldrich and was used without further purifi-
cation. TNB was synthesized by McMahon and co-workers using published
methods [23]. SiO2 substrates (5 mm diameter and 0.15 mm thick, Fisher Scien-
tific coverglass) were attached to a copper temperature stage using double-sided
conductive carbon black tape (SPI Supplies). The temperature of the substrate
during deposition was controlled using a Lakeshore 340 temperature controller. To
deposit the IMC and TNB glass films, a quartz crucible containing the crystalline
material was heated inside a vacuum chamber (*10-8 torr) such that the desired
rate of 0.2 nm/s was achieved. The rate of deposition was controlled by main-
taining the temperature of the quartz crucible, and the rate was monitored with
a quartz crystal microbalance (Sycon instruments). Deposition continued until a
10–15 lm film was deposited. After deposition, the substrates were removed from
the deposition chamber, placed in desiccant and stored in dry ice prior to analysis.
6.3.2 BLS Measurements
BLS was performed in a transmission geometry to obtain the phase velocity and
thus modulus values. The transmission, as opposed to the commonly used
reflective geometry, was used to remove the influence of refractive index on the
measurements, and the details of this geometry are described in detail elsewhere
[24–26]. For all experiments, the sample holder was initially flushed with argon
gas and desiccant was placed in the holder to eliminate any possible plasticization
by atmospheric water [27, 28]. The sample temperature was monitored with a
platinum RTD and controlled to within 0.5 K with a temperature controller which
was built in-house. For each data point in Fig. 6.2, the temperature was allowed to
stabilize for 15 min and then the BLS spectra were acquired for 5 min. The
heating/cooling rate between each temperature measurement was approximately
2–3 K/min. The temperature history establishes an effective heating/cooling rate
of approximately 0.1 K/min. For isothermal experiments, the temperature was
changed from room temperature to the annealing temperature in 15–20 min.
References
1. Dyre JC, Olsen NB, Christensen T (1996) Phys Rev B 53:2171

2. Olsen NB, Dyre JC, Christensen T (1998) Phys Rev Lett 81:1031
3. Miller RS, MacPhail RA (1997) J Chem Phys 106:3393
4. Royall PC, Williams SR, Ohtsuka T, Tanaka H (2008) Nat Mater 7:556
5. Forrest JA, Dalnoki-Veress K, Stevens JR, Dutcher JR (1996) Phys Rev Lett 77:2002
6. Keddie JL, Jones RAL, Cory RA (1994) Faraday Discuss 98:219
7. Fakhraai Z, Forrest JA (2008) Science 319:600
8. Bell RC et al (2003) J Am Chem Soc 125:5176
9. Ellison CJ, Torkelson JM (2003) Nat Mater 2:695
10. Swallen SF et al (2007) Science 315:353

11. Kearns KL et al (2007) J Chem Phys 127:154702
12. Kearns KL et al (2008) J Phys Chem B 112:4934
13. Kearns KL, Still T et al (2010) Adv Mater 22:39
14. Kearns KL et al (2009) J Phys Chem B 113:1579
15. Swallen SF et al (2008) J Chem Phys 128:214514
16. Yoshioka M, Hancock BC, Zografi G (1994) J Pharm Sci 83:1700
17. Plazek DJ, Magill JH (1966) J Chem Phys 45:3038
18. Novikov VN, Sokolov AP (2004) Nature 431:961
19. Bezot P, Hesse-Bezot C (1990) J Non-Cryst Solids 122:160
20. Carpentier L, Decressain R, Desprez S (2006) J Phys Chem B 110:457
21. Richert R, Duvvuri K, Duong L-T (2003) J Chem Phys 118:1828
22. Kovacs AJ (1963) Fortschr HochpolymForsch 3:394
23. Bonvallet PA et al (2007) J Org Chem 72:10051
27. Andronis V et al (1997) J Pharm Sci 86:346
28. Dawson KJ et al (2009) J Phys Chem B 113:2422
Chapter 7
Concluding Remarks
7.1 Conclusions
Colloidal self-assembly by vertical lifting is a potential and versatile method to

create and design artificial sub-micron periodic structures of hard (e.g., SiO2) and
soft (polymers) colloidal particles. The fabrication of high quality crystals as well
as the creation of artificial glasses by the introduction of structural disorder is
possible. Such structures have spatial dimensions of their components in the order
of the wavelength of visible light, hence they act as artificial opals. In addition,
they show interesting phononic properties in the hypersonic (GHz) range.
Brillouin light scattering (BLS) is found to be a powerful technique to measure
the elastic properties of nanostructured materials at hypersonic frequencies. In case
of dry, non-transparent, multiple scattering colloidal samples, BLS measures
q-independent localized modes of the individual spheres. In particular, for soft
spheres these spectra are found to be richer with increasing diameter. This finding is
explained by theoretical and geometrical considerations. Nearly all predicted modes
(especially those that appear at frequencies significantly below that of the acoustic
phonon in the back-scattering case) are found experimentally and assigned to their
corresponding spheroidal eigenmodes. The analysis of the eigenmode spectra yields
the longitudinal cl and transverse sound velocity ct, allowing the calculation of the
Young and shear elastic moduli at the nanoscale. Additionally, the analysis of the
line shape can provide information on the size polydispersity and on interactions
with surrounding medium [1, 2].
Based on the studies of several binary colloidal mixtures, the eigenfrequencies do
not depend on the crystalline order or on the kind and number of the neighboring
spheres (Sect. 4.2.2 in Chap. 4). Only in the extreme case of dilute large spheres in a
sea of small spheres a small blue shift was observed [1]. On the open discussion with
regard the eigenmode BLS activity [3, 4], density of states and finite element
calculations have shown that all eigenmodes (n, l) can contribute to the light
scattering spectrum of spheres with diameter in the order of magnitude of the
probing laser wavelength. The limitation on even ‘quantum numbers’ l, as claimed

132 7 Concluding Remarks
by others, is disputed (Sect. 4.2.1 in Chap. 4). Further, the utilization of densely
packed silica spheres in an index matching liquid led to the first verification of the
theoretical predictions for the q-dependent amplitude of the eigenmode spectrum;
the frequency of these localized modes is expectedly q-independent.
After these fundamental questions concerning the spectrum of the resolved
modes in homogeneous colloidal spheres have been convincingly addressed, the
effect of the elastic constants on the particle vibrational modes was best addressed
in the case of random copolymer (PMMA-PnBA) spherical particles, where
the elastic constants vary with the composition (Sect. 4.2.3 in Chap. 4). The
randomness of the copolymer and the amorphous state of the particle was verified
by the presence of single Tg by DSC, further confirmed by the single a-relaxation
obtained by dielectric spectroscopy, i.e., both kind of segments feel the same energy
landscape. The systematic red shift of the eigenfrequencies with increasing
composition of the softer (PnBA) component as well as the effective medium
velocities measured in bulk copolymer films were theoretically captured when a
higher ‘effective glassy sound velocity’ was assumed for the PnBA in the
copolymer than in its bulk state.
Two additional elaborate BLS experiments on nanostructured core–shell hybrid
systems were presented herein: (1) hard silica colloids surrounded by a soft
polymer shell show eigenmode spectra that are dominated by the scattering from
the soft polymer shell (larger compressibility). Nevertheless, due to the boundary
conditions and displacements in the composite material, it becomes feasible the
extraction of the elastic constants in parts, core and shell at the nano- and
mesoscale. The spatial confinement causes a hardening of the thin polymer
chain (Sect. 4.2.2 in Chap. 4) [5]; (2) in the inverse particle architecture, where the
polymeric core is embraced by a thin silica shell, the former dominates the
eigenmode spectrum. A temperature dependent study has revealed a striking
hardening of the polymeric core as compared to the bare polymer spheres implying
a modified glassy state under confinement. The glass transition temperature Tg of
the core, however, remained virtually constant whereas the thin silica shell assured
a robust shape persistent polymeric core at temperatures well above its Tg (‘nano
armor’, Sect. 4.3.1 in Chap. 4) [6].
Since the hypersonic phonons are in the order of magnitude of few hundreds
nanometers, they do not only probe the effective medium structured in a molecular
scale (like in the copolymers), but also in the mesoscale. By infiltration of the dry
samples with a liquid close to optical matching, it is possible to overcome the
multiple light scattering and, by that, to get access to the collective acoustic behavior
of the colloidal systems. Pure effective medium behavior is observed by BLS for
such systems in the long wavelength limit, i.e., for small q’s, where the phonon
wavelength exceeds the structure length of the colloidal system. Section 5.1.2 in
Chap. 5 presents a short overview over the most important established effective
medium theories, while in Sect. 5.2 in Chap. 5 their applicability on a system of
defect doped liquid infiltrated opals is presented.
At higher q’s, BLS can be used to record directly the dispersion relations that
led to the first demonstration of a distinct hypersonic Bragg gap at the edge of the
7.1 Conclusions 133
first Brillouin zone of a colloidal crystal by Cheng et al. [7]. In this thesis, the
resonant character of the particles was realized to demonstrate, for the first time,
the presence of an additional band gap—the hybridization gap, theoretically
predicted six years earlier. This HG originates from the interaction of the acoustic
band of the effective medium and bands from the multipole modes of
the individual particles (Sect. 5.3.1. in Chap. 5), i.e., the eigenmodes. Hence, its
realization demonstrates the strong correlations between the individual colloid’s
elastic properties (the ‘music’) and the phononic characteristics of the ensemble
(the ‘concert’). For polymer colloids infiltrated with an index matching liquid like
PDMS or silicon oil, the HG is found to open up at frequencies below that of
Bragg gap.
Increase of the mechanical mismatch by going from soft to hard opals based on
silica spheres significantly changes the propagation characteristics (Sect. 5.4 in
Chap. 5). This includes an unseen bending in the band diagram of polycrystalline
silica colloids in a liquid matrix as well as a Bragg gap, which is peculiarly robust
against structural deformation, in silica/rubber phoXonic films.
7.2 Outlook
Phononic materials might have interesting potential applications starting with the
obvious use as acoustic shields or vibration isolators. However, for most practical
applications, gap frequencies in the range of sonic or ultrasonic seem to be of
higher interest than those in the hypersonic range. On the other hand, if the
blocked phonon’s wavelength is comparable to the structure dimensions, i.e.,
k * d then the structure’s components for ultrasonic devices must be near to the
millimeter-range and even larger to stop sound between kHz and Hz. The ultimate
goal would be to create materials, which can realize gaps at wavelengths much
larger then the inherent length scale of the gap material (k d). Ping Sheng and
coworkers demonstrated an approach going in this direction [8]. They prepared
locally resonant sonic materials by combining a spherical lead core and a silicone
rubber coating in the millimeter size range and building up a crystal out of these.
The resulted superstructure exhibits a band gap around 400 Hz, which corresponds
to a wavelength two orders of magnitude bigger than the diameter of the spheres.
Like HG the explanation is due to localized modes mostly in the soft shells.
Advances in colloidal science can be utilized to provide materials with new
functions. For example core–shell particles with varying composition, hybrid
materials, or materials with even hierarchical order can allow tailoring the
phononic properties of small but smart materials to frequencies, where a broad
range of applications is conceivable.
When we turn our attention back to the manipulation of the phonon flow in the
GHz range, hypersonic phononic materials hold promise to control heat flow in
insulators and semiconductors. The typical phonon wavelength responsible for the
heat transport is even smaller (&10 nm) than the structures investigated in this
134 7 Concluding Remarks
thesis. If we will control the flow of elastic energy by building phonon guiding
devices in full analogy to the lossless light guidance in photonics or by building
devices with a strong anisotropy of phonon transmission, we can also attain new
knowledge in advanced heat management. In this direction, colloidal science could
offer new materials, e.g., multilayer structures with different particle sizes in each
structure that could split and direct the elastic energy. Other analogies to
photonics, e.g., acoustic superlenses already exist [9], but still impose challenges
for high frequency acoustics. It should be mentioned that hypersonic phononics
can simultaneously act as photonics for wavelengths in the visible spectrum
allowing the realization on photonic and phononic band gaps at the same time
(PhoXonics) [10]. The coupling of optical and mechanical degrees of freedom can
be applied to influence the optomechanical dynamics of small systems, utilizing
the radiation pressure of light [11]. The upper frequency limit for such systems is,
so far, in the MHz-range, however, colloidal systems like those presented in this
thesis may expand the investigation of such effects to the GHz-range, giving raise
to novel fascinating phenomena.
In parallel to the development of materials with new phononic functions, there
is also a need for new experimental techniques to measure the band diagrams of
non-transparent structures exhibiting also strong phononic gaps. BLS is a powerful
technique for transparent samples with serious shortcomings in opaque systems
such as dry opals and in measuring hypersonic transmission spectra. The main
problem is the availability of a selective and continuous generation of phonons in
the GHz-range, which will be hopefully overcome in the near future.
References

4. Li Y et al (2008) Chem Phys Lett 461:111
6. Still T, D’Acunzi M, Vollmer D, Fytas G (2009) J Coll Int Sci 340:42
9. Zhang XD, Liu ZY (2004) Appl Phys Lett 85:341
10. Maldovan M, Thomas EL (2006) Appl Phys Lett 88:251907
11. Kippenberg TJ, Vahala KJ (2008) Science 321:1172
Chapter 8
Appendix: Scattering Geometry
8.1 Transmission Case
Since Brillouin light scattering (BLS) is a quasi-elastic scattering technique,

incident and scattered wave vector can be regarded as of equal length (Fig. 8.1):
2pn
jki j ¼ jks j ¼ : ð8:1Þ
k
With Snell’s law:

1
n sin b ¼ sin a ) b ¼ sin1 sin a ; ð8:2Þ
n

1 1
n sin c ¼ sin h a ) c ¼ sin sin h a : ð8:3Þ
n
The length of the scattering wave vector q is the length of the third side of the
isosceles triangle spanned by ki and ks : Dividing it into two right-angled triangles
allows to apply simple trigonometric rules:
2pn bþc
q ¼ jqj ¼ 2 sin ð8:4Þ
k 2
and with Eqs. 8.2 and 8.3

4pn 1 1 1
q¼ sin sin1 sin ðh aÞ þ sin1 sin a : ð8:5Þ
k 2 n n
Under the distinct condition that the incident angle a is half the scattering angle, q
becomes equal to its component parallel to the surface, and Eq. 8.5 can be further
simplified for this special case:
q ¼ qpara ¼ jqpara j for c ¼ b ) h ¼ 2a ð8:6Þ

136 8 Appendix: Scattering Geometry
n0=1
n
ks
ki
2
q
1 qpara
n
n0=1
Fig. 8.1 Transmission geometry
2pn
qpara ¼ ðsin b þ sin cÞ ð8:7Þ
k
2p 4p h
qpara ¼ ðsin a þ sin ðh aÞÞ ¼ sin : ð8:8Þ
k k 2
In the last equation, valid only in this special case, the refractive index disappears.
8.2 Reflection Case
Starting from the same point as in the transmission geometry,

2pn
jki j ¼ jks j ¼ ; ð8:9Þ
k
applying Snell’s law and the internal angle of a triangle, we can define a couple of
angles shown in Fig. 8.2:

1
b ¼ sin1 sin a ð8:10Þ
n
c ¼ 180 a ð8:11Þ
q ¼ 180 ð90 bÞ ¼ 90 þ b ð8:12Þ
¼ 90 d ð8:13Þ
n sin d ¼ sin 180 h a ð8:14Þ
Now it is possible to express the angle c in the sample between ki and ks as a

function of a and h
8.2 Reflection Case 137
n0=1
n
ks
qperp
q
ki
n
n0=1
Fig. 8.2 Reflection geometry

11 1 1
c ¼ 180 90 sin sin a þ 90 sin sin ð180 h aÞ
n n

1 1
¼ 180 sin1 sin a sin1 sin ðh þ aÞ ;
n n
ð8:15Þ
and q turns out to become:

4pn 1 1 1
q¼ sin 180 sin1 sin a sin1 sin ðh þ aÞ
k 2 n n
ð8:16Þ
4pn 1 1 1 1 1
¼ cos sin sin a þ sin sin ðh þ aÞ
k 2 n n
The angle g between q and ki or ks is
180 c
g¼
2 ð8:17Þ
1 1 1
¼ sin1 sin a þ sin1 sin ð180 h aÞ :
2 n n
Using this equation, the relation between q and its component perpendicular to the
surface, qperp, can be expressed again as a function of the two angles h and a,
qperp
¼ cos ðb gÞ
q

1 1 1 1 1 1 1
¼ cos sin sin a sin sin a þ sin sin ðh þ aÞ
n 2 n n

1 1 1
¼ cos sin1 sin a sin1 sin ðh þ aÞ
2 n n
ð8:18Þ
138 8 Appendix: Scattering Geometry
and qperp becomes
qperp ¼ q cos ðb gÞ

4pn 1 1 1 1 1
¼ sin 180 sin sin a sin sin ðh þ aÞ
k 2 n n ð8:19Þ

1 1 1 1 1
cos sin sin a sin sin ðh þ aÞ :
2 n n
By comparison with Eq. 8.16 it is clear that q ¼ qperp if the last term on the right
side becomes one. Due to the periodicities of the inverse sine function, this is the
case for
h 180
a¼ : ð8:20Þ
2
i.e., when chosing this special geometry in the BLS experiment, the wave vector
lies fully perpendicular to the substrates surface. However, the careful comparison
with Fig. 2.6 in Chap. 2 shows that the fulfillment of this geometry would lead to a
reflection of the laser directly into the Fabry–Pérot. Therefore, only geometries
very near to that can be chosen in practice, in order not to destroy the detector, and
the qs probed in reflection geometry are only approximately equal to the qperp s.
Curriculum Vitae
Tim Still
Date of Birth: April 3, 1982
Place of Birth: Neuwied, Germany
Nationality: German
ResearcherID: A-3286-2009
Since November 2009

Max Planck Institute for Polymer Research, Mainz, Germany
Postdoctoral Researcher
March 2007–November 2009
Max Planck Institute for Polymer Research, Mainz, Germany
Promotion by Johannes Gutenberg University Mainz
Degree obtained: Doctor in Natural Science (Dr. rer. nat.)
April 2002–January 2007
Institute for Physical Chemistry, Department of Chemistry, Philipps University
Marburg, Germany
Degree obtained: Diploma in Chemistry (Dipl. Chem.)
Thesis: Inelastische Neutronenstreuung an geschmolzenen Natriumhaliden
Publications
2010
(8) T. Still, D. Kiefer, G. Fytas, N. Stefanou,... Collective Hypersonic Excitations

in Strongly Multiple Scattering Colloids, under review.
139
140 Curriculum Vitae
(7) F. Stadler, T. Still, G. Fytas, C. Bailly, Elongational Rheology and Brillouin

Light Scattering of Entangled Telechelic Polybutadiene Based Temporary
Networks, under review.
(6) T. Still, M. Retsch, U. Jonas, R. Sainidou, P. Rembert, K. Mpoukouvalas,
G. Fytas, Vibrational Eigenfrequencies and Mechanical Properties of Mesoscopic
Copolymer Latex Particles. Macromolecules 2010, 43, 3422.
(5) K. Kearns, T. Still (equal contribution), G. Fytas, M. D. Ediger, High-Modulus
Organic Glasses Prepared by Physical Vapor Deposition. Adv. Mater. 2010, 22, 39.
2009
(4) T. Still*, M. D’Acunzi, D. Vollmer, G. Fytas, Mesospheres in Nano-Armor:

Probing the Shape-Persistence of Molten Polymer Colloids. J. Coll. Interf. Sci.
2009, 340, 42.
2008
(3) T. Still, R. Sainidou, M. Retsch, U. Jonas, P. Spahn, G. P. Hellmann, G. Fytas,

The Music of Core-Shell Spheres and Hollow Capsules: Influence of the Archi-
tecture on the Mechanical Properties at the Nanoscale. Nano Lett. 2008, 8, 3194
(with ‘Top down, Bottom up’ in Nature Nanotech. 2008, 3, 614).
(2) T. Still, W. Cheng, M. Retsch, U. Jonas, G. Fytas, Colloidal Systems: A
Promising Material Class for Tailoring Sound Propagation at High Frequencies.
J. Phys. Condens. Matt. 2008, 20, 404203.
(1) T. Still, W. Cheng, M. Retsch, R. Sainidou, J. Wang, U. Jonas, N. Stefanou,
G. Fytas, Simultaneous Occurrence of Structure-Directed and Particle-Resonance-
Induced Phononic Gaps in Colloidal Films. Phys. Rev. Lett. 2008, 100, 194301.
Presentations in International Conferences
2009
ICPPP 15, Leuven, Belgium, 07/20–23/2009:

T. Still, A. Sato, G. Fytas (oral presentation): Tailoring Phonon Propagation by
Mesoscopic Engineering of Soft Matter
APS March Meeting, Pittsburgh, PA, USA, 03/16–20/2009:
T. Still, R. Sainidou, G. Hellmann, G. Fytas (oral presentation): The ‘Music’ of
Silica-Poly(methyl methacrylate) Core-Shell Spheres: Eigenvibrations and
Mechanical Properties at the Nanoscale
Curriculum Vitae 141
K. L. Kearns, T. Still, G. Fytas, M. D. Ediger (poster): Brillouin Light Scattering

Measurements of High Modulus Glasses Produced by Physical Vapor Deposition
Makromolekulares Kolloquium 2009, Freiburg, Germany, 02/26–28/2009:
M. Retsch, T. Still, W. Cheng, R. Sainidou, U. Jonas, N. Stefanou, G. Fytas
(poster): Phononic Properties of Colloidal Particles and Nanostructured Films
2008
16th Ostwald Colloquium, Leipzig, Germany, 09/22–24/2008:

T. Still, M. Retsch, R. Sainidou, U. Jonas, G. Fytas (oral presentation): High
Frequency Acoustics in Colloidal Systems by Spontaneous Brillouin Light
Scattering
Son et lumière 2008, Cargèse, Corse, France, 01/01–13/2008:
T. Still, M. Retsch, R. Sainidou, W. Cheng, U. Jonas, G. Fytas (oral presentation +
poster): GHz Acoustics in Colloidal Systems by Spontaneous Brillouin Light
Scattering
Acoustics’08, Paris, France, 06/29–07/04/2008:
T. Still, M. Retsch, R. Sainidou, U. Jonas, G. Fytas (oral presentation): High
Frequency Acoustics in Colloidal Systems by Spontaneous Brillouin Light
Scattering (J. Acoust. Soc. Am. 2008, 123, 3555)
Index
A E
Anderson localization, 116 Effective medium, 7, 66, 69, 91, 94, 98, 100ff,
Anisotropy, 134 114ff, 132
Eigenmode, 15, 29ff, 47, 53ff, 96, 98, 107,
116, 120, 131ff
B Electron microscopy, 38f, 56f, 61ff, 72ff, 101f
Band diagram, 7, 43, 45ff, 89, 91ff, 108ff,
133f
Band gap, 1ff, 36, 45, 54, 89ff, 104ff, 133 F
Bragg gap, 1ff, 42, 45, 89, 97ff, 104, 116, 119, Fabry–Pérot, 6, 20f, 27ff, 138
132 Finesse, 26ff
Bulk modulus, 13ff, 125 Finite element modelling, 44, 51, 54ff, 63, 76,
131
Free spectral range, 26ff, 56, 70
C
Calculation methods, 43ff G
Centrifugation, 5, 36, 69f, 84, 86 Glass/glass transition, 4, 6, 7, 25, 30, 35, 37,
Colloidal mixture, 42, 111ff, 131 57, 63ff, 69ff, 105, 109, 123ff, 131ff
Compressibility, 23f, 68, 72
Core–shell particle, 5, 37, 48, 55f, 63, 71ff,
77ff, 85f, 117ff, 121, 132f H
Correlation, 38f, 78, 128, 133 Hybrid, 5, 32, 37, 55, 61ff, 72, 77, 81, 84, 91,
94, 98, 105ff, 111ff, 132f
Hybridization gap, 7, 54, 91, 94, 98, 105ff,
D 111, 113, 119, 133
Dense packing, 37, 62, 67, 96, 114
Density, 2, 3, 14, 21, 23, 33, 38, 40f, 53ff, 64f,
71, 74, 77, 80ff, 89, 94, 101ff, I
113ff, 124, 131 IMC, see IMC
Density of states, 54, 80, 83, 109, 131 Indomethacin, 124
Differential thermo analysis, 40
Differential scanning calorimetry, 40, 64, 73f,
125ff L
Disorder, 61, 63, 104ff, 131 Lamé coefficients, 11, 13, 30, 46, 89
Dispersion relation, 6, 20f, 29, 32, 43, 51, 61ff, Laser, 5f, 20f, 25, 29, 30, 32, 38, 53, 57, 58,
70, 91ff, 105ff, 132 90, 105
143
144 Index
Layer multiple scattering, 44, 50, 116 Resonance vibration, 20, 45, see eigenmode
Linewidth, 26, 73, 81, 114f, 120 Rigidity, 33, 55, 63ff, 72, 73f, 77
Rubber, 37, 67ff, 73ff, 117, 120, 133
M
Melt compression, 37, 117f S
Molecular weight, 43, 64, 84 Scattering angle, 4, 18ff, 23, 29, 32, 38, 41, 65,
Multiple scattering, 7, 32, 55, 57, 58, 65, 69, 69ff, 105, 125, 135
79, 96, 116, 117, 131 Scattering cross section, 43, 44ff, 56, 58, 60,
Music, see eigenmodes 63, 67
Scattering geometry, 30ff, 39, 56, 65, 99f, 105,
111, 125, 129, 135ff
N Setup (BLS), 25ff, 29
Nanostructure, 5, 21, 53, 55, 71ff, 94, 110, Shear modulus, 11, 13, 15, 57, 66, 68, 73f, 80f,
123, 131f 125
Silica, 4, 6, 7, 20, 37, 55f, 60, 62, 66ff, 77ff,
85f, 100ff, 113ff, 132f
O Spherical waves, 9, 15, 33, 104
Opal, 4, 20, 35, 37, 54ff, 94ff, 133f Strain, 9ff, 119f
Order, 104ff, see disorder Stress, 9, 11ff, 32f, 54
Organic glass, 123ff Supercooled liquid, 123ff
Suspension, 5, 55, 62, 69ff, 78, 98, 105,
114ff
P Synthesis, 4, 63, 73, 77, 84ff, 120f, 128f
Phononic crystal, 2ff, 46, 49, 51ff, 90, 97, 98,
104
Photonic crystal, 1ff, 7, 38, 89 T
PhoXonics, 4, 6, 117, 133f T-Matrix, 48f, 94f
Plane wave, 17f, 34, 43ff, 81f, 91 TNB, see TNB
Poisson’s ratio, 13, 14, 57, 65, 68, 69, 80, 120, tris-Naphtylbenzene, 164ff
126
Poly methyl methacrylate, 55, 63ff, 72, 77ff,
97f, 105ff, 121, 132 U
Poly n-butyl acrylate, 63ff UV/Vis spectroscopy, 40ff, 101f
Polydispersity, 61ff, 82, 85, 109, 131
Polystyrene, 4, 31, 37, 45, 55f, 67ff, 72, 79, 84,
95ff, 105ff V
Pump-probe, 79 Vertical lifting deposition, 35ff, 65, 78, 97, 99,
101, 105, 111, 120, 131
R
Raman scattering, 5, 16, 25, 32, 53ff W
Reciprocal space, 1, 41, 46, 93, 97ff, 107 Wide angle X-ray scattering, 41, 64
Reflectivity, 5, 25ff, 53
Refractive index, 1ff, 18, 21, 25, 31, 42, 57,
69, 96, 102, 107, 129, 136 Y
Relaxation, 64, 123ff, 132 Young’s modulus, 30ff, 65, 125ff

High Frequency Acoustics in Colloid-Based Meso - and Nanostructures by Spontaneous Brillouin Light Scattering (Springer Theses) - Tim Still

Uploaded by

Copyright:

Available Formats

High Frequency Acoustics in Colloid-Based Meso - and Nanostructures by Spontaneous Brillouin Light Scattering (Springer Theses) - Tim Still

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

High Frequency Acoustics in Colloid-Based Meso - and Nanostructures by Spontaneous Brillouin Light Scattering (Springer Theses) - Tim Still

Uploaded by

Copyright:

Available Formats

Springer Theses

Recognizing Outstanding Ph.D. Research

For further volumes:

Theses are accepted into the series by invited nomination only

• They must be written in good English.

High Frequency Acoustics

Doctoral Thesis performed at Max Planck Institute

ISSN 2190-5053 e-ISSN 2190-5061

ISBN 978-3-642-13482-1 e-ISBN 978-3-642-13483-8

Springer Heidelberg Dordrecht London New York

Library of Congress Control Number: 2010929574

Springer-Verlag Berlin Heidelberg 2010

Cover design: eStudio Calamar, Berlin/Figueres

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Marie von Ebner-Eschenbach

T. Still, W. Cheng, M. Retsch, R. Sainidou, J. Wang, U. Jonas, N. Stefanou,

T. Still, W. Cheng, M. Retsch, U. Jonas, G. Fytas, J. Phys. Condens. Matt. 2008,

T. Still, R. Sainidou, M. Retsch, U. Jonas, P. Spahn, G. P. Hellmann, G. Fytas,

K. Kearns, T. Still (equal contribution), G. Fytas, M. D. Ediger, Adv. Mater. 2010,

T. Still, M. Retsch, U. Jonas, R. Sainidou, P. Rembert, K. Mpoukouvalas, G. Fytas,

Mainz, March 2010 Prof. Dr. H.-J. Butt

Mainz Tim Still

2 Basics and Brillouin Light Scattering . . . . . . . . . . . . . . . . . . . . . . . 9

3.4.2 The Plane Wave Method . . . . . . . . . . . . . . . . . . . . . . . . 45

4 The Vibrations of Individual Colloids . . . . . . . . . . . . . . . . . . . . . . . 53

5 Phononic Behavior of Colloidal Systems . . . . . . . . . . . . . . . . . . . . . 89

6 Smaller than Colloids: Characterization of Stable Organic Glass. . 123

6.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

8 Appendix: Scattering Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . 135

Curriculum Vitae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

LMS Layer multiple scattering method

1.1 General Introduction

T. Still, High Frequency Acoustics in Colloid-Based Meso- and Nanostructures 1

(organic) polymers. However, colloid science is not generally limited to spherical

the one whose phonon wavevector k is identical with q as a consequence of

1.2 Aims and Motivation

The structure of this thesis is as follows:

2.1 Elastic Waves in Condensed Matter

2.1.1 Elasticity Theory Basics

T. Still, High Frequency Acoustics in Colloid-Based Meso- and Nanostructures 9

Therefore, it is possible to write

dl02 ¼ dl2 þ 2uik dxi dxk : ð2:5Þ

dl02 ¼ ð1 þ 2uð1Þ Þdx21 þ ð1 þ 2uð2Þ Þdx22 þ ð1 þ 2uð3Þ Þdx23 : ð2:7Þ

Fig. 2.1 Illustration of x

By writing the following identity

rik ¼ ciklm ulm : ð2:16Þ

ciklm ¼ ckilm ¼ cikml ¼ ckiml : ð2:17Þ

That is in an anisotropic body there are at most 21 independent components,

Tensor notation 11 22 33 23, 32 13, 31 12, 21

With that notation one can rewrite Eq. 2.16 as

C11 ¼ C12 þ 2C44 ; ð2:20Þ

k ¼ C12 and l ¼ C44 ; ð2:21Þ

2.1.2 Elastic Waves in Isotropic Media

respectively, while the bulk modulus K depends on the longitudinal as well as on

2.1.3 Spherical Waves

ðk þ 2lÞrðr uÞ lr  r  u þ .x2 u ¼ 0: ð2:37Þ

ðr2 þ kl2 Þl ¼ 0; ðr2 þ kt2 Þm ¼ 0; ðr2 þ kt2 Þn ¼ 0; ð2:38Þ

ðk þ 2lÞrðr uÞ lr r u þ .x2 u ¼ 0: ð2:37Þ

with I0 jE0 j2 and