NANOSTRUCTURED MATERIALS
Selected Synthesis Methods,
Properties and Applications
Other Books in Series
DIAMOND: Electronic Properties and Applications
Lawrence S. Pan and Don R. Kania, Editors
ION IMPLANTATION: Basics to Device Fabrication
Emanuel Rimini, Author
SOLID STATE BATTERIES: Materials Design and Optimization
C. Julien and G. Nazri, Authors
SOL-GEL OPTICS: Processing and Applications
L.C. Klein, Editor
PHOTOREFRACTIVE EFFECTS AND MATERIALS
David Nolte, Editor
MATERIALS FOR OPTOELECTRONICS
Maurice Quillec, Editor
PIEZOELECTRIC ACTUATORS AND ULTRASONIC MOTORS
Ken Uchino, Author
WIDE-GAP LUMINESCENT MATERIALS: Theory and Applications
Stanley R. Rotman, Editor
THIN FILM FERROELECTRIC MATERIALS AND DEVICES
R. Ramesh, Editor
MICRO-ACTUATORS: Electrical, Thermal, Optical, Magnetic,
Mechanical and Acoustic
Massood Tabib-Azar, Author
HIGH-TEMPERATURE SUPERCONDUCTORS: Materials,
Properties and Applications
Rainer Wesche, Author
NANOCRYSTALLINE METALS AND OXIDES: Selected Properties
And Applications
Philippe Knauth, Joop Schoonman
THE KLUWER INTERNATIONAL SERIES IN:
ELECTRONIC MATERIALS: SCIENCE AND TECHNOLOGY
Series Editor
HARRY L. TULLER
Massachusetts Institute of Technology
NANOSTRUCTURED MATERIALS
Selected Synthesis Methods,
Properties and Applications
edited by
Philippe Knauth
Professor
Université de Provence
Marseille, France
Joop Schoonman
Professor
Delft University of Technology
Delft, The Netherlands
KLUWER ACADEMIC PUBLISHERS
NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW
eBook ISBN: 0-306-47722-X
Print ISBN: 1-4020-7241-4
©2004 Kluwer Academic Publishers
New York, Boston, Dordrecht, London, Moscow
Print ©2002 Kluwer Academic Publishers
All rights reserved
No part of this eBook may be reproduced or transmitted in any form or by any means, electronic,
mechanical, recording, or otherwise, without written consent from the Publisher
Created in the United States of America
Visit Kluwer Online at: http://kluweronline.com
and Kluwer's eBookstore at: http://ebooks.kluweronline.com
Dordrecht
CONTENTS
Nanomaterials Production by Soft Chemistry
M.-P. Pileni
Hydrothermal Synthesis of Nanomaterials
O. Schäf, H. Ghobarkar, and P. Knauth
EHDA in Particle Production
T. Ciach, K.B. Geerse, and J.C.M. Marijnissen
Dynamic Compaction of Nano-Structured Ceramics
M.J.G. Jak
Structural and Electrical Properties of Nanostructured
and Coarse Manganese Spinel
J. Molenda and J. Marzec
Metal-Polymer Nanocomposites: Formation and Properties
near the Percolation Threshold
S. Zavyalov, A.Timofeev, A.Pivkina, and J.Schoonman
Nanocrystalline Layers of CdSe Produced by Means
ofa Multilayer Approach
D. Nesheva, Z. Levi, I. Bineva, and H. Hofmeister
X-Ray Diffraction from Nanostructured Materials
J. Pielaszek
High Resolution Electron Microscopy of Surfaces and Interfaces
H. W. Zandbergen
Nanoelectronics
G. Allan, C. Delerue, C. Krzeminski, and M. Lannoo
Index
23
43
55
1
2. Properties and Applications
73
97
115
127
145
161
185
1. Synthesis and Processing
PREFACE
In the framework of the rapid development of Nanoscience and
Nanotechnology, the domain of Nanostructured Materials is attracting more
and more researchers, both academic and industrial. Synthesis methods are a
major prerequisite for achievement in this rapidly evolving field. This book
presents several important recent advances in synthesis methods for
nanostructured materials and processing of nanoobjects into macroscopic
samples, such as nanocrystalline ceramics. The chapters do not cover the
whole spectrum of possible synthesis techniques, which would be limitless,
but present highlights especially in the domains of interest of the editors.
M.-P. Pileni presents “chimie douce” approaches for preparation of a large
variety of nanostructured materials, including metals, alloys, semiconductors
and oxides. Normal micelles, i.e. oil in water droplets, stabilized by a
surfactant, and reverse micelles, i.e. water in oil droplets, are used as
nanoreactors. Spherical nanoparticles and nanocrystals with a shape
anisotropy, such as nanorods, can be obtained.
O. Schäf et al. demonstrate that hydrothermal synthesis with water as solvent
and reaction medium can be specifically adapted to nanostructured materials,
if crystal growth can be avoided and a high degree of supersaturation is
maintained, thereby enhancing the rate of nucleation. Different modes of
operation are presented, including rapid expansion of supercritical solutions,
rapid thermal decomposition of precursors in solution and static high-pressure
hydrothermal synthesis.
T. Ciach et al. present “electrospraying” as a powerful new route for the
preparation of nanoparticles, especially of oxides for electroceramics. The
principle of the electro-hydro-dynamic atomization technique and bipolar
coagulation, i.e. mixing of two electrospray droplets, are explained and
examples of nanoparticle production are given.
M. Jak then shows how nanoparticles can be processed into nanostructured
ceramics, by using dynamic compaction techniques. This chapter covers the
explosive compaction and the room-temperature magnetic-pulse compaction
techniques. Examples of ceramic oxides and nanocomposites are presented,
which show the broad applicability of these techniques and the advantages in
comparison with classical compaction.
The following chapters are devoted to selected properties and applications of
nanostructured materials and will be a good complement to those already
presented in our previous volume in this series (P. Knauth, J. Schoonman, ed.,
Nanocrystalline Metals and Oxides : Selected Properties and Applications,
Kluwer, Boston, 2002).
J. Molenda et al. discuss the electrical and electrochemical properties of
nanostructured manganese spinel, that is potentially important as cathode
material in rechargeable lithium-ion batteries.
S. A. Zavyalov et al. use co-condensation of metal nanoparticles in a polymer
matrix to prepare nanocomposites. The percolation threshold of metal
nanoparticles is important for the electrical, optical, and chemical properties of
the nanocomposites.
D. Nesheva et al. study evaporated multilayers with continuous
nanocrystalline CdSe layers, or with discontinuous nanocrystals of CdSe,
depending on the thickness and morphology of the other layer ( or
ZnSe). Quantum size effects are attributed to 1D carrier confinement in the
continuous CdSe layers and quasi-3D confinement in the discontinuous
“composite” films.
J. Pielaszek discusses how X-ray diffraction can be applied to study the
structure of nanostructured materials. The determination of an average
crystallite size from reflection broadening, especially using Scherrer’s
equation, and the Rietveld refining, and atomistic modeling of X-ray
diffraction patterns from nanostructured materials are described.
H. Zandbergen introduces High Resolution Electron Microscopy as a powerful
tool for studying nanostructures. The imaging process of HREM, the relations
between the micrography and the material structure are described and
examples of HREM studies of grain boundaries are shown. Ways to
investigate the chemical composition and electronic effects are outlined.
Given the tremendous importance of this field for the future development of
the industrialized world and mankind in general, the final chapter of this book
is devoted to « nanoelectronics ». G. Allan et al. present a prospective on
further size reduction in microelectronics and on the future of molecular
electronics. The first part treats devices built on inorganic materials and
quantum effects, such as tunneling junctions. The second part introduces the
concepts of molecular diodes, wires, and circuits, experiments on molecules in
solution and imaging and fabrication of molecular objects.
We thank all the colleagues who spend considerable time and effort in writing
these high-level contributions. We are also pleased to acknowledge the
support of the series editor Prof. Harry Tuller and of Greg Franklin, senior
editor at Kluwer Academic Publishers.
P. K. gratefully acknowledges the support by the "Centre National de la
Recherche Scientifique (CNRS)", the North Atlantic Treaty Organization
(NATO), the European Union (COST 525 project), and the National Science
Foundation (NSF) that helped to realize this and other projects on
nanostructured materials. J. S. acknowledges the European Science
Foundation-NANO program and the Delft Interfaculty Research Center
“Renewable Energy” for support of exchange visits and nanoparticle research.
Marseille, France, and Delft, The Netherlands, June 2002.
P. Knauth and J. Schoonman
viii
Nanomaterials Production by Soft Chemistry
M.P.Pileni
Laboratoire LM2N, Université P. etM. Curie (Paris VI), B.P. 52, 4 Place Jussieu, F - 752 31 Paris
Cedex 05, France.
Abstract. In this paper, various ways to make nanocrystals are presented.
It is possible, by using colloidal self-assemblies as nanoreactors, to
produce a large variety of nanoparticles, semiconductors, metals, oxides
and alloys The limitations in using these colloidal solutions to produce
nanomaterials are pointed out.
2
I. Introduction.
During the last decade, due to the emergence of a new generation of high
technology materials, the number of groups involved in nanomaterials has
increased exponentially1,2
. Nanomaterials are implicated in several
domains such as chemistry, electronics, high density magnetic recording
media, sensors and biotechnology. This is, in part, due to their novel
material properties, that differ from both the isolated atoms and the bulk
phase. An ultimate challenge in materials research is now the creation of
perfect nanometer-scale crystallites (in size and shape) identically
replicated in unlimited quantities in a state than can be manipulated and
that behave as pure macromolecular substances.
The essential first step in the study oftheir physical properties and the use
of nanomaterials in various technologies is their production. Physical and
chemical methods were developed: ball milling3,4
, a flame by vapor phase
reaction and condensation5
, chemical reduction1
and coprecipitation1,6-10
have been employed to control the particle size (up to 10 nm) and
morphology. More recently, new preparation methods have been developed
such as sonochemical reactions11
, gel-sol12
, microwave plasma13
and low
energy cluster beam deposition14
. In 1988, we developed a method based
on reverse micelles (water in oil droplets) for preparing nanocrystals15
. In
1995, we were able to control size and shape of nanocrystals by using
colloidal solutions as templates2
. Using these methods we fabricated
various types of nanomaterials : metals
semiconductors (CdS, CdTe etc...) and alloys (Fe-Cu, CdMnS, CdZnS).
Normal micelles make it possible to produce ferrite magnetic fluids16
.
Some of the techniques described above enable preparing amorphous
nanoparticles whereas others favor formation of highly crystallized
nanoobjects.
One of the challenges is to produce anisotropic nanocrystals. Colloidal
solutions can be used as a nano-reactor17
whose shape partially plays a role
in the shape of the nanomaterial produced (see below). In the last two
3
years, a large number of groups have succeeded in making nanorods.
Various nanomaterials such as silver18,19
, gold19-23
, platinium24,25
, copper26
and semiconductors27,28
were produced. In most cases, a surfactant, usually
cethyltriethylammonium bromide or its derivatives, is added to the
preparation solution. However, the colloidal solution is a mixture of several
compounds and its structure is unknown. The surfactant is used as a
polymer and plays an important role in formation of nanorods. The aspect
ratio (length to width ratio) markedly depends on the amount of surfactant,
but the key factor in favoring the nanocrystal growth in a given direction is
not known.
Self-assembled nanocrystals have attracted increasing interest over the last
five years 17,29,30
. The level of research activity is growing seemingly
exponentially, fueled in part by the observation of physical properties that
are unique to the nanoscale domain. The first two- and three-dimensional
superlattices were observed with and CdSe nanocrystals29,30
. Since
then, a large number of groups have succeeded in preparing various selforganized
lattices of silver31-34
, gold35-38
, cobalt39
, and cobalt oxide40
. With
the exception of CdSe30
and cobalt39
nanocrystals, most superlattice
structures have been formed from nanocrystals whose surfaces are
passivated with alkanethiols. When nanocrystals are characterized by a low
size distribution, they tend to self-organize in compact hexagonal
networks. Conversely, if the distribution is too large they are randomly
dispersed on a substrate. This is obtained when a drop of solution
containing the nanocrystal is deposited on the substrate lying on a paper.
Conversely, when the substrate is fixed with anti-capillary tweezers, rings
made ofnanocrystals, and surrounded by bare substrate, are formed42
. This
is observed with silver, gold, CdS and ferrite nanocrystals. These
phenomena were attributed to either wetting34
or magnetic43
properties. In
fact, they are due to Marangoni instabilities42
. Under other deposition
conditions, large "wires" composed of silver nanoparticles have been
observed, in which the degree of self-organization varies with the length of
4
the alkyl chains coating the particles44
. Interestingly, it has been recently
demonstrated that the physical properties of silver45,46
, cobalt39,47,48
and
ferrites49,50
nanocrystals organized in 2D and/or 3D superlattices differ
from those of isolated nanoparticles. These changes in the physical
properties are due to the short distances between the nanocrystals. Such
collective properties are attributed to dipole-dipole interactions.
Furthermore, the electron transport properties drastically change with the
nanocrystal organization51
.
In this paper a colloidal solution with a well-known structure is used as a
template to produce nanocrystals. The size and shape of the materials are
partially controlled by that ofthe template.
II. Results and discussions.
Colloidal solutions favor partial control of the size and shape of nanomaterials.
In the following, colloid methods developed to make nanocrystals are
described.
II. 1. Spherical nanoparticles
Several techniques are used to produce nanomaterials by soft chemistry. Either
amorphous or crystalline nanoparticles are obtained. Whatever the procedure is,
the major factors involved in controlling the size or shape ofnanoparticles are
confinement, electrostatic interactions, reactant solubilities and large local
amounts of reactant. After or during the production, nanoparticles have to be
passivated to prevent coalescence. One of the first approaches to make
nanoparticles was based on the variation of the solubility product of reactant
with temperature52
. This controls the particle size. Simultaneously, syntheses of
nanoparticles were developed in aqueous solution in presence of a charged
polymer that strongly interacts with one of the reactants53,54
. In the following,
data obtained by using colloidal self-assemblies as a nano-reactor to control the
nanocrystal sizes are given.
5
i) Reverse micelles
Reverse micelles are water in oil droplets stabilized by the surfactant55
. The
water to surfactant concentration ratio, linearly controls the size
of the droplet56
. Reverse micelles are subjected to Brownian motion and during
these collisions the droplets combine to form a dimer with an exchange oftheir
water contents. The dimer dissociates to again form reverse micelles. These two
properties (control of the droplet size56
and the exchange process57
) make
possible the use of reverse micelles as a nano-reactor. Two micellar solutions
are prepared. Each contains one of the reactants. By mixing these solutions, a
chemical reaction takes place58
and nanomaterials are produced1,2
. The droplet
size, which is controlled by w, controls the size of the particle. This procedure
has been used to obtain a large variety of materials such as semiconductors1,59
,
metals1,40,60
and oxides61
.
When the two reactants are present as salts, amorphous nanomaterials are
formed. Conversely, when one of the reactants is a functionalized surfactant
(the reactant is the counter ion of the surfactant) nanocrystals are produced. In
the latter case, it is possible to make well defined alloys like Cd1-yZnyS62
,
Cd1-yMnyS63
whereas it is not possible with salt ions solubilized in the droplets.
All the chemical reactions occurring in aqueous solution cannot be obtained in
micelles. This was well demonstrated with telluride derivatives. CdTe is made
as described above64
. But it is not possible to produce Cd1-yMnyTe65
whereas
Cd1-yMnyS nanocrystals are formed. The replacement of sulfur by the telluride
derivative induces formation of rods of telluride and CdTe nanocrystals65
.
Similarly, it is possible to produce ZnS and not ZnTe. Again telluride rods are
formed. These data clearly show that chemistry in homogeneous solution
(aqueous) differs from that ofcolloids (water in oil droplets).
In the following, one ofthe reactants is a functionalized surfactant. On
increasing the water content, i.e., the size of the nano-reactor (water in oil
droplet), the particle size increases (Fig. 1). However the variation of the
nanocrystal diameter depends on the type of produced material. For II-VI
6
semiconductors59
such as CdS, ZnS, CdTe it is possible
to control the particle diameter from 1.8 nm to 4 nm. Conversely for silver
sulfide66
and copper67
nanocrystals it can be varied from 2 to 10 nm. The major
change in the particle size is obtained at low water content: On increasing the
water content, the particle size increases to reach a plateau around w=20. Ifit is
assumed that the largest particle size and that of the water molecule volume is 1,
CdS, ZnS, CdZnS, CdMnS, PbS, Co, Ag nanocrystals and the water molecule
volume behave similarly as shown in Figure 2. This indicates that the crystal
growth is related to the water structure inside the droplet, which is confirmed by
the change in the O-H vibration ofthis water68
(Fig. 2). This is valid for most of
the nanocrystals produced. Discrepancies are observed with silver sulfide and
silver nanocrystals. With silver sulfide nanocrystals, a linear increase in the
particle size with the water content is seen66
. With silver nanocrystals, the
behavior observed in Figure 2 is obtained when the reducing agent is sodium
borohydride. On using hydrazine as the reducing agent, the behavior markedly
7
changes. At low water content (w=2), the average size of nanocrystals is 5 nm
with a low size distribution (12%). On increasing the water content, the size
distribution markedly increases (40%) and the average diameter is around 3.5
nm. It does not change with increasing w from 10 to 40. Such changes in the
behavior can be related to the fact that with sodium borohydride the chemical
reaction is very fast whereas with hydrazine it is rather slow. It also must be
noted that it is possible to make larger nanocrystals by using a plastic vessel
which again prevents deposition ofa silver nanocrystal film on the surface.
The crystalline structure of nanocrystals usually differs from that of the bulk
phase. As an example, the II-VI semiconductors in the nano-scale are
characterized by a hexagonal structure (Wurtzite)69,70
whereas the bulk phase is
cubic (Zinc Blende). In some cases, the structure of nanocrystals in the phase
obtained in the nanoscale is unstable compared to the bulk phase. In the bulk
phase, silver iodide has, at room temperature, two stable phases and and
an unstable one On the nanoscale range, the and phases are stable and
the phase cannot be detected. These structural changes are observed for a
large number of nanocrystals. Of course, this change in the structural behavior
is not general. For example, silver sulfide nanocrystals form a monoclinic phase
in the nano-scale and in the bulk materials.
8
By replacing a mixed surfactant by a functionalized surfactant, the phase
diagram drastically changes71,72
. To our knowledge, one of most quantitative
studies of the phase diagram obtained using functionalized surfactant was
with copper diethyl sulfosuccinate, water, isooctane. Similar
behavior was obtained with many other divalent surfactants such as
and Reverse micelles are formed in two regions for the phase
diagram: 0<w<5 and 30<w<40. Syntheses in these two regions of the phase
diagram induce formation of copper nanocrystals with average particle sizes
smaller in the 0<w<5 region (12-nm) than those at higher water content (7.5nm)
(Fig. 3)73
. This is explained by the fact that the redox potential changes
with the water structure inside the water pool. At low water content, the
number ofnuclei formed is rather small and large nanocrystals are produced.
On increasing the water content, the number of nuclei increases inducing
formation of smaller nanocrystals. By replacing with in
the water- rich region (30<w<40) cobalt nanocrystals are produced74
. At the
same water content (w=40), the average nanocrystal size. Again this is smaller
in mixed micelles (5.6nm)39
than in pure functionalized surfactant (8 nm)74,
this has to be related to the number of nuclei formed during the chemical
reduction.
ii) Normal micelles
Normal micelles are oil in water droplets and with these the length of the
surfactant alkyl chain controls the size of the droplets. Most of the
nanomaterials produced are made with functionalized surfactants. This induces
a marked increase in the local amount ofreactants and a supersaturation regime
is formed. Conversely to reverse micelles, there is no confinement of the
reactants. Normal micelles act as a polymer to control the nanoparticles size. It
has been possible to make CdS75
and copper76
nanocrystals and magnetic
fluids77
. In the latter case, transition metals involved in the ferrofluids form the
counter ions ofthe micelles. This technique makes it possible to produce a large
variety of alloys78,80
with a composition similar to that of the reactant used for
the syntheses. The size of the nanocrystals, which can vary from 2 to 8 nm, is
controlled by the micellar concentration.
II.2 Nanocrystals having a shape anisotropy.
Colloidal solutions are good candidates for controlling the shape of
nanocrystals. For a given surfactant-water-oil system, it is well known that a
large variety of self-organizations of the surfactant can be produced55,81
. This is
demonstrated from a careful structural study of the phase diagram. Such joint
structural and synthesis studies were done with water - isooctane
solution in the oil-rich region ofthe phase diagram. Figure 4 shows the various
structures of colloidal self-assemblies. At low water content, w<5, reverse
micelles are formed. On increasing the water content, the system evolves from
spheres to interconnected cylinders (5<w<9). A further increase in the water
content results in the appearance of a lamellar phase in equilibrium with
interconnected cylinders (9<w<10). At higher water content a pure lamellar
phase is formed (10<w<15.5). These changes in the colloidal structure are
related to the change in the surfactant head polar group hydration71
. A further
surrounded by interconnected cylinders72
. Then the system evolves to
interconnected cylinders (26<w<31) and spherical reverse micelles (31<w<40).
Chemical reduction ofthe surfactant produces copper nanocrystals73
and, as
9
10
shown in Figure 5, the shape ofthe colloid plays an important role in the shape
of these nanocrystals. In reverse micellar solution (w<5 and 31<w<40),
spherical copper nanocrystals are formed. In the region of interconnected
cylinders (5<w<10 and 26<w<31) cylinders and spheres are produced. It can be
seen that the average size of these nanocrystals is the same in the various w
regions73,82
. In the stable emulsion region various shapes of nanocrystals are
produced (square, triangular, cylindrical). From comparison of Figure 4 and
Figure 5, it can be concluded that the shape ofthe template has to be taken into
account to control the particle size and shape. A question arises : Is the template
the only parameter which plays a role in controlling the shape of
nanocrystals83
?
Let us consider the region corresponding to 9<w<10, where most of the
nanocrystals produced are spheres with a very small number of cylinders
(Fig. 6A). Addition ofvarious salts such as NaCl, NaBr, KC1, KBr
11
and in the mesostructured fluids does not induce, from the
macroscopic viewpoint, drastic changes in the phase diagram. With
overall salt concentration, the three phases remain. The relative volume of
isooctane remains constant and is rather large compared to the others.
Slight changes, close to experimental errors (10%), in the relative volumes
of birefringent and isotropic phases are observed. From SAXS
measurements ofthe isotropic phase, it is seen that there is a slight increase
in the characteristic diameter and a decrease in the head polar area by salt
addition. However no changes in the parameters were observed by
changing the salt. The scattering curves obtained with various salts are
superimposed and from this it is concluded that the macroscopic structure
ofthe template remains unchanged.
12
Synthesis in presence of NaCl induces formation of very long rods
(Fig. 6B) and along with these, nanocrystals having various shapes (square,
triangular, etc...) are produced. The number of rods is too small to be
described statistically. However, the shortest length and width of the rods
ar 650 nm and 15 nm whereas the maximum values are 1600 nm and 25
nm, respectively. These rods are highly crystalline84
(Fig. 7A). The
HRTEM images of a sample tilted at 0° ±18° and ±36° are in Fig. 7B.
After tilting these cylindrical particles, a periodic sequence can be
observed, i.e., every 36° a complete repetition ofthe high-resolution image,
clearly displayed in the power spectrum, is achieved. A second different
configuration is obtained by tilting around ±18°. From the real image, it
can be seen that for 0° and ±36° tilt angles, the central part of the image
shows atomic resolution with strong contrast. However, the particles tilted
by ±18° only show lattice planes on the left or the right part of the image.
Indeed, the reflections obtained with the different tilt angles correspond to
the same lattice parameters and the angles between planes are also in good
13
agreement. The structure of copper rods is a truncated decahedron with 5fold
symmetry. A similar structure with a 5-fold symmetry is observed
with cylinders shown in Figure 5 and Figure 6A. Hence, cylinders and rods
of copper nanocrystals do not have the same structure as the bulk phase.
This is rather surprising because it was established that nanomaterials
having a size larger than 10 nm were characterized by a crystalline phase
similar to that ofthe bulk. This claim is probably wrong. In fact, similar 5
fold symmetries were observed recently for silver85
and gold86
cylindrical
nanocrystals.
By replacing NaCl by KC1, similar behavior is observed with formation of
long rods (Fig. 8A). Rods are also obtained with or CTAC (Figs. 8B
and 8C). It can be noticed that there are still spherical and other shapes
14
(square, triangular, etc.). However, no small cylinders are observed in
presence of chloride ions which clearly indicates that only the cylinders
grow to form rods. This is well demonstrated by following the rod
formation kinetics. Figure 9 shows that the length of the cylinders
increasespreferentially compared to the width. This indicates that chloride
ions inhibit the growth along the [100] direction and favor that along the
[110] direction. Of course this phenomenon depends on the amount of
chloride ions present in solution. On increasing the chloride ion
concentration, the relative length versus width of the rods increases and
then decreases as shown in Figure 10. The maximum is reached for
chloride ions. The same behavior is observed with and CTAC as
additives 87
. The length increases with chloride concentration and then
decreases above whereas the width increases progressively. This
15
indicates that above a given chloride concentration, the inhibition
efficiency along the [100] direction decreases. On annealing the rods, two
behaviors are observed88,89
:
i) The rods remain stable at high temperature (925°C). We have to keep in mind
that the melting temperature ofbulk (phase) copper is 1083°C.
ii) The rods form small cylinders having the same structure as the long
rods. Hence even after annealing, the rods do not crystallize in the FCC
structure ofthat of copper in the bulk phase.
16
By replacing NaCl by NaBr at the same salt concentration the
number of rods markedly decreases (Fig. 6C). The longest length and
width are 400 nm and 56 nm whereas the shortest are 230 nm and 24 nm,
respectively. It can be seen that that a relatively large amount of square flat
nanocrystals appear. The major difference between additions of NaCl and
NaBr is the size of the rods and the appearance of this large amount of
square nanocrystals. In presence of no rods are formed and spheres
are produced (Fig.6D). These data clearly show that the anions added to the
microstrucrure play an important role in the nanocrystal shape. Two
phenomena take place:
i) The production yield of copper nanocrystals drastically increases by salt
addition. This does not seem to change markedly with the salt used and is
related to data obtained in presence of CTAC in mixed reverse micelles87
.
This could be attributed to screening of the copper ions interactions with
the anionic head polar group of the surfactant. This induces changes in the
redox potential of copper ions.
ii) The chloride ions adsorb specifically on a given direction of the lattice
plane ofthe nanocrystals and favor the growth in this direction (110). This
adsorption process differs with the salt anion.
References
17
1- Pileni, M.P.; J. Phys. Chem.; 1993, 97, 6961
2- Pileni, M.P.; Langmuir; 1997, 13, 3266
3- Goya, G.F.; Rechenberg, H. R.; J. Mag. Mag. Mat. 1999, 196-197, 191.
4- Fatemi, D. J.; Harris, V. G.; Browning, V.M.; Kirkland, J. P.; J. Appl.
Phys. 1998, 83, 6867
5- Elmore, W.C.; Phys.Rev. 54, 1938, 309
6- Sato, T.; lijima, T.; Seki, M.; Inagaki, N.; J. Mag. Mag. Mat. 1987, 65,
252
7- Ueda, M.; Shimada, S.; Inagaki, M. J.; Eur. Cer. Soc. 1996, 16, 685
8- Yokoyama, M.; Sato, T.; Ohta, E.; Sato, T.; J. Appl. Phys. 1996, 80,
1015.
9- Chen, Q.; Zhang, Z. ; J. Appl. Phys. Lett. 1998, 73, 3156.
10- Vayssières, L.; Chanéac, C.; Tronc, E.; Jolivet, J.P.; J. Coll. Int. Sci.
1998, 205, 205.
11- Shafi, K.V.P.M.; Koltypin, Y.; Gedanken, A.; Prozorov, R.; Balogh, J.;
Lendvai, J.; Felner, I. ; J. Phys. Chem. B 1997, 101, 6409.
12- Sugimoto, T.; Shimotsuma, Y.; Itoh, H. ; Powder Tech. 1998, 96, 85.
13- Vollath, D.; Szabo, D.V.; Taylor, R.D.; Willis, J.O.; J. Mat. Res. 1997,
12,2175.
14- Perez, A.; Melinon, P.; Dupuis, V.; Prevel, B.; Bardotti, L.; TuallonCombes,
J.; Maselli, B.; Treilleux, M.; Pellarin, M.; Lerme, J. ; Cottancin,
E.; Broyer. M. ; Janet, M.; Negrier, M.:; Tournus, F.; Gaudry, M.;
Materials Transactions, in press (2001)
15- Petit, C. and Pileni, M.P., J. Phys. Chem., 1988, 92, 2282,
16- Moumen, N. and Pileni M.P.; J. Phys. Chem.; 1996, 100, 1867.
17- Pileni, M.P.; J. Phys. Chem.; 2001, in press
18- Jana, N.R.; Gearhearth, L.; Murphy, C. ; J. Chem. Commun, 2001, 617
19- Cepak, V.M.; Martin, C.R.; J. Phys. Chem.; 1998, 102, 9985
20- Wang, Z.L.; Gao, R.P.; Nikoobakht, B.; El Sayed, M.A.; J. Phys.
Chem.; 2000, 104, 5417
18
21- Bianca, M.; Van der Zande, I.; Böhmer, M.R.; Lambertus, G.; Fokkink,
J.; Schönenberger, C.; J. Phys. Chem.; 1997, 101, 852
22- Yu, Y.Y.; Chang. S.S.; Lee, C.L.; Chris Wang, C.R. ; J. Phys. Chem.;
1997, 101, 6661
23- Jana, N.R.; Gearhearth,L.; Murphy, C.J.; J. Phys. Chem.; 2001, 105,
4065
24- Kyotani, T.; Tsai.L.F.; Tomita, A.; Chem. Commun.; 1997, 701
25- Henglein, A.; Giersig. J. Phys. Chem.; 2000, 104, 6767
26- Toimil Molares, M.E.; Buschmann, V.; Dobrev, D.; Neumann, R.;
Scholz, R.; Schuchert, I.U.; Vetter, J.; Adv. Mater.; 2001, 13,62
27- Wang,S.; Yang, S.; Langmuir; 2000, 16, 389
28- Chen, C.C.; Chao, C.Y. Lang, Z.H.; Chem. Mater.; 2000, 12, 1516
29- Motte L., Billoudet F., Pileni M.P.; J Phys Chem.; 1995, 99, 16425
30- Murray. C.B., Kagan C.R., Bawendi M.G.; Science ; 1995, 270, 1335
31- Harfenist S.A., Wang Z.L., Alvarez M.M., Vezmar I., Whetten R.L.; J.
Phys. Chem.; 1996, 100, 13904
32- Taleb A., Petit C., Pileni M.P.; Chem. Mater. ; 1997, 9, 950
33- Harfenist S.A.; Wang Z.L.; Whetten R.L.; , I.; Alvarez, M.M. ; Adv.
Mater.; 1997, 9, 817
34- Ohara, P.C.; Heath, J.R.; Gelbart, W.M.; Angew Chem. Int. Engl.;
1997, 36, 1078
35- Brust, M.; Bethell, D.; Schiffiin, D.J.; Kiely, C.; Adv. Mater.; 1995, 9,
797
36- Hostetler, M.J.; Stokes, J.J.; Murray, R.W.; Langmuir; 1996, 12, 3604
37- Vijaya Sarathy, K.; Raina, G.; Yadav, R.T.; Kullkarni, G.U.; Rao,
C.N.R.; J. Phys. Chem. B ; 1997, 101, 9876
38- Badia, A.; Cuccia, V.; Demers, L.; Morin, F.; Lennox, R.B.; J. Am.
Chem. Soc.; 1997, 119, 2682
39- Petit, C.; Taleb, A.; Pileni, M.P.; J. Phys. Chem.; 1999,103, 1805
40- Yin, J.S., Wang, Z.L; Phys. Rev. Lett.; 1997, 79, 2570
19
41- Vossmeyer, T.; Chung, S.; Gelbart, W.M.; Heath, J.R.; Adv. Mater. ;
1998, 10,351
42- Maillard, M.; Motte, L.; Ngo, A.T.; Pileni, M.P.; J. Phys. Chem. B;
2000, 104, 11871
43- Kurrika, V.; Shafi, P.M.; Felner, I.; Mastai, Y.; Gedanken, A.; J. Phys.
Chem.; 1999, 103, 3358
44- Motte L.; Pileni M.P.; J. Phys. Chem. B ; 1998, 102, 4104
45- Taleb, A.; Petit, C; Pileni, M.P.; J. Phys. Chem. B ; 1998, 102, 2214
46- Taleb, A.; Russier, V.; Courty, A.; Pileni, M.P.; Phys. Rev. B.; 1999,
59, 13350
47- Petit, C.; Taleb, A.; Pileni, M.P.; Adv. Mater.; 1998, 10, 259
48- Russier, V.; Petit, C.; Legrand, J.; Pileni, M.P.; Phys. Rev. B; 2000, 62,
3910
49- Ngo, T. and Pileni, M.P.; Adv. Mater.; 2000, 12, 276,
50- Ngo, T. and.Pileni, M.P.; J. Phys. Chem.; 2000, in press
51- Taleb, A.; Silly, F.; Gusev, O.; Charra, F.; Pileni, M.P.; Adv. Mater.;
2000, 12, 119
52- Brus, L.E.; J. Chem. Phys.; 1983, 79, 5566
53- Henglein, A.; Chem. Rev.; 1989, 89, 1861
54- Fendler, J.H.; Chem. Rev., 1987, 87, 877
55- “Structure and Reactivity in Reverse micelles” Pileni, M.P.; (ed.) Pub
Elsevier ,1989, Amsterdam, Oxford, New York, Tokyo
56- Pileni, M.P.; Zemb, T.; Petit, C.; Chem. Phys. Lett.; 1985, 118, 414
57- Jain, T.K.; Cassin, G.; Badiali, J.P.; Pileni, M.P.; Langmuir; 1996, 12, 2408
58- Fletcher,P.D.I.; Robinson, B.H; Ber. Bunsenges. Phys. Chem.; 1981, 85,
863
59- Pileni, M.P.; Catalysis Today; 2000, 58, 151
60- Pileni, M.P.; Langmuir; 2001, in press
61- Pileni, M.P.; Adv. Funct. Mater.;2001, in press
62- Cizeron, J. and Pileni, M.P.; J. Phys. Chem.; 99, 1995, 17410
20
63- Levy, L.; Hochepied, J.F. and Pileni, M.P.; J. Phys. Chem.; 1996, 100,
18322
64- Ingert, D.; Feltin, N.; Levy, L.; Gouzert, P. and Pileni, M.P.; Adv. Mater.;
1999, 11,220
65- Ingert, D. and Pileni, M.P., Adv. Funct;. Mater.; 2001, 11, 136
66- Motte, L. Billoudet, F. and. Pileni, M.P.; J. Phys. Chem.; 1995, 99, 16425
67- Lisiecki, I. and Pileni, M.P.; J. Amer. Chem. Soc.; 1993, 115, 3887
68- “Hydrogen bound Networks ” Motte, L.; Lisiecki, I. and Pileni M.P.; 1994,
447, Ed. Dore, J. and Benissan, M.C.; NATO publisher
69- Cizeron, J.; Pileni, M.P.; J. Phys. Chem.; 1995, 99, 17410
70- Vogel, W.; Urban, J.; Kundu, M; Kulkarni, S.K.; Langmuir, 1997, 13, 827
71- Lisiecki, I.; André, P.; Filankembo, A.; Petit, C.; Tanori, J.; GulikKrywicki,
T.; Ninham, B.W. and Pileni, M.P.; J. Phys. Chem.; 1999, 103, 9168
72- Lisiecki, I.; André, P.; Filankembo, A.; Petit, C.; Tanori, J.; GulikKrywicki,
T.; Ninham, B.W. and Pileni, M.P.; J. Phys. Chem.; 1999, 103, 9176
73- Tanori, J. and Pileni, M.P; Langmuir; 1997, 13, 639
74- Legrand, J.; Petit, C. and Pileni, M.P.; J. Phys. Chem.; 2001, 105, 5643
75- Petit, C.; Jain, T.K.; Billoudet, F. and Pileni, M.P; Langmuir; 1994, 10,
4446,
76- Lisiecki, I.; Billoudet, F. and Pileni, M.P; J. Phys. Chem.; 1996, 100, 4160,
77- Moumen, N. and Pileni, M.P.; J. Phys. Chem.; 1996, 100, 1867
78- Ngo, A.T.; Bonville, P. andPileni, M.P.; Eur. Phys. J. B.; 1999, 583
79- Hochepied, J.F. and Pileni, M.P.; J. Phys. Chem.; 2000, 104, 905
80- Hochepied, J.F. and Pileni, M.P.; J. Applied Phys.; 2000, 87, 2472
81- “The language ofshape Ed. Hyde,S.; Andersoon, S.; Larsoon, K.; Blum, Z.;
Landth, T.; Lidin, S.; Ninham, B.W.; Pub. Elsevier: Amsterdam, New York,
Oxford, Shannon, Tokyo, 1997
82- Pileni, M.P.; Ninham, B.W.; Gulik-Krzywicki, T.; Tanori, J.; . Lisiecki, I.
and Filankembo, A.; Adv. Mater. ; 1999, 11, 1358
83- Filankembo, A. and Pileni, M.P.; J. Phys. Chem.; 2000, 104, 5867
21
84- Lisiecki, I.; Filankembo, A.; Sack-Kongehl, H.; Weiss, K.; Pileni, M.P.
and Urban, J.; Phys. Rev. B.; 2000, 87, 1
85- Nepijko, S.A.; Levlev, D.N.; Schulze, W.; Urban, J. and Ertl, G.; Chem.
Phys. Chem.; 2000, 3, 140
86- Wang, Z.L.; Gao. R.P.; Nikoobakht, B. and El Sayed, M.A.; J. Phys.
Chem.; 2000, 104, 5417
87- Lisiecki, I.; Borjling, M.; Motte, L.; Ninham, B.W. and. Pileni., M.P.;
Langmuir; 1995, 11, 2385
88- Lisiecki, I.; Sack-Kongehl, H.; Weiss, K.; Urban and Pileni, M.P.;
Langmuir; 2000, 16, 8802,
89- Lisiecki, I.; Sack-Kongehl, H.; Weiss, K.; Urban, J. and Pileni, M.P.;
Langmuir, 2000, 16, 8807
HYDROTHERMAL SYNTHESIS OF NANO-
MATERIALS
O. Schäf, H. Ghobarkar* and P. Knauth
Université de Provence-CNRS, Laboratoire Matériaux Divisés, Revêtements,
Electrocéramiques (MADIREL, UMR 6121), Centre St Charles, Case 26, 13331 Marseille
Cedex 3, France
* Freie Universität Berlin, Institut für Mineralogie, Takustr. 6, 14195 Berlin, Germany
1. INTRODUCTION
Hydrothermal processing is a non-conventional method to obtain
nanocrystalline inorganic materials. A direct precursor-product correlation
exists allowing the tailoring of almost any materials synthesis without the
presence of further structure directing agents 1,2
.
In short, the synthesis method uses the solubility in water of almost all
inorganic substances at elevated temperatures and pressures, and
subsequent crystallization of the dissolved material from the fluid. As
implemented in the name, water at elevated temperatures plays an essential
role in the precursor material transformation. If one relies on the phase
diagram of water, only the branch between liquid and gaseous water is
essential for further considerations (see fig. 1): below 100°C the
equilibrium vapor pressure of liquid water is below 1bar; above the boiling
point the important hydrothermal pressure range is available. Two
different operation modes can be distinguished in general.
In the autogeneous pressure range, the process pressure is due to the
existence of a gas phase in equilibrium. Here, the pressure of the closed
system is given by temperature, corresponding to the equilibrium line in
figure 1. Above 374°C, corresponding to a pressure of 221bar, gas and
liquid phase can not be distinguished any more (critical point). However,
the autogeneous pressure at a given temperature is often not sufficient for
the necessities of hydrothermal processing, although a great number of
conventional synthesis experiments are conducted in the region below the
critical point.
In the high-pressure mode, water under an external pressure more
elevated than the equilibrium water vapor pressure is used. Under these
24
conditions remarkable synthesis results can be obtained because the
solubility of inorganic materials is generally rising with pressure.
Temperature, water pressure and the time of reaction are, therefore, the
three principal physical parameters in hydrothermal processing.
Generally, the synthesis method is claimed solvothermal3,4
, which
means that water, but also other solvents are used, in accordance with the
general process principle. It is also possible that additives are employed to
modify the initial properties of pure hydrothermal water. Polar solvents
(e.g. aqueous solutions containing HF, or other acids or bases to
trim pH) or even non-polar solvents (e.g. pure, supercritical ) can be
used for the dissolution–recrystallization process in order to widen the
range of applicability of this synthesis method. Although they present some
advantages, these solvents also have specific disadvantages concerning
mainly their toxicity and corrosion problems for the autoclave materials. In
addition, the simplicity and elegance of the original process is getting lost.
2. PRINCIPLES AND OPERATION MODES
Both physical and chemical parameters can be trimmed systematically
during the hydrothermal processing in such a way that a direct precursorproduct
correlation can be established. However, hydrothermal processing
is rather complex, because a great number of interacting parameters play
essential roles5,6
. The influence ofthese parameters can only be sketched as
the in-situ observation of the process still rests a problem, which has not
yet been solved satisfyingly.
Temperature plays an important role in kinetics of product formation
as well as on the thermodynamic stability ofthe product phase. Pressure is
essential for the solubility, the supersaturation range directing the
crystallization process as well as for the thermodynamic stability of the
product phase (e.g. a too high synthesis pressure leads to the crystallization
of denser phases). In order to be independent from the temperature driven
autogeneous pressure, the external pressure mode is actually the preferable,
as low temperature-high pressure syntheses are only available by this highpressure
hydrothermal synthesis (HPHS) mode. Time is also an important
parameter because the synthesis of kinetically stable phases are favored in
short term processes while the thermodynamically stable phases are
generally formed in long term experiments - corresponding to the chosen
temperature-pressure regime. This is caused by re-dissolution and recrystallization
of the already formed phases. However, as will be shown
later, further related reaction conditions such as convection within the
reaction vessel and the state of aggregation of the precursor materials (e.g.
ifthey are present as macroscopic crystals, nanocrystalline phases or glass)
must not be neglected in the effort to control the process parameters.
Based on experimental results, figure 2 shows the principle of the
hydrothermal process of dissolution-supersaturation and subsequent
crystallization. At a given hydrothermal temperature and pressure
convenient for the synthesis (e.g. 1kbar water pressure and temperatures of
about 300°C) the precursor material is continuously dissolved in the
hydrothermal fluid. The formation of gels is not observed at any time
during the process - even ifalumosilicate materials are used because bigger
molecular units are hydrolysed at elevated temperature and pressure7,8,9
:
In an aqueous solution under autogeneous pressure conditions far below
the critical point, different states of dissolution might be existent and, most
important, not only the basic structural building units can be present, but
even colloidal states. HPHS synthesis implements, therefore, a first step of
crackdown of possibly present “macromolecular” units by chemical
reaction, existing e.g. as colloidal solution, as precipitated colloidal
solution (crystalline, partially crystalline (e.g. gel), glassy and amorphous)
or solid state precursor materials of the same kind, because bigger units
exceeding the size present in true solutions are not stable under high
25
26
pressure hydrothermal conditions. The formation of a true solution is
therefore assumed in which the smallest possible structural building units
as well as cations with their respective hydration spheres are transported.
The high pressure hydrothermal conditions do not only increase the
solubility in the aqueous medium in comparison to e.g. autogeneous
pressure conditions, the region of supersaturated solution, which is
essential for continuous crystal growth, is also widened at a given pressure
Standard hydrothermal experiments are conducted under isothermal
and isobar conditions without agitation. At the beginning of the
experiment, the hydrothermal fluid consists only of water with solid
precursor materials (I –fig. 2); it gets more and more concentrated with
time. Even above the solubility limit of the product phase (point II,
supersaturation), the precursor materials can still be dissolved. Glassy
precursors are faster dissolved than crystalline materials of the same
27
composition. The zone of supersaturation is wider in case of glassy starting
materials than for respective crystalline materials and, in addition, pressure
dependent. At a certain level of supersaturation (marked as point III), a
spontaneous crystallization will finally occur, leading to a decrease of the
concentration in the hydrothermal fluid during the further run of the
experiment.
Jung and Perrut10
give an actual overview on particle design using
solvothermal methods. While most of these methods are referred to the
treatment of organic materials and pharmaceutical products, the synthesis
of inorganic materials has also been successfully performed using
solvothermal fluid processes under dynamic flow conditions avoiding the
stationary conditions described above. Among them the rapid expansion of
supercritical fluid solutions (RESS) and the rapid thermal decomposition of
precursors in solutions (RTDS) are the most relevant for the synthesis of
nanocrystalline inorganic materials. Studies have been performed using
these methods also with water as pressure and reaction medium.
2.1 RAPID EXPANSION OF SUPERCRITICAL FLUID
SOLUTIONS (RESS)
If the RESS process is used with water as the solvothermal medium,
the precursor material is dissolved under high pressure and high
temperature conditions. The dissolution process is stopped before
spontaneous nucleation occurs (see fig. 2). For practical purposes, this is
achieved under dynamic flow conditions. Nucleation without further
growth is performed by an abrupt decrease of the solubility by expansion
through a heated nozzle into a medium under ambient temperature and
pressure conditions, inducing spontaneous nucleation of the dissolved
materials without significant crystal growth11,12,13
. Pressure, temperature
and time of exposure to hydrothermal conditions are important parameters
in order to obtain the desired nanocystalline materials. In this way,
agglomerates of nanocrystalline and KI were obtained14
, but
also nanocrystalline Fe2O3
15
and ZrC2
6
. Either salts of high solubility or
oxide materials dissolved under high pressure hydrothermal conditions are
used as precursors.
28
2.2 RAPID THERMAL DECOMPOSITION OF PRECURSORS IN
SOLUTIONS (RTDS)
The method of rapid thermal decomposition of precursors in solutions
(RTDS) based on high pressure high temperature hydrothermal processing
uses water for hydrothermal reactions of precursor materials already
dissolved under environmental conditions, leading to a precipitation of a
nanocrystalline product. The only brief exposure of a few seconds of the
pre-pressurized precursor solution to high temperatures (ca. 300-450°C)
initiate a particle nucleation, but prevents a further, subsequent crystal
growth of the nucleated product. In contrast to the standard HPHS
conditions, the hydrothermal process is used to conduct reactions of the
solution, which means that it is performed under dynamic flow and
subsequent quenching to ambient pressure and temperature conditions.
The precursor material used depends on the material to be synthesized
and it has to be, to a certain extent, soluble in water under ambient
conditions. Therefore, inorganic salt precursors with moderate to high
solubility under ambient conditions are used. Table 1 shows the grain size
of products obtainable by RTDS, adapting precursor salt and the
hydrothermal parameters. More detailed descriptions of this technique can
be found in references 17,18,19,20,21.
2.3 HIGH PRESSURE HYDROTHERMAL SYNTHESIS (HPHS)
Achieving the limit of supersaturation as shown in figure 2 means that
spontaneous nucleation occurs from the solution. The nucleation and a
possible crystal growth are, however, dependent on the kinetics of the
water-precursor-system under investigation22
. A growing crystal nucleus is
always active because all the material transported to its surface and
29
compatible with its physical and chemical properties will be built in. This
might lead to intercalation or even ion exchange behavior. Growth
processes of the nuclei are only possible in the region of supersaturated
solution. Only if the isothermal-isobar conditions in the standard synthesis
mode outlined in figure 2 are carefully maintained during a long term
experiment of hours up to several weeks (60d) and the supersaturation limit
is neither exceeded (so that the formation of a second phase can be
observed by another spontaneous crystallization and growth) nor the
concentration falls below the saturation limit (so that etching of the already
formed crystals occurs), microscopic or even bigger single crystals will be
observable. Both nucleation and growth take place in the region of
supersaturated solution although nucleation is assigned to its upper limit
and the rate of nucleation and the crystal growth rate are strongly
dependent on the degree of supersaturation (see fig. 3). If synthesis
conditions are chosen in such a way that always a high degree of
supersaturation can be maintained so that a high rate of nucleation is
established, the hydrothermal synthesis of nano-materials is possible.
Of course the synthesis success will strongly depend on the physical
and chemical properties of the nano-material to be synthesized as crystal
growth might occur in competition. Furthermore the re-dissolution of
already formed nuclei to build up microscopic crystals can not always be
neglected. However, the synthesis of crystals with a size just above the
thermodynamically critical limit for stability (in the nanometer-scale) is in
principle a process relatively difficult to maintain.
30
The experimental set-up for the hydrothermal synthesis is shown in
figure 4. The precursor materials are placed in containers with a solid to
water proportion of about 1:10. The closed containers are placed into the
high pressure vessel, which is then hermetically sealed and put into a
furnace. Applying the desired hydrothermal synthesis temperature, an
autogeneous pressure is formed. The external pressure is adjusted as soon
as the temperature equilibrium is achieved within the autoclave. By
installation of the hydrothermal pressure, the reaction process takes place.
Temperature gradients within the containers have to be avoided, as they
cause convection and therefore a fluid transport which favors in
consequence, the crystal growth after nucleation has taken place.
Temperature fluctuations of the furnace do have comparable negative
consequences because a rise in temperature leads to a higher dissolution
rate disturbing the dynamic equilibrium of dissolution-crystallization,
while a lowering of temperature leads to a higher supersaturation.
Dependent on the rate of crystal growth and, therefore, on the specific
material to be synthesized this may lead to an unwanted crystal growth
process instead of the desired nucleation only. Additionally a fluctuation of
temperature does also imply changes of hydrothermal pressure in the
standard experimental set-up, leading to a reinforcement of already present
temperature effects.
31
32
3. HIGH PRESSURE HYDROTHERMAL SYNTHESIS
(HPHS) OF NANOCRYSTALLINE MATERIALS
High pressure hydrothermal synthesis (HPHS) of nanocrystalline
materials has to be controlled by both, the process parameters temperature,
pressure and time, as well as the physical and the chemical parameters of
the materials system under investigation. The physical parameters in a
stationary hydrothermal system as represented in figure 4 have already
been outlined. The tailoring ofthe physical and chemical parameters of the
materials under investigation will have to focus on precursor–product
correlation using only pure water as pressure and reaction medium. The
effect of pH on the dissolution-re-crystallization process can not be taken
into account, because the concentration of is not a solitary parameter
but always correlated with the respective counter ion present. The presence
of such a counter ion, however, plays a nearly as important role in the
process, interacting either during the dissolution process or acting as a sort
of ionic template - or even as a constituent during the re-crystallization23
.
Added ionic solutions influence the synthesis process in a similar way.
While one sort of ions is directly implemented in the crystallization
process, the counter ion acts via the solution. In both cases the
concentrations of the present dissolved ions will not be constant during the
whole course of the experiment, and the change of this chemical parameter
will have effects on the synthesis sequence.
Therefore, the main emphasis is on the transformation of solid
precursor materials under the influence of HPHS conditions. In this
context, the chemical composition and state of aggregation of the
precursors are essential for further considerations.
3.1 GLASSY PRECURSOR TRANSFORMATION TO NANO-
MATERIALS
The high pressure hydrothermal transformation of glassy precursor
materials has been successfully employed for the systematic synthesis of
zeolites almost under the natural formation conditions24,25,26
. Here, the
vitreous state of aggregation is essential for the successful synthesis,
because the dissolution of the precursor material of identical composition
as the zeolite product is a homogeneous source for the zeolite constitution
at any time of the experiment. Under these conditions the formed zeolite
phase is thermodynamically more stable than the glassy precursor. As can
be seen in figures 2 and 3, nucleation and subsequent growth will only take
place in the region of supersaturated solution. Experiments have shown
that homogeneous zeolite nucleation from the supersaturated solution at a
chosen hydrothermal pressure 1 kbar is possible in a wide temperature
33
range (at least between 30°C and 400°C)27
. In contrast to this, further
experiments28
have demonstrated, that using glasses as precursors under
standard HPHS conditions, the temperature range of crystal growth is
restricted. Long term experiments during 60 days gave the confirmation,
that a limited quantity of glass precursor available in the closed container
as well as the high stability of the formed zeolite product against redissolution
enable the synthesis of nanocrystalline materials.
At lower temperatures microscopic crystals were synthesized, because
the dissolution rate of the glassy precursor is at a level to establish a low
grade of supersaturation where crystal nuclei grow continuously
consuming the dissolved material at the same rate. Having carried out this
experiment at different temperatures, it can be shown that ifthe dissolution
rate becomes too high for a continuous crystal growth, repeated nucleation
can be observed instead (see fig. 5). As can be seen, at 200°C the
maximum crystal size observed is about 500nm. Under the same
experimental conditions at 220°C only nanocrystalline material can be
observed, too small in size to be resolved by scanning electron microscopy.
The agglomerated crystal nuclei are stable even under long term
conditions, which means that once formed by consuming the glassy
precursor material, no change occurs. In the experiment no subsequent redissolution
did take place– as this would have lead to the synthesis of
34
microscopical observable crystals while all nanocrystalline material would
have been consumed. In the zeolite system under investigation, redissolution
of the once formed zeolite phase is only possible above 365°C
(same pressure-time conditions). Spherical agglomerates of microcrystals
can be formed by re-dissolution and crystal growth of already formed
nanocrystalline materials as shown in figure 6.
If a glassy precursor is available, certain nanocrystalline materials can
be produced by HPHS during long term experiments. However, limitations
of the method might be given by the process parameters themselves,
because the hydrothermal pressure limits the materials solubility. A too
low hydrothermal pressure results in a too slow dissolution kinetics, while
a very high water pressure results in a fast solubility kinetics29
, but may
favor the crystallization of unwanted high pressure phases30
. The size of
the glassy precursor pieces (e.g. if powdered or pieces in mm size),
agitation and temperature gradient induced convection are important
process parameters as well. Although zeolites are an extraordinary example
in this context, as their stability against re-dissolution is extreme, this
synthesis principle can also be applied to other nanocrystalline materials.
35
Using the standard isothermal-isobar set-up, the experiment is
conducted at different discrete temperature levels. In this mode a scan in
the ratio of the dissolution to re-crystallization kinetics is a very important
experimental process tool in order to obtain nanocrystalline materials
during long term high pressure hydrothermal synthesis by controlled
elimination of crystal growth. Figure 7 shows a SEM micrograph of
nanocrystalline Li-Fe-Silicate obtained by 28d HPHS at 1 kbar, while at
lower, but also at higher temperatures, respectively different
microcrystalline materials were synthesized.
3.2 CONTROLLED RECRYSTALLIZATION OF NANO-
MATERIALS
Nanocrystallinity is present if no or only very minor crystal growth
has occured. Some inorganic materials like the class of hexacyanometallates
with transition metal ions as structural elements show only a
very reduced solubility in aqueous solutions under ambient pressures31
.
Among them the well known Prussian blue, is the most
investigated of mixed valence state materials of this type32
. Its extremely
low solubility implements that under wet chemical synthesis conditions
instantaneously colloidal agglomerates are formed, which are unable to
36
grow. Our experiments have shown that Prussian blue submitted to high
pressure hydrothermal conditions for up to 52 days at pressures between
500bar and 1kbar and temperatures between 40°C and 100°C is not
decomposed but re-crystallized. In figure 8 a SEM-micrograph of a typical
result of a re-crystallization experiment is shown.
The experimental results shown in figure 8 are explainable taking into
account the processes described in figure 2: even under HPHS conditions
the smallest structural building units of Prussian blue are not attacked.
However, the solubility is significantly higher than under ambient
conditions and, most important, the region of supersaturated fluid is
widened. This extension of the region of supersaturated fluid is the reason
for a growth of the Prussian blue nuclei, where some of the nuclei grow
while most of them are consumed to deliver the material for the growth
process. Finally, their size rests in the nanometer range, but it could be
shown in principle that the crystal growth of hexacyano-metallate
nanomaterials is possible.
37
3.3 SHORT TERM PRECURSOR TRANSFORMATION TO
NANO-MATERIALS
For reasons ofthe extreme synthesis conditions like the effect that all
product materials formed had to pass the dissolution re-crystallization
pathway, the hydrothermally prepared materials are formed near
equilibrium conditions. To keep the product particle size in the nanometer
range, all precursor reaction processes have to occur very rapidly, so that
the whole process is finished within several hours in order to avoid a
significant growth of the product particles. Different nanocrystalline
materials have been obtained in this way. Nanocrystalline zirconia (7-20nm
size) and iron oxide powders (55-80nm) already obtained by RTDS and
RESS are also obtainable by hydrothermal treatment of nitrate and oxide
precursors at 240°C, up to 6h at autogeneous pressure15
. Different other
nanocystalline oxides and oxy-hydroxides can be obtained in the same way
by reaction and nucleation of nitrate precursors using water as pressure and
reaction medium. Among them are AlOOH with particle sizes of 100-
600nm33
, of about 50nm particle size34,
, NiO of l00nm particle
size35
, with spherical particle size of 50nm35
and of particle
sizes up to 300nm36
. Nitrate free AlOOH nanocrystals were also obtainable
using nitrate precursors under “true” hydrothermal conditions (400°C,
350bar, 30min)37
. Other simple oxides have been obtained by the same
authors33,34,35
in an analogous way using non-oxide precursors. One can
mention of about 20nm size applying and precursors,
of spherical particle size of 10nm from precursors and
of up to 50nm sizeemploying or as starting materials. The
production of fine powdered oxides in the nanometer scale was also
reported by hydrothermal oxidation of the respected coarse grained
metals38
. Nanocrystalline of up to 100nm size was obtained using
tetraalkylammonium hydroxide peptized sols as precursors with water as
reaction medium for 2 hours under autogeneous pressures at 240°C39
.
Particles of an average diameter between 3 and 4nm could be obtained by
hydrothermal reaction between a gel and a hydroperoxide (180°C
autogenous pressure, 24h)40
. Our own investigations showed that the short
term high pressure hydrothermal treatment of at 500bar lead to a
nanocrystalline CrOOH product in a temperature range between 290°C and
450°C (maximum temperature of investigation). The higher the synthesis
temperature, the bigger was the average grain size.
Reports were also given that a few nanocrystalline oxides can be
obtained in a relatively simple way by treating the coarse grained oxide
precursors for a short time. Nanocrystalline Cr2O3
41,42
and HfO2
43,44
have
been obtained in this way.
38
Different methods to obtain more complex oxides by short term high
pressure hydrothermal synthesis using water (as pressure and reaction
medium) have been described. Nanocrystalline could be
synthesized using microcrystalline precursors of the respective
composition41,42
and yttria stabilized of 6-22nm particle size was
obtained from a precipitated yttria-zirconia hydroxide-hydrated precursor
by a simultaneous dissolution-reaction-crystallization process45
.
Preparation of fine barium-ferrite particle was reported46
decomposing
the just blended precursor materials hydrothermally. Hydrothermal
methods could also be applied in order to control the size and shape of lead
titanate47
. Different approaches are known to synthesize nanomaterials
by hydrothermal methods. It could be obtained by short term
(0.5-2h) autogenous pressure treatment of the mixed precursor solutions
using additives in order to obtain a homogenous product of about l00nm
size48
. Using stoichiometric mixtures of hydrated Barium-hydroxide
precursors and different Ti-sources, gives almost the same results49
.
Microwave assisted very short term hydrothermal synthesis of
nanocrystalline at 200°C (5-30min) was also reported using barium
nitrate, titanium tetrabutoxide and KOH as precursors50
. The results
confirmed that in principle a quantitative synthesis can be achieved within
30min if the appropriate precursor materials are chosen. with
different particle sizes from 50-100nm has been hydrothermally prepared
39
from different Sr and Ti precursors between 150 and 240°C at autogeneous
pressure51
. An influence of the precursor and additional solutions on the
particle size has been found. Even nanocrystalline phases of have
been synthesized from organic precursors at 170°C and autogeneous
pressure (2h)52
. Differences in homogenity of particle size and phases
present have been found between batches using just water and batches with
added acid.
These few examples might show the great potential of the process; a
more consequent and systematic application of the synthesis principles
accompanied with their refinement will lead to a further rapid extension of
the method.
4. CONCLUSIONS
As the different examples and principal synthesis strategies might
have shown, hydrothermal synthesis is an elegant and straightforward
method with a huge potential for the solution of future synthesis tasks. For
nano-materials synthesis, one principal problem has to be solved: once
obtained, how to prevent nuclei growth. The physical parameters water
pressure, temperature and time and the respective precursor-product system
have to be tuned in such a way that always a high degree of supersaturation
can be maintained so that a high rate of nucleation is established simultaneously
avoiding crystal growth.
Hydrothermal synthesis ofinorganic materials is a technique with a
relatively simple basic mechanism. Although the physical process
parameters are comparably easy to control, less is known about the
chemical and physical processes taking place in detail. More information
has to be gathered about size and constitution of the smallest structural
units transported in the aqueous fluid. The process kinetics associated with
the mechanisms of reaction - specific for different kinds of precursor
materials during the dissolution-re-crystallization process - play thereby a
key role. The aim will be a modeling of the parameters in order to make
predictions about the course of the experiment.
5. ACKNOWLEDGMENT
The authors gratefully acknowledge the support from the French
Centre National de la Recherche Scientifique, CNRS.
40
6. REFERENCES
1
H. Ghobarkar, O. Schäf and U. Guth, Prog. Solid St. Chem. 27(2-4), 29-73
(1999).
2
M.S. Whittingham, J.D. Guo, R. Chen, T. Chirayil, G. Janauer, P. Zavalij, Solid
State Ionics 75, 257-268 (1995).
3
F. Cansell, B. Chevalier, A. Demourgues, J. Etourneau, C. Even, Y. Garrabos, V.
Pessey, S. Petit, A. Tressaud and F. Weill, J. Mater. Chem. 8 (1998).
4
G. Demazeau, C.R. Acad. Sci. Paris, t. 2, Série IIc, 685-692 (1999).
5
K. Byrappa, I. Yoshimura, M. Yoshimura (eds.), Handbook of Hydrothermal
Technology, William Andrew Publishing (2001).
6
G. Demazeau, J. Mat. Chem. 9, 15 (1999).
7
M.M. Hoffmann, J.G. Darab, B.J. Palmer, J.L. Fulton, J. Phys. Chem. A 42,
8471-8482 (1999).
8
J.L. Fulton, D.M. Pfund, S.L. Wallen, M. Newville, E.A. Stern, Y. Ma, J. Chem.
Phys. 105,2161-2166(1996).
9
S.L. Wallen, B.J. Palmer, D.M. Pfund, J.L. Fulton, M. Newville, Y.Ma, E.A.
Stern, J. Phys. Chem. A 101, 9632-9640 (1997).
10
J. Jung, M. Perrut, J. Supercrit. Fluids 20, 179-219 (2001).
11
D.W. Matson, R.C. Petersen, R.D. Smith, Adv. Ceramic Mat. 1, 242-246 (1986).
12
R.D. Smith, R. Wash, U.S. Patent #4,582,731 (1986).
13
D.W. Matson, R.D. Smith, J. Am. Ceram. Soc. 72,871-881 (1989).
14
D.W. Matson, J.L. Fulton, R.C. Petersen, R.D. Smith, Indust. Eng. Chem. Res.
26, 2298-2306 (1987).
15
A.A. Burukhin, B.R. Churagulov, N.N. Oleynikov, Y.V. Kolen'ko, MRS
Symposium Proceedings 520, 171.
16
A.A. Burukhin, B.R. Churagulov, N.N. Oleynikov, High Press. Res. 20, 255-264
(2001).
17
D.W. Matson, J.C. Linehan, R.M. Bean, Mater. Lett. 14(4), 222-226 (1992).
18
J.G. Darab, D.W. Matson, J. Electron. Mater. 27(10), 1068-1072 (1998).
19
B.L. Armstrong, Mater. Manuf. Process 11(6), 999-1012 (1996).
20
D.W. Matson, J.L. Fulton, J.C. Linehan, R.M. Bean, T.A. Werpy, J.G. Darab,
U.S. Patent #5,652,192 (1998).
21
A.A. Burukhin, B.R. Churagulov, N.N. Oleynikov, High Press. Res. 20, 255-264
(2001).
22
G. Demazeau, J.M. Mollet, C. Cros, A. Largeteau, J. Alloys and Compounds
262-263, 271-274 (1997).
23
H.Ghobarkar, O. Schäf, P. Knauth, Angew. Chemie 113(20), 3948-3951 (2001);
Int. Ed. 40(20), 3831-3833 (2001).
24
H. Ghobarkar, O. Schäf and U. Guth, High Pressure Res. 20(1-6), 45-54 (2001).
25
H. Ghobarkar, O. Schäf and P. Knauth, Ann. Chim., Science des Matériaux 24,
209-215(1999).
26
H. Ghobarkar, O. Schäf, Verfahren zur Synthese von Zeolithen mit definierter
Stöchiometrie und Struktur. German Patent, AZ 198 24 184.4-41, disclosed 2. 12.
1999.
27
H. Ghobarkar, O. Schäf, Mic. Mes. Mats. 23(106), 55-60 (1998).
28
H. Ghobarkar, O. Schäf, unpublishedresults (2001).
41
29
K.Th. Wilke, Kristallzüchtung, VEB Deutscher Verlag der Wissenschaften,
Berlin (1973).
30
W. Eitel, Silicate Science Vol. IV: Hydrothermal Silicate Systems, Academic
Press (1966).
31
A.G. Shape, The Chemistry ofCyano Complexes ofthe Transition Metals,
Academic Press, London (1976).
32
M. Verdaguer, Polyhedron 20, 1115-1128 (2001).
33
T. Adshiri, K. Kanazawa, K. Arai, J. A. Ceram. Soc. 75(4), 1019-1022 (1992).
34
T. Adshiri, K. Kanazawa, K. Arai, J. A. Ceram. Soc. 75(9), 2615-2618 (1992).
35
T. Adshiri, S. Yamane, S. Onai, K. Arai, Proc. 3rd
Int. Symp. on Supercritical
Fluids, Strasbourg, France, Oct. 17-19, G. Brunner, M. Perrut (eds.), Tome 3, 241-
247(1994).
36
Y. Hakuta, H. Terayama, S. Onai, T. Adshiri, K. Arai, 4th
Int. Symp. on
Supercritical Fluids, Sendai, Japan, May 11-14, 255-258 (1997).
37
G. Li, R.L. Smith, H. Inomata, K. Arai, Materials Letters 53, 175-179 (2002).
38
M. Yoshimura, S. Somiya, Materials Chemistry and Physics 61, 1-8 (1999).
39
J. Yang, S. Mei, J.M.F. Ferreira, Mat. Sci. Eng. C 15, 183-185 (2001).
40
Y. Jiang, Y. Wu, B. Xie, Y. Xie, Y. Qian, Materials Chemistry and Physics 74,
234-237 (2002).
41
M. Yoshimura, S . Somiya, Am. Ceram. Soc. Bull. 59(2), 246 (1980).
42
S. Hirano, S. Somiya, J. Am. Ceram. Soc. 59(11-12), 534 (1976).
43
H. Toraya, M. Yoshimura, S. Somiya, Comm. Am. Ceram. Soc. 65(5), C-72
(1982).
44
H. Toraya, M. Yoshimura, S. Somiya, Comm. Am. Ceram. Soc. 65(9), C-159-
160(1982).
45
R.R. Piticescu, C. Monty, D. Taloi, A. Motoc, S. Axinte, J. Europ. Ceram. Soc.
21, 2057-2060 (2001).
46
K. Arai, T. Ajiri, S. Yuki, I. Ota, U.S. Patent #5,433,878, (1995).
47
J. Moon, M.L. Carasso, J.H. Adair,H.G. Krarup, J.A. Kerchner, J. Mater. Res.
14(3) 866 (1999).
48
S.W. Lu , B.I. Lee, Z.L. Wang, W.D. Samuels, J. Crystal Growth, 219, 269-276
(2000).
49
C.T. Xia, E.W. Shi, W.Z. Zhong, J.K. Guo, J. Crystal Growth 166, 961-966
(1996).
50
Y.B. Khollam, A.S. Deshpande, A.J. Patil, H.S. Potdar, S.B. Deshpande, S.K.
Date, Materials Chemistry and Physics, 71, 304-308 (2001).
51
C.C. Chen, X. Jiao, D. Chen Y. Zhao, Mat. Res. Bull. 36, 2119-2126 (2001).
52
I. Maclaren, P.A. Trusty, C.B. Ponton, Acta Mater. 47(3), 779-791 (1999).
EHDA IN PARTICLE PRODUCTION
Application of electrospray in nanoparticle production
T. Ciach, K.B. Geerse and J.C.M. Marijnissen
Particle Technology Group, Delft University of Technology, Delft, The
Netherlands
This chapter describes a route to nanoparticle production via
Electro-Hydro-Dynamic Atomization (EHDA). This process, also referred
to as electrospraying, is a process where a liquid jet breaks up into droplets
under the influence ofelectrical forces. EHDA as a method to produce
particles has a number of advantages. There is a great flexibility concerning
the particle composition, because precursors can be found for almost
any material. In contrast to particle generators applying high temperatures
and/or plasma, it is a gentle method in the sense that even easily decomposing
compounds remain stable. Even particles of biochemical molecules
such as insulin or peptides have been successfully produced. Finally, the
method has the potential of producing very narrow size distributions with
a simple way of controlling particle size. While this principle has frequently
been used for particle production, it has seldom been applied in
the nano-particle size range, but it could be seen here as one of the most
promising methods.
EHDA has formerly been used for ceramic particle production in
the micron range [1]. More recent research has shown that preparation of
nano-particles down to a few tens of nanometers is also possible using this
method [2,3]. By electrospraying and pyrolysis of a and
solution in ethanol, 20 nm ZnS particles were obtained.
INTRODUCTION
ELECTRO HYDRO DYNAMIC ATOMIZATION
The possibility of using electric fields to disperse liquids in order
to produce particles has attracted much interest in the scientific world. The
phenomenon of an electric effect on menisci is known since the sixteenth
century. William Gilbert [4] reported the interaction between a water
droplet and a piece of amber, leading to a conical shaped droplet. Zeleny
[5] showed in the beginning of this century that liquid meniscus subjected
44
to a high enough electric field, adapt a conical shape and emit a mist of
very small droplets. In the fifties, Vonnegut and Neubauer [6] tried to relate
the properties ofa liquid to its behaviour in electrospray.
Electrohydrodynamic atomisation refers to a process, where a liquid
jet breaks up into droplets under the influence of electrical forces. Depending
on the strength of the electric stresses in the liquid surface relative
to the surface tension stress and the kinetic energy of the liquid jet
leaving the nozzle, different spraying modes will be obtained. The modes
are classified into two general categories: those that exhibit a continuous
flow of liquid through the meniscus and those that do not. The former
consists ofthe simple-jet, the cone-jet and the ramified-jet, while the latter
consists of the dripping, the micro-dripping, the spindle and the intermittent
cone-jet modes. The latter are often referred to as pulsating modes.
shows the different modes graphically. Most authors have observed both
categories although the nomenclature has not been universally accepted.
For particle production, the cone-jet mode shows the best performance. In
this state the process is very stable and small, monodisperse particles are
formed. Figure 2 shows a photograph ofthis cone jet mode.
45
Taylor made the first mathematical description of the cone by balancing
the forces that are present figure 3.
The Particle Technology Section of Delft University of Technology [8]
developed a physical numerical model of EHDA in the cone jet mode.
The measurement results showed that monodispersity of the droplet produced
in the cone depends on the jet break-up mechanism, which was
found to depend on the ratio of the electric stress over the surface tension
stress. At low stress ratios, the jet breaks up due to varicose perturbations.
At higher stress ratios, the jet showed a whipping motion. Figure 4 shows
the break up in these two stress regimes.
For seven different liquids, the relation between varicose and whipping
breaks up and the stress ratio within the jet was investigated. This stress
ratio is given by:
46
where is the surface charge on the jet, is the surface tension of the
liquid considered, is the jet diameter at its breakup and is the
permittivity of vacuum.
Figure 5 shows the break up mechanisms as function of this stress
ratio for seven liquids with different conductivities, varying from 13 to 45
microsiemens per meter. The stress ratio for one liquid was changed by
changing the flow rate ofthe liquid.
To be in the varicose break up regime the stress ratio should be below 0.2.
In the physical model of Hartman [8], the influence of spraying
parameters on the cone shape, droplet size and droplet charge is investigated.
The shape of the liquid cone is calculated by solving the NavierStokes
equation in one dimension for a steady state situation. To be able
to solve these equations, the electric field strenghts, the surface charge and
the radial velocity profile inside the liquid have to be known. If a certain
cone shape is assumed, then the electric fields in and outside the liquid are
numerically calculated using Gauss' law. The surface charge follows then
from the current balance at the liquid-air interface and from the estimation
of the liquid velocity at the liquid-air interface. The one-dimensional
Navier-Stokes equation is then used to calculate a new cone-shape, which
is used as input for new electric field, and surface charge calculations.
This process is repeated until the input and output cone shapes have converged.
The break-up of the jet is calculated numerically by using models
of Weber and Melcher [9,10]. The calculations use the previously calcu-
47
lated jet shape. These models yield the jet diameter before break up, the
diameter of the combined main and secondary droplet and the droplet ve-
locity.
In the model of Hartman, the current through a cone jet system was modelled.
For the high conductivity case he found for the current:
where Q is the flow rate applied, I is the current through the jet, K and
are respectively the conductivity and the density of the liquid. The variables
a and b are scaling constants.
For a nozzle with a diameter of 8 mm and for n-butanol and ethylene glycol
with various conductivities, the scaling law was fitted to current measurements.
The result is as follows:
It has to be noted that the current also depends on the electrode configuration,
and the type of ions used. So, measurements might yield deviations
from the scaling laws up to 30%.
For the refining of the droplet size scaling law one of the tools
Hartman used was a High Speed Imaging System. Hartman found the
following expressions for the droplet size in the varicose break up regime:
48
The constant c was fitted to the experiments and was found to be equal to
2.05 for the measurements with the camera.
The process of particle formation by EHDA was extensively
studied, both fundamentally and experimentally at TU Delft. In particular,
the group has a lot of experience in using EHDA for ceramic and organic
microparticles production [1]. Figure 6 shows an example.
The particles shown were made in the setup shown in figure 7.
The spraying nozzle operating in the cone-jet mode is fixed in the aerosol
reactor. Liquid is supplied via a syringe pump. A ring electrode is placed
below the nozzle. The voltage on this electrode is lower then the voltage
on the nozzle. This electrode is used to stabilize the EHDA process. On
the bottom of the reactor, there is a grounded tip, on which a corona discharge
takes place. Ions of a polarity opposite to the droplet charge are
produced this way. They neutralize the droplets, to reduce the loss due to
the high space charge.
Sprayed particles dry in the air and then pass the heater where
thermal decomposition takes place. As a precursor for particles
solution in absolute ethanol was used. The temperature of
the heater was 400 °C. The chemical reaction of thermal decomposition of
this precursor is shown below:
49
The use ofchlorine containing precursor always leaves some impurities in
the product.
BIPOLAR COAGULATION
The bipolar coagulation process consists of the mixing of two electrospray
droplets of opposite charges. The principle of this process is
shown in the figure 8. These two electrosprays are produced by EHDA of
liquids. Because electrical attraction forces will dominate the coagulation
and because it is possible to control the size and charge of droplets produced
by EHDA, it should be possible to control the coagulation of droplets.
The potential of such a process is enormous. It should enable to mix
products on a micro-scale and to make chemical reactions in a microenvironment.
This process would gives a new degree of freedom to the classical
spray processes.
50
An example of the bipolar coagulation is the method of production
of particles. The powder was produced in the large configuration
ofthe reactor (long residence time), with a titanium solution in anhydrous
butanol and a water solution in absolute butanol. The temperature of
the tubular furnace was 400°C. The particles were measured online with
the aerosizer. The geometric mean diameter of the particles was 5.42 micron
and the geometric standard deviation was 1.13. The particles on the
photos appear to be smaller than 5.4 micron. They have a size around 2.5
micron. This discrepancy is due to the unknown density of these particles.
The geometric mean diameter mentioned above is given by the aerosizer.
To obtain this, a density of the particle measured must be entered into the
software. The density of the produced particles is unknown so the used
density was one. It gave the geometric mean diameter of 5.42 microns.
Another example of particle that shows intermediate stage of bipolar coagulation
is shown in the figure 10.
EHDA can be also applied in the production of nanoparticles. In
the foregoing we have seen that EHDA is able to yield monosized particles.
To apply this for the production of nano-particles the starting conditions
must be such that the size ofthe produced droplets, together with the
concentration of the original liquid gives the required particle size. For
mono-dispersity the break-up of the jet should take place in the varicose
breaking-up regime. To satisfy the size requirements we have to make the
flow rate, Q, as small as possible and at the same time the conductivity, K,
rather big. In the other parameters there is not much flexibility. Care
should be taken to choose a solvent with a relatively low surface tension,
e.g. ethanol, since for liquids with a high surface tension like for water the
field, required to establish the cone-jet mode is so high that sparking oc-
curs.
To satisfy the second requirement, monodispersity, the stress ratio
should be less than about 0.3. As an example we have produced NaCl
nanoparticles from a solution (fig. 11.).
It was shown that nanoparticles production by the method is feasible
[11]. Particles of sodium chloride of about 5nm diameter were produced
using the EHDA principle. These particles were obtained by
spraying a mixture of water, ethylene glycol and sodium chloride. For
producing these ultra fine particles, the liquid flow had to be kept low and
precisely controlled. The conductivity of the solution should be in the
proper range.
Organic and inorganic nanoparticles can be produced this way. As
a precursor for the particles liquid or dissolvable compounds containing
the desired elements can be atomised first in the EHDA reactor. After
evaporation of the solvent, nanoparticles are formed. Particles will get
their desired chemical composition after oxidation, hydrolysis or thermal
51
APPLICATION IN NANOPARTICLE PRODUCTION
decomposition of the precursors. Examples of the chemical reactions
during those processes are:
For the production of chemically complex nano-particles, a major
advantage of EHDA can be used, namely the possibility of producing
mixtures of well-defined composition. Mixed oxides, metal-oxide and
oxide-coated metal nano-particles can be made.
For the theoretical analysis of production of the nanoparticles currently
available semi-empirical models of the EHDA process must be
evaluated and eventually modified, since they have been developed for
processes forming particles in the micron range. In close interaction with
the experimental activity, it is also a goal of current research to find approaches
to modeling of nano-particle formation. Theoretical modeling
will be also needed to describe particle formation during thermal decomposition.
This is especially important in the case of coated particle production.
If two precursors decomposing at different temperatures are present,
we expect that a solid core be formed. If the second component is liquid
at some stage, it may wet the solid core, forming a coated particle. The
desired composition is established, after the reactions are completed. The
wetting process strongly depends on the surface energies involved. To
find the proper conditions, the composition of the chemicals applied and
the temperature history has to be carefully controlled.
Spray pyrolysis has already been used in thin film production. In
particular, the method provides a high enough purity to produce thin film
gas sensors [11]. The technique is liable to open new perspectives in the
particulate film domain, because it contains the additional features of producing
particles of uniform and variable size down to the regime of a few
nanometers. Thin films are used in solar cells, semiconductors, superconductors,
supermagnetic cores and as various coating materials, e.g. in
smart glasses. Thin particle films of semiconductors change their electrical
resistance due to presence of adsorbed gases and are used
in gas sensors.
Nano-particles obtained via the EHDA route can find a big variety
of other applications in electronics, optics and medicine [12]. They can be
used for producing “quantum dots” e.g. for high efficiency tunable LED,
low noise semiconductor lasers or high capacity optoelectronic memories.
Possible applications also include manufacturing of solar cells. Currently
production of this type of energy sources is very polluting and employing
it as a way of large-scale energy production will not be environmentally
52
friendly. New materials as well as production routs are searched for. Titanium
dioxide is a very promising compound in this field.
Coated nanoparticles of iron oxide are a promising material for
very high frequency magnetic cores due to their superparamagnetic behavior.
They can also be used in magnetic refrigeration technology based
on the magnetocaloric effect. In both cases magnetic particles need to be
coated. Anotherapplicationofchemically morecomplicated nanoparticles
is in the production of high temperature superconductors. Methods based
on spraying have been proven successful in producing chemically complex
ceramic nanoparticles as required for superconductor production.
EHDA could open new perspectives here, again due to the feature of uniform
and variable size downto a few nanometers.
Finally, nanoparticles in medicine already represent a market of
many billions ofEuros. Nanoparticles extend the range of materials applicable
as medicine to compounds that are almost insoluable in water.
EHDA could enter here as a flexible method of producing particles of
fragile biochemical compounds. Targeted drugs based on special coated
particle designs could also be produced by this method.
53
REFERENCES
Vercoulen P.H.W., Camelot D.M.A., Marijnissen J.C.M., Pratsinis S. and Scarlett B.
SnO2 production by an electrostatic spray pyrolysis process. Synthesis and measurement
of ultrafine particles. Proceedings of the International Workshop on the
Synthesis of Ultrafine Particles; 1993; Delft University Press.
Dudout B., Marijnissen J.C.M. and Scarlett B. Use of EHDA for the production of
nanoparticles. J. of Aerosol Sci. 1999; 30:687-688
Lenggoro W., Okuyama K., De la Mora J.F. Preparation of ZnS nanoparticles by
electrospray pyrolysis. S. ofAerosol Sci. 2000; 31:121-136
Gilbert W. De magnete, magneticisque corporibus, et de magno magnete tellure;
physiologia nova, plurimis argumentis, experimentis demonstrata London, 1600.
Zeleny J. Instability ofelectrified liquid surfaces. Phys. Rev. 1917; 10:1-16
Vonnegut B. and Neubauer R.L. Production of monodisperse liquid particles by
electrical atomization. J. Colloid Sci. 1952; 7:616-622
Cloupeau M. and Prunet-Foch B. Electrohydrodynamic spraying functioning modes:
A crtitical review. J. Aerosol Sci. 1994; 25:1021-1036
Hartman R.P.A., Brunner D.J., Camelot D.M.A., Marijnissen J.C.M. and Scarlett B.
Electrohydronamic atomization in the cone-jet mode: physical modeling of the liquid
cone and jet. J. ofAerosol Sci. 1999; 30:823-849
Weber C. Zum Zerfall eines Flüssigkeitsstrahles. Z. für angew. Math, und Mech.
1931; 11:136
Melcher J.R. Field-coupled surface waves. Cambridge, Mass. M.I.T. 1963.
Korotchenkov G., DiBattista M., Schwank J. and Brinzari V. Structural characterization
of SnO2 gas sensing films deposited by spray pyrolysis. Mat. Sci. and Eng. B
2000; 77-1:33-39
Kruis F.E., Fissan H. and Peled A. Synthesis of nanoparticles in the gas phase for
electronic, optical and magnetic applications – a review. J. of Aerosol Sci. 1998;
29:511-535
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
DYNAMIC COMPACTION OF
NANO-STUCTURED CERAMICS
Michiel J.G. Jak
Delft Institute for Sustainable Energy, Laboratory for Inorganic Chemistry,
Julianalaan 136 2628 BL, Delft University of Technology, The
Netherlands
1: Introduction
The densification methods ofceramic materials are the same for many
years now. Traditionally the densification takes place in a two-step process:
shaping of the green body by using static pressing and subsequent sintering.
This well-known and widely spread technique results in high densities and large
samples with different geometries can be densified. High temperatures and long
sintering times are necessary. The sintering time can be reduced by using a high
starting density and by applying an external pressure.
However, when high temperatures are used in order to densify nanostructured
materials, grain growth will occur and the material can even
decompose, react, or melt and lose the optimum properties they were selected
for. Therefore, densification of nano-structured ceramics by using high
temperatures should be applied with care.
As an alternative for traditional densification using high temperatures,
this chapter describes the use of high-pressure techniques. High-pressure
techniques can be divided into static and dynamic techniques. With static
compaction a constant pressure is applied onto a sample for a certain period of
time. Due to the repulsive forces between the nano-sized particles and surface
contaminations that hinder intimate contact between the particles, static
pressing is not sufficient in many cases. Dynamic compaction uses a pressure
pulse with a certain pulse duration, resulting in a pressure wave that travels
through the sample. In both cases the pressure can by applied in a uniaxial,
biaxial, radial, or isostatic/isodynamic mode. The basic principle of dynamic
compaction is the way pressure waves, necessary to densify the sample, are
generated. One of the best-known methods is by using explosives, but
compaction by magnetic pulses is also used. In the following sections these two
methods, i.e., Explosive Compaction (EXC) and Magnetic Pulse Compaction
(MPC) are discussed. Both methods can be used for the shaping ofgreen bodies
but also final products can be made in a single step process.
56
2: The difference between dynamic and static compaction
The main difference between dynamic and static compaction is the
pressure wave that is generated with dynamic compaction. The difference lies in
the distribution and nature ofthe stresses that cause compaction [1]. Due to the
pressure wave velocity, the particles of the powder to be compacted in the
pressure wave front obtain a certain velocity. Hence, the impulse of the
particles during dynamic compaction is much higher compared with static
compaction resulting in higher local pressures that cause more plastic
deformation and fracture of the particles. Due to the movement of the particles
and the heat generation by friction the surface contaminants are removed and
fresh surface areas are available. The high local pressure may result in particle
attrition as well leading to new, clean surfaces. These processes, called dynamic
cleaning, result in very sinter active powders.
Various methods to dynamically compact powdered materials are
classified and reviewed by Clyens and Johnson [2]. A few basic principles hold
for all these different methods. The consolidation of powder materials by
dynamic compaction can be divided into three stages [2], as shown in Fig. 1.
Stage I: Transitional restacking: powder particles are flowing along one
another and repacking. During this stage, the particle surface
instantaneously reaches extremely high temperatures due to interparticle
friction effects. However, the temperature gradient between the hot surface
and the cold center is huge and the cooling rate can be up to one million
degrees per second [3]. Due to the very high heating and cooling rate, no
diffusion or segregation of material is possible. The interparticle bonding
occurs when the interparticle loads become sufficiently high to break up
possible present surface contaminants and to bring the particles in intimate
contact.
57
Stage II: Plastic deformation: transitional restacking ceases when the
compaction pressure reaches a value such that the particles are locked.
Depending on the powder being compacted, transitional restacking is followed
by the compaction stage in which particles deform plastically to fill the
porosities in the powder.
Stage III: Cold working: in this final stage cold working (fixation of
dislocations which increases the hardness of the material) and/or particle
attrition takes place.
At very high pressures a fourth stage may occur that comprises
phase transformation. The occurrence of this fourth stage depends on the
pressure and the stability ofthe crystal structure [4].
Figs. 2a-d demonstrates the three stages for a Fe powder sample dynamically
compacted with different pressures [5]. It must be emphasized that stages II and
III not necessarily take place and stage II is of minor importance for ceramic
materials.
During static compression external stresses are applied which cause
particles to move to a denser particle stacking. After this initial compaction
(stage I) the density is further increased by fracture and plastic deformation of
the particles at contact areas (stages II and III). In the case of dynamic
58
compaction, the externally applied pressure travels through the powder as an
anisotropic pressure wave with very large stress gradients. These anisotropic
stresses, however, do not solely result in deformation near the contact areas, but
also force other particles, which are not in direct contact, to move toward each
other. Furthermore, due to the anisotropic nature of a pressure wave and the
large stress gradients in the pressure wave, the amount of plastic deformation
during dynamic compaction is very large compared to static pressing [1,6]. The
same holds true for explosive compaction, but instead of a pressure wave a
shock wave (with a higher velocity) is generated, leading to higher particle
velocities and stress gradients. In both cases an additional pressure effect is
caused by the impulse of the particles.
It should be stressed that homogeneous and high densities with a
minimal crack formation can only be accomplished by compacting samples
with a high starting density. Furthermore, the temperature increase of the
sample is smaller when using a high starting density due to lower heat
generation caused by friction forces.
The general pressure ranges and pressure wave velocities for four
different compaction techniques are shown in Fig. 3. Dynamically: Explosive
Compaction (EXC) and Magnetic Pulse Compaction (MPC), statically: Cold
Uniaxial Pressing (CUP) and Hot Uniaxial Pressing (HUP).
3: Explosive Compaction (EXC)
Explosive compaction [for instance 1-3, 7-9] uses explosives or
exploding gases for the generation of a pressure wave with high velocity (shock
wave). In this way high pressures can be obtained and large samples can be
compacted. The pressure wave is a shock wave with a typical velocity of a few
59
kilometers per second. The detonation velocity depends on both the chemical
structure of the explosive used and the density of the explosives. Higher
densities result in higher detonation velocities [10]. Basically, two explosive
compaction configurations are possible, i.e., cylindrical and flat (uniaxial). The
cylindrical configuration, also referred to as explosive isodynamic compaction,
will be described in more detail using Fig. 4.
Compaction is achieved due to the pressure caused by the detonation
(A) ofthe explosives along the inner tube (B), and has a radial direction toward
the center of the tube. The compacted powder region is marked (C) and the
material to be compacted (D). Samples with a diameter up to 15 cm can be
compacted and the length ofthe sample is in principle unlimited, but depends on
the capacity of the bunker used.
Typical dimensions ofsamples compacted using the cylindrical setup are a tube
diameter of0.5 to 5 cm (maximum of 15 cm) and a length of 5 to 25 cm made
ofaluminum or stainless steel.
After compaction the metal tubes including the compact are normally
heat-treated for a few hours at about 200°C. This heat treatment decreases
present internal stresses, which are introduced due to the dynamic compaction
process.
The second method of compaction is the flat configuration, also referred
to as explosive uniaxial compaction [9]. The experimental setup is shown in
Fig. 5. Compaction is achieved by placing a metal plate, the so-called target
plate, on the powder and the explosives placed on top ofthe target plate.
A metal tube is filled with the powder to be compacted The tube is
closed using metal plugs to avoid loss of powder during the compaction
and protection ofthe powder against the detonation products. This metal tube is
centered using disks in another but larger container This container or
envelope is only used to hold the explosives and, therefore, can be made of
60
any material. However, metals are to be preferred as they lead to higher
pressures during compaction. The explosives are ignited by a detonator
placed on top ofthe container.
The advantage of the cylindrical compaction method is that a
homogeneous compaction can be obtained, because absorption of the shock
wave and enhancement of the pressure takes place at the same time. It should be
borne in mind that the shock wave travels from the tube wall to the center of the
powder, and that enhancement of the pressure occurs due to a reduction ofthe
surface area on which it acts. This balancing can be more difficult using the flat
compaction configuration due to reflection of shock waves from the die.
4: Magnetic Pulse Compaction (MPC)
Another elegant way of generating a pressure wave is by using a
magnetic pulse that accelerates a metal target plate or target tube [for instance
5, 8, 11-15]. A mechanical pressure pulse is obtained by conversion of
electrical energy into a magnetic field.
The magnetic field is generated by charging a capacitor bank (typical maximum
energy storage about 100 kJ) to a preset voltage of typically 1 to 10 kV. The
pressure acting on the powder is approximately linearly proportional to the
preset voltage. Via a spark gap switch the stored energy is discharged over an
induction coil that generates a magnetic field. An electronically conducting
material in this magnetic field will pick up this magnetic field and will move in
the direction of the Lorentz force. In the case of a spirally wound induction coil
the Lorentz force is parallel with the coil's center. As a result, a metal target
plate in front of the induction coil is repelled. Consequently, the target plate will
61
interact with a piston which is pushed onto the powder to be compacted (Fig.
6).
The Magnetic Pulse Compaction (MPC) method is easy to control and
safe. The powders can be compacted in a controlled atmosphere including
vacuum and also (pre)heating is possible. Typical pressure wave velocities are
10-200 m/s and pressures up to 5 GPa, but higher pressures and target plate
velocities are possible when proper molds are used.
A cylindrical configuration is possible as well. In this case the to be
compacted sample is placed in a metal tube that is placed in the center of a
spirally wound induction coil. By discharging the charged capacitors over the
spirally wound induction coil a magnetic field is generated. Due to the fast
change in magnetic field strength, eddy currents are generated in the metal tube.
These eddy currents generate an induced Lorentz force that acts in the opposite
direction of its origin (Lenz's law) perpendicular to the tube surface. This
induced Lorentz force results in the radial compaction of the metal tube and its
contents. For a successful compaction the following should be achieved:
1)
2)
3)
4)
5)
6)
homogeneous magnetic field
no air gap between the coil and the tube. The magnetic field strength
decreases quadratically with increasing distance between the coil and the
tube.
high electronic conductivity of the tube in order to obtain high eddy
currents
homogeneity in electronic conductivity of the tube, i.e., no tube wall
variation
high starting density of the powder
short rise time of the magnetic field, i.e., a fast discharge of the capacitor
banks
Similar remarks and considerations concerning the cylindrical and flat
62
configurations, as discussed for EXC, can be made here. Table 1 lists the
specifications of the two different dynamic compaction techniques.
In general the lower pressure wave velocity of the magnetic pulse
compaction might cause a less uniform density distribution and a lower
mechanical strength of the compacts due to weaker interparticle bonding
[16,17]. The effective pressure that acts on the particles of the material to be
compacted depends on the nominal pressure as well as on the local pressure
gradients. The local pressure gradient depends strongly on the pressure wave
velocity.
Some typical experimental aspects for the flat and cylindrical
configuration are described below.
Flat configuration: Fig. 6 gives an outline of the MPC set-up used. The
diameter of the mold in which the powder is collected varies between 10 and 50
mm and sample thickness varies between 0.1 and 1 cm. Molds made of
stainless steel or stainless steel alloyed with tungsten carbide (WC-Co) can be
used.
Cylindrical configuration: The cylindrical configuration uses a spirally
wound induction coil. The material to be compacted is placed in a metal tube,
which is closed at the ends. The inner diameter of the coil (which is the
maximum sample diameter) is 25 mm and the effective compaction length
about 80 mm. The effective compaction length depends on the thickness and the
hardness of the material of the target tube and the starting density of the
material to be compacted. A thicker and higher hardness of the target tube
results in a more homogeneous compaction length, but higher pressures are
necessary. A higher starting density of the material to be compacted also
contributes to the homogeneity of the compaction. The generated pressure is in
general lower compared to the flat configuration.
To summarize, two different methods for dynamic compaction were
described, i.e., explosive and magnetic pulse compaction. Both methods can be
used in a uniaxial and a cylindrical mode. With explosive compaction high
pressures can be generated resulting in high densities, while densification of
large samples is feasible. This very reproducible technique has drawbacks in
terms of safety and infrastructure. For application of this compaction technique
63
bunkers, explosives, and specialized personnel are required, which limit the
accessibility to the technique. Furthermore, the compaction rate is quite low due
to the long preparation time of the setup. In addition, the setup is not reusable.
On the other hand, magnetic pulse compaction has a high production rate of up
to one compaction per second without any special safety requirements. The
pressure is easy to control and to adjust and compactions at elevated
temperatures and in different gas environments or vacuum are possible. As a
disadvantage the lower pressure and inherent lower density or smaller sample
size should be considered.
It should be stressed that homogeneous and high densities with a minimal
crack formation can only be achieved by compacting samples with a high
starting density. Furthermore, the temperature increase of the sample is smaller
when using a high starting density due to lower friction forces.
5: Examples and applications of dynamic compaction
Dynamic compaction can be applied for the compaction of a variety of
materials and is particulary interesting for the compaction of materials which
properties negatively change when high temperatures are used. In this respect
one can think about amorphous metals, polymers, ceramics, glasses, nanostructured
materials, and all their composites. Dynamic compaction is a
technique that, in general, does not change the crystal structure of the starting
material. Dynamic compaction can also be used for connecting different
components, especially layers of different materials [15,18-21], when high
temperatures are undesirable or static pressing is inadequate.
In this section the compaction of a selection of nano-structured ceramics
and composites will be described.
Nano-structured powder has been compacted using
uniaxial magnetic pulse compaction [12,22]. The powder has a mean particle
size of 20 nm and has been produced by the exploding wire technique [5,13,23]
resulting in with a specific surface area of and a
theoretical mass density (TMD) of Prior to compaction the powder
was heated to about 450°C and evacuated down to 1 Pa in order to remove any
absorbed and included gases. The initial density was (i.e., 22%TMD)
and the sample diameter was 15 mm and the sample height prior to compaction
was 4 mm. After compaction, performed at room temperature, the samples are
homogenous and crack-free. Fig.7 shows a scanning electron microscopy
micrograph of the MPC fracture surface.
64
X-ray diffraction spectra do not reveal any phase change. In Fig. 8 the density
after compaction as a function of the applied pressure is plotted and a
maximum density of 82% has been obtained at 2 GPa.
As a comparison the same powder has been compacted using static
pressing at room temperature. The density as a function of the applied pressure
is shown in Fig. 8.
From the comparison with the MPC samples a significantly lower density (up
to 15%) can be observed. The reason for this difference is the so-called dynamic
cleaning of the particle surfaces due to the friction forces and heat generation
that occurs during dynamic compaction [12].
The MPC samples of can be used for further sintering in order
to increase the density and the hardness. powders were compacted using
MPC using different pressures resulting in different densities. These green
samples were sintered using high power millimeter wave radiation from a 30
GHz gyrotron [24-26]. Heating rates of 30 to 50°C/s to a sintering temperature
of 1150°C were used in order to prevent excessive grain growth. Fig. 9 shows
the final density and the average grain size after sintering as a function of the
green density (after MPC). From this figure the importance of a high starting
65
density for the sintering process is clearly shown. Combined with the densities
obtained with MPC and static pressing as presented in Fig. 8 it is obvious that
MPC is a very promising technique for the preparation of green bodies that can
be used for sintering up to a high density without excessive grain growth [25].
The powder used has a mean particle size of 16 nm and has been
produced by the exploding wire technique [22]. The theoretical mass density
(TMD) is (about 1:1 rutile:anatase). The material was compacted
using uniaxial magnetic pulse compaction. Prior to compaction the powder was
heated up to about 450°C and evacuated down to 1 Pa in order to remove any
absorbed and included gases. The initial density was (i.e.,
18%TMD) and the sample diameter was 15 mm and the sample height prior to
compaction was 4 mm. After compaction, performed at room temperature, the
samples are homogenous and crack-free. Fig. 10 shows a scanning electron
microscopy picture of the fracture surface of compacted X-ray
diffraction spectra do not reveal any phase change.
In Fig. 11 the density after compaction as a function of the pressure is
66
plotted and a maximum density of 86% was obtained at 1.5 GPa [12].
Li-doped is a ceramic Li-ion conducting electrolyte that can be
used in all-solid-state Li-ion batteries [27-32]. The crystallite size, determined
from the X-ray diffraction line width using the Scherrer equation, is about 30
nm, but the particles are heavily agglomerated to particles of up to 1 µm. Fig.
12 shows a scanning electron microscope picture of the as-synthesized material.
The crystallite size can be adjusted by changing the synthesis parameters
[31]. Sintering of these nano-structured powders leads to drastic grain growth
that results in a huge decrease of the ionic conductivity. The densification of
nano-structured 7 mole% Li-doped has been studied as a function of
pressure and different compaction techniques [9,28]. These techniques are:
uniaxial magnetic pulse compaction (MPC), uniaxial explosive compaction
(EXC), hot uniaxial pressing (HUP), and cold uniaxial pressing (CUP). The
HUP experiments were performed at 1500°C for 5 minutes including heating
and cooling. Figs. 13 and 14 show the SEM micrographs of the MPC and HUP
sample, respectively. The MPC sample shows a rather dense sample with the
nano-structured particles still present. The HUP sample however, shows
massive grain growth with particles up to 1 µm in diameter.
67
As a result ofthe grain coarsening the density ofthe HUP sample is
about 90 %TMD.
Fig. 16 shows the X-ray diffraction spectra of the starting powder and
the compacted samples [28]. The spectra show no peak shift and/or additional
peaks, proving that no structural changes occurred during compaction. The
only effect that the different compaction techniques have on the spectra is a
change in peak width and thus on the particle size. The peak width of the
starting powder and the dynamically compacted samples are comparable
(crystallite size is 26±1 nm calculated from the peak width), while the crystallite
size of the HUP and CUP sample is 89 and 43 nm, respectively. These results
show that dynamic compaction preserves the nano-structured morphology and
does not alter the crystal structure [28].
68
Table 2 shows the importance of preserving the nano-structured
morphology in terms of ionic conductivity. A severe decrease in ionic
conductivity is observed with increasing crystallite size. The reason for this
decrease in conductivity is that Li-doped is a grain boundary conductor,
i.e., the smaller the particles, the more grain boundaries are present, the higher
the ionic conductivity will be [28]. Hot pressing of the sample decreases the
ionic conductivity with more than 2 orders of magnitude despite the high density
of the HUP sample.
Nano-nano composites
The advantages of using MPC for compacting composites compared
to static pressing will be demonstrated by an example of a zirconiahematite
nano-nano composite [33]. The crystallite size of the hematite
phase was 41 nm and 8 nm for the zirconia phase. After compaction with
either uniaxial pressing (UP) at 1000 MPa, or isostatical pressing at 400
MPa (IP), or MPC (1500 MPa), the green bodies were pressure-less
sintered in air for two hours at different temperatures (heating rate
2°C/min). The relative densities after sintering the different green bodies
are shown in Fig. 17 as a function of sintering temperature. It can be
clearly seen that the MPC sample shows the highest density of 94 %TMD
already at 1000°C.
69
6: Conclusion and discussion.
In this chapter several routes of densifying nano-structured ceramics
have been briefly described. The advantages of dynamic compaction over
static compaction techniques have been shown and explained. The dynamic
compaction technique can be used to make final products or half-fabricates
due to the relatively high density and mechanical strength. Both properties
are related to the intimate particle contact obtained.
It has been demonstrated as well that the dynamically compacted
samples can be used as green bodies that can be used for further sintering.
Extremely high densities are reached by combining dynamic compaction
and a subsequent sintering step while the sintering time and temperature
can be much lower than using conventional sintering only. After
densification the material can be still nano-structured due to the high
starting density.
Nano-structured ceramics are one class of materials that can
successfully be densified by dynamic compaction, but the technique is
certainly not limited to these materials. Magnetic pulse compaction is a
commercially used technique for near-net-shape densifying metal powders
into all kinds of shapes with densities close to the theoretical density. The
most widespread application is the shaping of gear wheels for the
automotive industry, but also connecting different metals (cold welding)
and polymer-metal connections are performed commercially using MPC.
The advantage of commercial equipment is the ability to compact one
sample per second.
Explosive compaction and cladding are commercially applied as
well. Artificial diamond is produced on a large scale for abrasive
applications using explosive compaction of graphite. Explosive welding
70
and cladding are rapid growing markets for combining cheap or
mechanically stable materials with an expensive or weak material with
special properties. In addition, explosive compaction generates much higher
pressures than magnetic pulse compaction which can be used to achieve
higher densities and for the densification of larger compacts. On the other
hand is the low production rate and the necessary infrastructure a major
drawback for commercial application.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
P. Boogerd, “Dynamic Compaction of Ceramics”, Ph.D. Thesis, Delft University
ofTechnology, The Netherlands (1995), Ch. 1.
S. Clyens and W. Johnson, Mater. Sci. Eng., 30 (1977), 121.
R. Prümmer, “Explosivverdichtung pulvriger Substanzen” (Springer Verlag,
Berlin, 1987), p.56. (in German)
Ref. [1], Ch. 6
A. Weisenburger, “Anwendung der Hochleidtungsimpuls- und
Mikrowellentechnik zur Herstellung nanokristalliner Pulver und Festkörper”,
Ph.D. Thesis, University of Karlsruhe, Germany (1999). (in German)
E.P. Carton, “Dynamic Compaction of Ceramics and Composites”, Ph.D. Thesis,
Delft University of Technology, The Netherlands (1998), Ch. 2.
D.K. Dijken, “Dynamic and Isostatic Densification of Powder Materials” Ph.D.
Thesis, University of Groningen, The Netherlands (1994).
M.J.G. Jak, “Dynamic compaction of Li-ion battery components and batteries”,
Ph.D. Thesis, Delft University of Technology, The Netherlands, 1999.
M.J.G. Jak, E.M. Kelder, J. Schoonman, V.N. Lashkov, A.V. Strikanov, A.A.
Selezenev, and A.A. Potanin, J. Appl. Phys., 86-10 (1999), 5455.
Ref. [6], Ch. 3.
T. Negishi, T. Ogura, T. Masumoto, T. Goto, K. Fukuoka, Y. Syono, and H. Ishii,
J. Mater. Sci., 20 (1985), 399.
V. Ivanov, S. Paranin, A. Vikhrev, R. Böhme, and G. Schumacher, Proceedings
4th
Euro Ceramics, Riccione, October 2-6, 2 (1995), 169.
V. Ivanov, Y.A. Kotov, O.H. Samatov, R. Böhme, H.U. Karow, and G.
Schumacher, Proceedings International Conference Nanostructured Materials,
Stuttgart, October 3-7, 1994.
R. Böhme, M.J.G. Jak, G. Link, and A. Weisenburger, Abstracts ESF-NSF
symposium “Aerosols for Nanostructured Materials and Devices”, Edinburgh,
September 12, 1998, 42.
M.J.G. Jak, F.G.B. Ooms, E.M. Kelder, W.J. Legerstee, A. Weisenburger, and J.
Schoonman, J. Power Sources, 80 (1999), 83.
E.A. Doroshkevich and O.V. Roman, Rep. 8th Union Powder Met. Conf., Minsk
(1966), 61.
J.D. Campbell, Mater. Sci. Eng., 12 (1973), 3.
Ref. [8], Ch. 3.
M.J.G. Jak, E.M. Kelder, and J. Schoonman, “A method for the manufacturing of
a lithium battery”, Patent PCT/NL/96/00356, 1996.
M.J.G. Jak, E.M. Kelder, M. Stuivinga, and J. Schoonman, Solid State Ionics, 86-
88 (1996), 897.
M.J.G. Jak, E.M. Kelder, A.A. van Zomeren, and J. Schoonman, Proceedings of
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
the Electrochem. Soc. “Symposium on Exploratory Research and Development of
Batteries and Supercapacitors for Electric and Hybrid Vehicles”, Eds. W.A.
Adams, A. Landgrebe, and B. Scrosati, 1996, p. 58-70,
Ref. [5], Ch. 6.
I.V. Beketov, Y.A. Kotov, A.M. Murzakaev, O.M. Samatov, V. Volkov, R.
Böhme, and G. Schumacher, Proceedings 4th
Euro Ceramics, Ed. C. Galassi,
Riccione, October 2-6, 1 (1995), 77.
G. Link, A. Weisenburger, R. Böhme, M. Thumm, in “Mikrowelleneinsatz in der
Materialprozesstechnik, der chemischen Verfahrenstechnik und in der
Festkörperchemie”, ed. M. Willert-Porada, Shaker Verlag, Aachen (1998), p. 320
(in German).
G. Link, A. Weisenburger, R. Böhme, M. Thumm, Proceedings 9th
Int. Conf. on
Modern Materials & Processing, CIMTEC '98, Firence (1998)
G. Link, V. Ivanov, S. Paranin, V. Khrustov, R. Böhme, G. Müller, G.
Schumacher, and M. Thumm, Materials Research Society Symposium
Proceedings, 430, Microwave Processing of Materials, p. 157, Eds. M.F.
Iskander, J.O. Kiggans, and J-Ch. Bolomey (1996).
Ref. [8], Chs. 4-6.
M.J.G. Jak, E.M. Kelder, N.M. van der Pers, A. Weisenburger, and J.
Schoonman, J. of Electroceramics, 2 (1998) 127.
E.M.Kelder, M.J.G. Jak, F.de Lange, and J.Schoonman, Solid State Ionics 85
(1996) 285.
M.J.G. Jak, E.M. Kelder, and J. Schoonman, J. of Solid State Chem., 142 (1998),
74.
M.J.G. Jak, E.M. Kelder, S.J. Everstein, and J. Schoonman, J. of Power Sources,
81/82 (1998), 808.
M.J.G. Jak, E.M. Kelder, Z.A. Kaszkur, J. Pielaszek, and J. Schoonman, Solid
State Ionics, 119 (1998), 159.
T.P. Raming, “The synthesis of nano-nano dual phase ceramic composites”,
Ph.D. Thesis, Twente University, The Netherlands (2000), Ch. 5.
71
STRUCTURAL AND ELECTRICAL
PROPERTIES OF
NANOSTRUCTURED AND COARSE
MANGANESE SPINEL
J. Molenda and J. Marzec
Faculty of Materials Science and Ceramics, Staszic University of Mining and
Metallurgy, Al. Mickiewicza 30, 30-059 Cracow, Poland
manganese spinel is a promising cathode material for 4V
lithium accumulators [1-3]. Its reversible capacity of 100 – 130 mAh/g is
comparable to the capacity of but it has higher potential, is cheaper
and less environment polluting than attracts an enormous
and still growing interest of the leading laboratories in the world and the
number of related papers is increasing recently very fast indicating, that this
material will be soon commercialized. The studies of this lithium manganese
oxide reported in literature concentrate mainly on the determination of
operating parameters of batteries, lithium chemical diffusion coefficient
measurements, structural parameters, and attempts to find admixtures that
improve battery operating characteristics. The obtained characteristics often
concern materials that are not well defined from the point of view of ionic
and electronic defects structure. Microscopic mechanism of charge transport
that is responsible for the efficiency ofthe intercalation process is practically
not known.
Manganese spinel is synthesized mainly by a traditional high-temperature
(~800°C) solid state reaction. Thus obtained stoichiometric manganese
spinel undergoes a phase transition from regular (Fd3m) to
orthorhombic (Fddd) phase near room temperature (~290K). The transition
is connected with the Jahn -Teller distortion of the ions surrounding.
Basic structural studies performed by Rodriguez-Carvajal and other groups
1. INTRODUCTION
74
[4-6] with the use of neutron and electron diffraction methods have shed new
light on the nature ofthis phase transition and have indicated its relation to a
specific charge ordering in the manganese sublattice. This ordering, called
“electronic crystallization” [4], consists in a columnar (in the c-axis
direction) ordering of ions surrounded by ions and affects
transport properties of a cathode material. Unquestionably, the existence of a
phase transition in near room temperature, i.e., close to the
working temperature of the battery may cause a change of its important
operational parameters with temperature. In literature there is a common
opinion that the phase transition present at around room temperature worsens
the properties of the cathode material, however there is no experimental
proof for it. In addition there is still no data on the dependence of the phase
transition on the deintercalation degree x in There is also lack of
data on the dependence of electrical properties of the deintercalated spinel on
deintercalation degree x. Only limited data on the properties of
stoichiometric at temperatures close to the phase transition are
available.
There are particularly poor conditions for electron transport in manganese
spinel at room temperature as compared with other widely used cathode
materials in Li-ion batteries. Polaron conduction mechanism with activation
energy of 0.2-0.3eV cannot explain observed high efficiency of intercalation
process.
The issue is what are the physicochemical properties of deintercalated
spinel, what is a factual electron transport mechanism in operating spinel
cathode and how the phase transition affects the efficiency of intercalation
process.
As the conductivity is surely one of the most important features of the
cathode material, systematic investigations on electrical transport and related
phenomena in together with structure and DSC
characterisation at subsequent stages of deintercalation process were
performed.
Literature reports indicate that manganese spinel obtained by lowtemperature
soft chemistry methods yields better electrochemical
characteristics of batteries [7]. Using these methods better homogeneity of
materials is obtained as substrate mixing is done in a solution that means on
the atomic level. In addition, grains having significantly smaller size are
obtained, which has an influence on diffusion processes rate in a cathode
material.
Nanostructured manganese spinel was obtained by a sol – gel
method and its structural, heat and electrical properties were investigated.
Analogous characteristics were determined for the same samples after
annealing them at 800°C in air for 24 hours. For comparison a sample of
75
made from commercial
Selectipure SP30 powder (Merck Co.) was also tested. Subsequently all
these samples were delithiated and characterised in the same way as the
starting materials.
2. STUDIED MATERIALS
2.1. Nanostructured manganese spinel – “series A“
Stoichiometric amounts of (Aldrich, SigmaUltra 99.99%) and
(Aldrich 99.99%) were dissolved in a minimal amount
of deionized water. Concentrated ammonia (Aldrich, A.C.S. Reagent 28%)
at 25°C was added drop by drop to the mixture until was reached, at
which a precipitation of colloidal deposit was observed. After had
been reached the mixture was left for 1h continuously stired. Thus formed
colloid was next dried at 90°C during 48h and subject to calcination in air at
300°C for 24h. The heating and cooling rates were 50°C/min. The whole solgel
processes were described in details in [8].
2.2. Strictly stoichiometric “series B”
Series B materials were obtained by annealing series A samples at 800°C
in air for 24h and quenched.
2.3. with low lithium excess
“series C”
samples were obtained by annealing for 24h in air at 800
°C pellets of commercial Selectipure SP30 powder (Merck Co.) and then
quenched.
2.4. Delithiated materials
Lithium was extracted by two methods – electrochemical and chemical.
Electrochemically deintercalated samples of
and 0.40) were obtained in a cell (with in
DMC/DME as electrolyte). No additives were doped to the cathode
materials. Deintercalation processes were carried out at room temperature
with spinel pellets of all A, B and C series as cathodes and the cells were
charged at a constant current density
Hunter’s method [9] of lithium extraction as a partial dissolution of
manganese spinel in sulfuric acid was used for chemical deintercalation of
the C series. This process is described by reaction:
76
Such Li – extracted materials were already studied by XRD, IR
spectroscopy [10] and also characterized by laser diffraction, BET method,
scanning electron microscope and electrochemical behaviour [11].
Figs. 1a and 1b present SEM photographs of the surface of the spinel
obtained by the sol – gel method and after annealing it at 800°C (series A
and B). A comparison of the pictures reveals that annealing of the samples
at high temperatures leads to an increase in the grain size. In the case of
spinel obtained by the sol – gel method a mean size of the grains is equal to
while that for the sample annealed at 800°C in air for 24h increases
to Fig. 2 present SEM pictures (a) and particle size distribution (b) of
the C series.
77
78
3. STRUCTURAL PROPERTIES
The fundamental lattice of the spinel structure can be considered as a face
centred cube of ions and the cations are located in the interstitial
positions in the anion lattice. In the spinel structure there are two
kinds of cation sites. The 8a sites are occupied by ionstetrahedrally
coordinated by four anions. The interstices of the coordination
tetrahedron are small so the large ions enlarge the 8a position shifting the
anions outward. Another cation sites 16d, each coordinated by six are
occupied by equal number of and ions and the electric charge is
+3.5 on average. These sites can be considered as a diamond type lattice of
tetrahedrons formed by 16d cation sites. These tetrahedrons sharing corners
with each other are lined along direction building Kagomé type lattice.
3.1. XRD analysis
3.1.1. Series A
For the nanostructured manganese spinel sample obtained by the sol – gel
method powder X-ray spectra were recorded at 300K and 70K. It was found
that at both temperatures the sample remains a single phase one and that its
high-temperature cubic structure is maintained also at 70K. Both of the
measured X-ray patterns reveal identical and very strong broadening of the
lines, which is a sign of small dimensions of the coherently scattering
domains and, as a consequence, small grain size or (and) the existence of
strong stresses in the sample resulting from heterogeneity of the composition
or structure faults. A part of the X-ray pattern recorded at 300K is shown in
Fig.3a. Apart from broad lines of the spectrum contains three very
narrow lines of pure germanium which was used here as an internal standard
(giving in practice apparatus line width). As determined by Rietveld
analyses of the XRD pattern, the lattice parameter a of the sample at room
temperature equals to 8.165(4) Å. The obtained value, unexpectedly low as
compared with the anticipated 8.25 Å for exactly stoichiometric material
indicates that possibly not all structural sites in the material obtained by the
low temperature method are occupied. Thermogravimetric measurements
point to irreversible oxygen loss on heating, starting at 400°C (1.6% overall
mass loss), in accordance with ASA results yielding high average manganese
valency equal 3.83 in the sol-gel material [8]. The series A material
exhibiting oxygen excess is not a stoichiometric manganese spinel. The
structure of the material can be considered as well formed oxygen
framework but with vacancies in cation sublattice leading to the observed
low lattice parameter and high manganese valency.
Following annealing at 800°C leads to oxygen loss until the
stoichiometric composition is achieved (series B).
79
3.1.2. Series B
Fig.3b presents a part ofthe X-ray pattern measured at 300K for the
series B sample. One can see a distinct narrowing of the lines as compared
with the series A, what indicates significant increase of coherently scattering
domains related to the expected growth of the grains and a general
improvement of the quality of the sample crystal structure. Lattice
parameter a is equal to 8.2479(5) Å at room temperature, characteristic for a
stoichiometric phase, what is in accordance with the obtained from
ASA experiment 3.53 average manganese valency in the sample.
80
However one can notice differences in relative widths of the lines
from Fig.3b resulting from the fact that the sample is a two-phase one. This
is due to the phase transition from a high-temperature cubic to a lowtemperature
orthorhombic structure appearing for a stoichiometric
at around 290K. At 300K the sample is in the temperature region of a
structure transformation. This means that the true line width for this sample
is still smaller and determined by the width of the narrowest lines in the
spectrum (excluding germanium lines).
In order to follow the process of crystal structure transformation a
series of X-ray spectra were recorded between 320K and 220K both on
heating and cooling of the sample. Fig.4 presents a temperature dependence
of the (400) line width at half maximum (FWHM). The results show both
the sharpness of the structure change and the presence of temperature
hysteresis, which indicates that the transition is of the first order type. The
results remain in excellent agreement with the literature data [12], where for
a stoichiometric sample a first order phase transition from cubic to
orthorhombic structure was also observed in this temperature range with a
hysteresis of about 20K. At low temperatures the entire sample lattice
undergoes the orthorhombic distortion.
81
3.1.3. Series C
For the composition of manganese a
broadening ofthe diffraction lines with lowering temperature was also
observed. The broadening was caused by fine splitting of the cubic
structure lines resulting from the orthorhombic distortion. In Fig.5
two selected peaks from XRD pattern fitted with both cubic and
orthorhombic components are given. One can notice that the fit nicely
follows the experimental peak shape. Fig.6 presents half-maximum
width (FWHM) of the (004) line as a function of temperature for
An analysis using the Rietveld method shows that the
best fit to the experimental X-ray patterns is obtained assuming a
coexistence of cubic and orthorhombic phases in relative proportions
varying with temperature. In Fig.7 is shown the relative content of the
cubic phase as a function of temperature. These results indicate a
presence of continuous transition (progressing distortion) from cubic
to an orthorhombic phase (of lower symmetry) as the temperature is
decreased, and contrary, this leads to a decrease of the disordered
phase with rising temperature. The temperature value seems only to
determine the ratio of both phases, like it takes place in a second order
order–disorder transitions. However, observed transition does not
lead to a total transformation of the initial cubic structure till below
70K.
82
83
4. ELECTRICAL AND THERMAL PROPERTIES
4.1. Series A and B
Figs. 8 and 9 present temperature dependencies of thermoelectric
power, electrical conductivity and DSC effect for manganese spinel obtained
by a sol-gel method (A series) and after annealing it at 800°C in air for 24h
(B series), respectively.
84
At room temperature the values of electrical conductivity for manganese
spinel obtained by a sol – gel method are of the order of (Fig.8b)
which is lower as compared to for the annealed sample (Fig.9b).
The obtained conductivity activation energy (0.1 eV) related to the activation
energy for small polaron movement is lower than for the spinel annealed at
800°C (0.2 eV in the temperature range 285 – 310K). This is connected with
a significantly smaller value of a parameter in the case of low-temperature
spinel as compared to the annealed one. Lower absolute values of electrical
conductivity in the case of spinel obtained by a sol – gel method are related
to lower effective carrier concentration (proportional to the
product) as well as scattering of the carriers on intergranular boundaries.
Manganese spinel annealed at 800°C has different electrical
conductivity characteristics (Fig.9b). A jump of the conductivity by one
order of magnitude is seen at 285K, i.e. at the phase transition temperature
(Figs.4, 9c). Comparing the results of Rodriguez-Carvajal [4] who also
detected a jump by only ½ order of magnitude, one can conclude that the
conductivity decrease by one order of magnitude observed here indicates a
high purity of the B series sample and a strictly stoichiometric composition
of it.
Figs.8a and 9a present the temperature dependencies of
thermoelectric power for A and B series, respectively. The stoichiometric
manganese spinel (B series) shows an anomalous characteristics of
thermoelectric power (Fig.9a). A sharp peak having magnitude of the order
of 500µV/K is observed at the temperature at which the phase transition is
detected (compare Figs.4, 9b, 9c). In the case of the nanostructured spinel
produced by the sol – gel method (A series) a flat line characteristics is
obtained, typical of polaron conduction mechanism (Fig.8a).
4.2. Series C
The phase transition commencing at 280K in the C series sample
(figs.5-7) is also visible in conductivity, thermoelectric power and DSC
measurements. In figs. 10a,b,c a small peak in thermoelectric power, an
obvious change in electrical conductivity and thermal effect can be noticed
as associated with phase transition. The observed temperature hysteresis in
thermal and conductivity effects can be attributed to different
cooling/heating ratios (10 deg/min and 0.3 deg/min rate for DSC and
conductivity respectively). For the slowest temperature change as in XRD
measurements (0.03 deg/min) no shift was observed (fig.6). However for
the same temperature scan ratio for the B series sample a clear hysteresis
was observed, indicating a modification of the phase transition nature for the
C series material towards a second order type transition.
85
5. EVOLUTION OF THE PHASE TRANSITION IN
MANGANESE SPINEL UPON DELITHIATION
The presented results show that in manganese spinel at room temperature
conditions for charge transport are rather poor. Its electrical conductivity at
300K (i.e. under the conditions of battery use) is very low, of the order
Under such circumstances the deintercalation/intercalation process
should be extremely difficult. Nevertheless it proceeds with a good
efficiency. Therefore a question arises: what is the real electron transport
mechanism in functioning spinel cathode and is the phase transition still
present in a deintercalated cathode material?
Commonly it is known that the phase transition occurs for the critical
50% concentration on Jahn-Teller ions Therefore it should not be
visible in deintercalated spinel in which the concentration falls. In fig.
11 the dependence of phase transition temperature range on lithium content
was shown. For comparison also the results obtained for the strictly
stoichiometric, annealed sample (series B) were also presented.
From fig. 11 follows that electrochemical deintercalation of
86
series C spinel down to shifts the temperature of the beginning of
the phase transition to 300K, but it becomes much wider on temperature and
leads only to a 50% phase change even down to T=70K. For higher
deintercalation degree rather a disordered structure is obtained
with various lithium occupancy of the 8a structural sites and in the XRD a
line broadening is observed in the whole temperature range (fig. 11). A
thermal effect indicating a “glass–like” behaviour (i.e. appearing of local
orders) on DSC curves was registered on cooling (fig 14c). Also EPR
spectra for those samples are broadened what can be attributed to a local
field distributions arising from a disordered structure.
6. TRANSPORT AND THERMAL PROPERTIES OF
DELITHIATED MATERIALS
6.1. Series C – electrochemically deintercalated
In figs. 12-15 temperature characteristic of the electrical conductivity and
thermoelectric power for the compositions 0.75 and 0.40 (C
series) are shown together with DSC measurements. One can see measured
sharp maxima with values as high as 600 – 10 000 µV/K in Seebeck
coefficient, stepwise changes in the electrical conductivity and
corresponding thermal effects on DSC curves. For and 0.94
compositions these anomalies can be associated with the orthorhombic
distortion of cubic spinel lattice taking place at around room temperature.
The temperature positions of these irregularities do not change with
subsequent cycles of heating and cooling. For 0.4 compositions
the DSC and electrical conductivity results differ (figs. 14, 15). For the
sample on the DSC curve for heating subsequent processes of
87
“freezing” (exothermic) and “melting” (endothermic) are observed while for
there is no sharp anomaly ( figs 14c and 15c respectively). For both
these compositions only a slight change in conductivity is observed (figs.
14b and 15b). In XRD patterns of these samples lines are generally broad in
the whole temperature range indicating a presence of a defected structure
and no suppositions can be made on appearing ofthe distorted phase.
88
The Seebeck coefficient generally reflects the electron energy
spectrum weighted by the conductivity. Measured peak values of of the
order of 10mV/K (as well as the peaked dependence on temperature) cannot
be ascribed to any conventional electronic mechanism. Relatively narrow
and high value peaks in thermoelectric power measurements can originate
from two-phase system occurring in the measurement procedure when the
whole sample is not at the same temperature. The applied temperature
gradient at a certain initial temperature causes the beginning of the order-
89
disorder phase transition at one end. The instability in phonon spectrum
(acoustic mode softening) of the arising phase can lead to the appearance of
an additional charge carrier concentration gradient as a result of a process
analogous to well-known phonon drag mechanism, leading to such high
values of in narrow temperature range.
The DSC measurements can be interpreted within a framework of
order–disorder transition (therefore rather considered as second order type).
Especially for on cooling curve a glass–like behaviour is clearly
visible indicating freezing ofa short-range order (fig.14c). When the sample
was cooled down to 170K afterwards on heating DSC trace the subsequent
“freezing” and “melting” processes were observed.
6.2. Series C – chemically deintercalated
Chemical deintercalation of manganese spinel of the C series leads to a
material without the phase transition as it was not seen in XRD patterns and
on DSC curves (fig.16c). Consequently no anomalies in electrical properties
were observed for those samples (fig.16a,b).
90
A slight decrease of the activation energy of electrical conductivity with
deintercalation degree was only observed as expected in this system with
polaron mechanism of electrical conduction where the lattice parameter
decreases from starting 8,239C down to 8.039C for the (fig. 17).
The performed studies revealed significant differences in physical
properties between chemically and electrochemically deintercalated samples
of manganese spinel. That can be attributed to a different microscopic
mechanisms of lithium extraction from the spinel structure: a selective one
as in the electrochemical process and the relatively rapid, structure
destroying when dissolved in the sulfuric acid. The first method leads to a
material in which the manganese sublattice remains almost unchanged and
that is the reason the material still exhibits the traces of the phase transition,
while the samples obtained with the second process possess highly defected
the Mn sublattice what disables the occurrence of the structure distortion and
all the phenomena related with it.
6.3. Series B – electrochemically deintercalated
In figs. 18 and 19 are shown temperature dependencies of
thermoelectric power, electrical conductivity and DSC for deintercalated
samples from series B with compositions and As
can be seen also these samples shows anomalous electrical and thermal
properties characteristic of the phase transition, similarly as for
stoichiometric samples from B series. Further deintercalation leads to
crushed samples what made electrical measurements impossible.
91
6.4. Series A– electrochemically deintercalated
For the nanostructured cathode material, with no additives characterized
by low electrical conductivity (fig.8) and high content the
electrochemical deintercalation route is pretty difficult. In figs. 20 and 21
are shown temperature dependencies of thermoelectric power electrical
conductivity and DSC for deintercalated samples from series A with
compositions and For the first composition no
anomalous effects were observed in electrical and thermal properties as for
92
the starting material (fig. 8). Further deintercalation degree (x=0.98) leads to
irregularities in temperature dependencies of electrical conductivity and
thermoelectric power (fig. 21).
6.5. Lithium chemical diffusion coefficient
In fig. 22 results of two independent measurements of the lithium
chemical diffusion coefficient are shown as a function of deintercalation
93
degree of manganese spinel from the C series. This parameter is directly
responsible for the speed of intercalation process. In general a quite good
intercalation rate is obtained in wide composition range
These two different measurements of diffusion coefficient yield values
differing of the order of magnitude, anyway from both it is clear that the
diffusion coefficient is evidently lower for the starting sample.
The high value of is obtained for composition for which
a considerable structure disorder are observed (as assumed from XRD),
glass-like behaviour on DSC curve, greatest EPR signal broadening, the
lowest (0.l8eV) activation energy in conductivity and the largest peak in
thermoelectric power (up to 10mV/K) (fig. 14). It follows from above the
disordered structure does not limit the intercalation process. Moreover it
seems to be propitious to a conductivity mechanism that seems to take place
in working spinel cathode, differing from that present in equilibrium state
material [13].
It was noticed a huge difference of the resistivity of working spinel
cathode as estimated from the loading circuit and that of the same material
picked up from the cell and submitted to standard four probe dc conductivity
method. The measured values differ by two orders of magnitude – the
electrical conductivity in working cathode is much higher. Similar results
can be obtained when Nernst-Einstein relation was used for estimating
electrical conductivity of operating spinel cathode basing on determined
lithium diffusion coefficient.
94
7. SUMMARY
The phase transition and related effects can be described in terms of
localised polaron condensation being an appearance of a short and a long
distance orders of “molecular polarons”. A dynamical Jahn-Teller active
cluster composed of four 16d manganese positions as well as four
neighbouring oxygens plays role ofa self- trapped small polaron, interpreted
as “molecular polaron”.
Below the “phase transition temperature” the interaction between
molecular polarons becomes dominant resulting in observed long distance
charge ordering accompanied by small lattice distortions. Within the cluster
charge density fluctuations related with lattice distortion are present. This is
a vibronic state of molecular polaron in which the motion of the electron and
atomic displacements are dynamically coupled.
For strictly stoichiometric sample with good, low in defects initial
structure (series B) the ordering process can occur in a narrow temperature
range, as was shown. In this case the phase transition is of the first order
while for the C series it looks like a second order transition with no
temperature hysteresis.
In Mn sublattice every 16d site belongs to two tetrahedrons so the
charge at the corners should be divided equally between them. The
Coulomb energy between tetrahedrons is lowest when the charge ofthe each
is the same +7e (it means two pairs of and occupy the four
corners – then a Coulomb energy within the tetrahedron is also a minimum
too). A condition that the charge of the tetrahedron is constant all over the
crystal is called Anderson’s condition and is a basis for electronic structure
of spinel type oxides. It is worth noting there is no charge transport in the
crystal if all the tetrahedrons fulfil Anderson’s condition. If
pair belongs to the same tetrahedron an electron jump does not change the
total charge of it. A finite amount of energy is necessary to violate
Anderson’s condition and thus to make an electron move from one atom to
the other.
Clearly different situation is created when ion leaves the 8a
position in working cathode. Lithium cation leaves an electron, locally
violates the Anderson’s condition and the left charge can easily flow within
manganese sublattice of a grain as a sequence of electron jumps.
The motion of electrons is mainly governed by the short range order and not
by the long range order of Mn ions.
The analysis of electrical conductivity and thermoelectric power
characteristics of deintercalated spinel in the vicinity of room temperature
(i.e. working temperature of the battery) together with XRD, DSC and
95
lithium chemical diffusion data, enables to conclude that good operation of
the cathode may be related to the proposed alternative charge
transport mechanism involved in working manganese spinel cathode. This
mechanism ensures easier electron charge transport and being present in a
wide range of compositions ensures a high efficiency of the
deintercalation process.
The performed studies indicates that the low temperature, sol-gel
preparation method leads to a cathode material with low electrochemical
activity in the 4V region due to high content of and low electrical
conductivity. Annealing at 800°C of the material leads to the strictly
stoichiometric manganese spinel with better transport properties
and electrochemical activity.
ACKNOWLEDGMENT
The “sol-gel” samples were obtained by M. Molenda from the Faculty of
Chemistry, Jagiellonian University, Cracow, Poland. The chemically
deintercalated samples were prepared by D.R. Simon in Delft Interfaculty
Research Center: Sustainable Energy, Delft University of Technology, The
Netherlands.
The work is supported by the Polish Committee Scientific Research
under grants 7T08A 049 18 and PBZ KBN 013 / T08 /12.
REFERENCES
M.M. Thackeray, M.F. Mansuetto, D.W. Dees and D.R. Vissers, Mat. Res. Bull, 31 No.
2 (1995) 132
Yuan Gao and J.R. Dahn, J. Electrochem. Soc., 143 No. 1 (1996)100
E.M. Kelder, M.J.G. Jak, J. Schoonman, M.T. Hardgrave, S.Y. de-Andersen, Journal of
Power Sources, 68 (1997)590
J. Rodriguez-Carvajal, G. Rousse, C. Masquelier, M. Hervieu, Phys. Rev. Letters
81(1998)4660
G. Rousse, C. Masquelier, J. Rodriguez-Carvajal and M. Hervieu, Electrochem. and
Solid-State Lett., 2 (1999)6
M.R. Palacin, Y. Chabre, L. Dupont, M. Hervieu, P. Strobel, G. Rousse, C. Masquelier,
M. Anne, G.G. Amatucci and J.M. Tarascon, J. Electrochem. Soc., 147 (2000) 845
L. Chen, X. Huang, E. Kelder, J. Schoonman, Solid State Ionics, 76(1995)91
R. Dziembaj, Solid State Ionics (2002) in press
J.C. Hunter, J. Solid State Chem. 39(1981)42
L.He, S. Tang, Z. Chen, H. Du, Solar Energy Materials Solar Cells 62(2000)117
L.He, S. Tang, Z. Chen, H. Du, Solar Energy Materials Solar Cells 62(2000)125
H. Hayakawa, T. Takada, H. Enoki, E. Akiba, J. Materials Science Letters,
17(1998)811
J. Marzec, J.Molenda, D.R. Simon, E.M. Kelder,
J.Schoonman, Solid State Ionics 146(2002)225
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
METAL-POLYMER NANOCOMPOSITES:
FORMATION AND PROPERTIES
NEAR THE PERCOLATION THRESHOLD
S.A.Zavyalov1
, A.A.Timofeev2
, A.N.Pivkina3
, and J.Schoonman4
1
Karpov Institute of Physical Chemistry, Vorontsovo Pole, 10, 103064 Moscow, Russia;
2
Moscow State Engineering-Physical Institute, Kashirskoe Shosse, 31, 115409 Moscow,
Russia;
3
Semenov Institute of Physical Chemistry, Russian Academy of Science, Kosygin st. 4
119991, Moscow, Russia;
4
Delft University of Technology, Delft Institute for Sustainable Energy, P.O. Box 5045,
2600 GA Delft, The Netherlands.
ABSTRACT
Nanostructured metal (Pd, Sn, Cu, Al)-polymer (poly-para-xylylene)
and metal-oxide-polymer composites reveal synergism of properties of the
initial components, which gives rise to specific electrical, mechanical,
optical and chemical properties related to an ordered distribution of
nanoparticles over the matrix volume. Various methods of preparing a
metal-polymer nanocomposite exist and from the point of view of
homogeneity, methods where polymerization and formation of the
nanoparticles are performed simultaneously, are very promising. This
chapter reports on the experimental set-up for the production of
nanocomposites by vacuum co-condensation of the metal and monomer
(paracyclophane) vapours. The morphology, optical, electrical properties
and oxidation behavior of the deposited composites have been studied.
The influence of metal concentration on composition, structure,
morphology, and particle size of Al/PPX nanocomposites has been studied
utilizing transmission electron microscopy (TEM). After air exposure the
inorganic particles were identified as aluminium or alumina, depending on
metal concentration in vacuum. The size of nanoparticles within the
polymeric matrix was measured in the range of 10 to 55nm. The
temperature coefficient of resistance and the oxidation behaviour of the
synthesized composites were measured.
The collective effects of charge transfer within ensembles of metal
nanoparticles in the vicinity of the percolation threshold is proposed to be
a reason of observed electro-physical and chemical properties of the
synthesized nanocomposites.
98
INTRODUCTION
To date, nanostructured metal-polymer and metal oxide - polymer
composites are the subject of increased attention of researchers in
different areas of science and technology. Among the perspectives for
using these materials are novel types of solar cells and chemical gas
sensors. The electrical, mechanical, and optical properties of the
nanocomposites are improved by dispersing a metallic phase in a
polymeric matrix.
At a particular concentration of nanoparticles close to the
percolation threshold the nanocomposite materials exhibit properties
specific to the nanoparticles. These properties differ considerably from
those for bulk materials and the composite acquires these during
composite formation. As a result such materials reveal synergism of
properties of the initial components. Using different polymers adds
elements of self-organisation to the formation of the composite, for
inorganic particles are located in cells of the polymer between grids and
layers [1] which gives rise to specific properties of the composite related
to an ordered distribution of nanoparticles over the matrix volume.
Experiments show that unique properties of some thin-film
nanocomposites are due to cooperative effects amongst nanoparticles. The
improved gas sensitivity and photosensitivity of
nanocomposites were measured [1]. The capacity of this material was
found to be 40F/g [3]. The maximum of the catalytic selectivity of Pdcontaining
nanocomposites in the isomerization reaction of 3,4dichlorobutene-1
to 1,4-dichlorobutene-2 was observed at a Pd
concentration close to the percolation threshold [2].
It has been reported that in the vicinity of the percolation
threshold thin-film nanocomposites exhibit properties that differ from
isolated nanoparticles and from the bulk material [7, 8]. This supports the
cooperative, non-equilibrium effects of charge transfer between metal
nanoparticles at threshold concentrations [6]. The charge transfer between
nanoparticles can be connected with two types of conductivity, i.e. metallike
contact electrical conductivity and tunnel conductivity via a nonconducting
polymer layer.
To confirm the realization of the collective effects in metalpolymer
nanocomposites, aluminium/poly-para-xylylene nanocomposites
have been synthesized via a vacuum co-deposition technique. The
properties of the composites are studied by comparing the structure,
morphology, electrical properties and the oxidation kinetics.
99
EXPERIMENTAL PROCEDURE
Figure 1 shows the setup for depositing composite films by cocondensation
of metal vapour and an active precursor, which then
polymerizes. The set-up consists of four principle units: the zone of
treating the polymer precursor, the zone of evaporating metals, the reactor
and the deposition zone.
The polymeric matrix is synthesised as follows: in the evaporation
zone the precursor (paracyclophane [4]) is heated up to 380 - 420K. The
paracyclophane pyrolyses at 930K giving rise to biradicals of
paraxylylene. These molecules then deposit without polymerisation onto a
cooled (down to 77K) substrate. The flux of metal atoms evaporated from
a bulk sample condenses onto the substrate together with the monomer.
The condensate consists of nanoparticles of the metal and the monomer.
Upon heating the substrate to ambient temperature the monomer
polymerizes to poly-para-xylylene. The structure thus obtained is a porous
matrix with dispersed nanoparticles in it. Below the properties of
nanocomposites containing metal or metal-oxide nanoparticles in the
polymeric matrix are presented. Manipulating the synthesis conditions,
i.e., the distance between the vapour source and the substrate, the tilt angle
of the beam, and the deposition time, allowed for optimising the
deposition regime. Measuring the electrical resistance of the condensate
and composite permitted the control of the film formation.
Three types of substrates have been used in the experiments
reported here (Fig. 2), i.e., (1) a polished quartz substrate of size 5x5mm
100
and 1 mm thick with Pt-contacts for electrical measurements; (2) a
polished NaCl single-crystalline substrate of the same size and thickness
for TEM analysis; (3) a polished quartz substrate of size 10x20mm and
2mm thick for optical and AFM investigations. To obtain identical
samples both types of substrates were fixed close to each other onto a
cooled surface of the sample holder (4).
Oxidation kinetics during the air exposure after vacuum synthesis
was measured by means of data acquisition board L-1250, connected to a
PC. The temperature coefficient of the electrical resistance in vacuum (the
slope of versus temperature dependency) was measured
after re-cooling of the synthesized composite from 293K to 77K.
The morphology of nanocomposites was studied with transmission
electron microscopy (JEM-2000 EX-II) at 200kV. Samples for TEM were
prepared by standard procedures including separation of the
nanocomposite layer from the NaCl substrate in water and the film
deposition onto a Cu grid for further investigations. The metal content in
the composites was calculated by atomic absorption analysis using a
Perkin-Elmer 503 spectrometer. The surface morphology and composite
101
structure were studied by atomic force microscopy (SOLVER-P47) with
an etched silicon cantilever having a tip radius of 10nm and 20° apex
angle. Data were taken in ambient air with a contact force of The
optical spectra of the films were recorded with a spectrophotometer
Shimadzu UV-3100 in the wavelength range 200 - 2000nm.
RESULTS
Structure and morphology
Freshly synthesised tin-containing samples are normally of grey
colour with a metallic lustre. On contacting with the ambient air during 2
minutes the samples became transparent for composites with a Sn
concentration below the percolation threshold of 10Vol.%, but the
samples with a tin content beyond 10Vol.% do not change their colour
during several months (Fig. 3).
The surface morphology of Sn/PPX nanocomposites with different
content of the filler from 8Vol.% to 16Vol.% (Table 1) was investigated.
AFM analysis of model Pd-based composites (Fig. 4) shows that the
surface of the nanocomposites is sufficiently uniform. The spherical
polymeric globules of a size up to 200nm are well distinguishable. The
polymeric spherulites are uniform in size and shape and form a porous
rather than a continuous structure. The pore size does not exceed 100nm.
In addition, at the spherulite surfaces imperfections are distinguishable
deviating their shape from sphere. The phase-contrast regime of the AFM
(Fig. 5a) allows for identifying these surface imperfections as the particles
of the inorganic phase, establishing location, and evaluating their sizes.
According to Fig. 5b, where the cross section ofthe composition profile is
presented, the inorganic particle size is 7 - 10nm.
102
Analysis of the composites reveals that for the metal content in asprepared
samples below or at the percolation threshold (samples 1, 2,
Table 1) the inorganic particles are isolated and the interparticle distance
varies from 5 to 20nm (Figs. 6, 7). Slightly above the percolation
threshold (sample 3) the particles form continuous filaments of varying
diameter, but the maximal diameter never exceeds that of the single metal
nanoparticle (Fig. 8). Beyond the percolation threshold (sample 4), the
nanoparticles form aggregates located on the boundaries between the
polymer spherulites (Fig. 9). Hence, the metal-polymer and metal-oxidepolymer
nanocomposites are the systems wherein the inorganic particles
form structured subsystems with respect to the polymeric matrix. In
increasing the metal content the nanoparticles localise along the borders of
polymeric spherulites accompanied with the formation of conducting
chain structures. Further increase in the metal content gives rise to
aggregation of nanoparticles and their coalescence.
103
104
Freshly synthesized Al-containing samples (Table 2) exhibit normally
a dark colour with a metallic luster in vacuum. On contacting the
composite with the ambient air sample 7 became transparent, whereas
samples 5 and 6 do not change their colour.
105
Electron diffraction patterns show a high degree of crystallinity of the
investigated samples. Figures 10-12 show the TEM micrographs of the
synthesized Al-containing nanocomposites and the particle size
distribution as observed by TEM. It was found that all the investigated
samples comprise rather homogeneously dispersed nanoparticles within
poly-para-xylylene globules.
106
107
Table 2 shows the crystal phase and the average particle size of Al
and comprising synthesized nanocomposites. The resulting
micrographs for sample 5 (Fig. 10) indicate that a composite containing
aluminium nanoparticles with an average size of 10nm and no other
crystal phases have been formed. On the contrary sample 7 (Fig. 12)
contains rhombohedral (corundum) with the crystal parameters
a=4,7592Å and c=12,992Å. Sample 6 consists of aluminium and alumina
nanoparticles embedded into the polymeric matrix (Fig. 11).
108
Electrical properties
Fig. 13 shows the temperature coefficient (the slope of
R(T)/R(20°C) versus Temperature dependency) of the electrical resistance
in vacuum of tin, copper and aluminium-containing as-prepared
composites for metal contents from 4Vol.% to 12Vol.%. The composites
with a metal content of 12Vol.% show a positive temperature-coefficient
indicating a conductivity determined by a continuous network with metallike
contacts between the nanoparticles. In contrast, when the metal
concentration is 8Vol.%, the temperature-coefficient becomes
semiconductor-like. This indicates a loss of metal-like contacts between
particles of the metal phase. This is to be expected from percolation
behaviour of the composites on the metal filler content. The percolation
threshold can be determined by the variation in the temperature
dependence of the electrical resistance, which is for the present case
10Vol.% of metal.
109
Optical spectra
The reference optical spectrum of the polymeric matrix measured
in the absence ofa filler (pure PPX, see Fig. 10) practically reproduces the
literature data [5]. A strong absorption is observed at For
a less intense band corresponding to absorption of PPX is
observed. Low-intensity absorption is observed at
decreasing with wavelength.
The inorganic filler entirely changes the absorption spectrum that
depends now on the filler concentration. The absorption peaks at 405, 530
and 760nm are observed for 8% metal particles in the polymer matrix
(sample 1, Table 1). When the metal concentration reaches 9% the peaks
move toward shorter wavelengths and the interpeak distances become
smaller. This shift is different for different wavelengths. For the peak at
405nm the shift is 25nm, for the peak at 530nm it is 45nm, and for that
centered at 760nm the shift is equal to 80nm. The broadest absorption
peak is observed within the wavelength range 1200 to 1800nm.
Absorption maxima are located at 1250nm (sample 1) and 1500nm
(sample 2). In further increasing the filler content the total absorption
increases in the spectral range of 300 to 1000nm. On surpassing the
percolation threshold (samples 3, 4) the spectral peaks become less and
less intense. At a filler concentration of 16% a mirror metallic film forms
at the sample surface. The changes in composite structure on a nanometer
scale are crucial for optical absorption. These changes reveal the
collective resonance effects in nano-ensembles distributed in a polymeric
matrix.
110
Oxidation behaviour
Substantial differences in oxidation behaviour exist between the
investigated samples. Figures 15-17 show the resistivity change of freshly
synthesized samples 5-7 if exposed to air at 1atm. According to the above
TEM results, nanoparticles in sample 5 are aluminium crystallites, which
are clearly reflected in the small increase of the electrical resistance (Fig.
15) during air exposure However, the electrical resistivity
of sample 7 increases dramatically during several seconds (Fig. 17). In
fact, nanoparticles in this sample are pure alumina dielectric material. The
high resistivity of the nanocomposite is caused by the high resistivity of
the alumina particles and large distance between them. TEM micrographs
of sample 6 revealed alumina and aluminium crystallites within the
polymeric matrix. The dramatic increase in resistivity is followed by sharp
decrease after 84s of air exposure (Fig. 16), which could be attributed to
new current path formation to by-pass areas with the high alumina
content.
111
Discussion
As shown earlier [7, 8], synthesized metal/polymer
nanocomposites have an open porosity, and all the nanoparticles are fully
accessible to oxygen. However, the experimental data show that the
112
equivalent oxygen accessibility does not lead to oxidation for some
composites.
Metal nanoparticles of sample 5 do not form a continuous net of
connecting particles; the average distance between nanoparticles is 50nm.
The nanocomposite resistance is due to subsequently connected
resistances of nanoparticles and polymer layers between them. Because
the sign of the temperature coefficient of the resistivity for sample 5
corresponds to the bulk metal, the nanocomposite resistance is a function
of particles resistance and polymer layer resistance could be neglected.
Thus, the interparticle charging and cooperative effects occur within an
ensemble of Al nanoparticles.
For sample 7 all the metal nanoparticles oxidize in air. The Al
metal oxidation is a highly exothermic process, so the polymeric matrix
needs to be heated. Considering the quite low glass temperature of polypara-xylylene,
i.e., 60°C [9], a high oxide nanoparticle mobility and
aggregation is proposed. This fact is reflected in the relatively large
particle size of alumina (55nm) within sample 7. The oxidizing process is
extremely fast; the resistivity increase is about 600%.
Sample 6 contains nanoparticles of two types, i.e., in type I,
particles are interconnected and lead to percolation, whereas nanoparticles
of type II form isolated clusters. During oxidation the latter react and
oxidize very fast, similar to sample 7. The average number of metal
particles to form one alumina particle is estimated to be 100-500.
Nanoparticles of type I do not oxidize in air because of the cooperative
effects discussed above.
Thus, the observed experimental data show that cooperative
effects define the chemical properties ofmetal/polymer nanocomposites.
SUMMARY
Samples of nanoporous composites of a metal and the polymer PPX
have been synthesised in the form of thin films. AFM studies show the
metal nanoparticles to have a size of 7 to 10nm. Within the composite the
polymer forms more or less spherical globules with a maximum size of
about 200nm. The interfacial regions between neighbouring polymeric
spherulites contain nanoparticles of inorganic filler. Depending on the
metal content in as-prepared samples, three different types of surface
morphology are distinguished, i.e., isolated nanoparticles localized on the
polymer spherulite surfaces, formation and growth of nanoparticle chains,
and spreading of interconnected chains.
TEM analysis of a nanocomposite with Al content above the
percolation threshold shows spherical pure metal nanoparticles with a
mean diameter of about 10nm, while below the percolation threshold the
composite contains agglomerates of rhombohedral (corundum) with
a mean size of55nm. A sample with a metal content just at the percolation
threshold contains metal nanoparticles of 10nm and alumina aggregates of
28nm in diameter. The inorganic phase is homogeneously dispersed
113
within the polymeric matrix in all of investigated samples. It has been
shown that the nanocomposite structure determines the oxidation
behaviour of Al nanoparticles within the polymeric matrix under air
exposure.
Measurements of electrical resistance as a function of temperature in
vacuum reveal metal-like conductivity of samples above the percolation
threshold, whereas samples below the threshold have semiconductor-like
conductivity. The percolation threshold ofthis system is evaluated.
The optical absorption spectra for the sample beyond the percolation
threshold differ from that of the sample with a nano-metal content below
the threshold.
Results clearly reveal that cooperative effects arising in the
metal/polymer nanocomposites in the vicinity of the percolation threshold
determine the electro-physical and chemical properties.
ACKNOWLEDGEMENTS
The authors are grateful to the Netherlands Organization for
Scientific Research (NWO grant #047.011) for financial support of the
project.
LITERATURE REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
Galyamov B., Zavyalov S., Zavyalova L, Morozova O., Kupriyanov L.,
Roginskaya Yu. Russian J. Phys. Chem. 1995; 69:971.
Vorontsov P., Gerasimov G., Golubeva E., Grogoriev E., Zavyalov S.,
Zavyalova L., and Trakhtenberg L. Russian J. Phys. Chem. 1998; 72; 1912.
Tsirlina G. A., Roginskaya Yu. E., Postovalova G. G., Vasil’ev C. A. (in
Russian). Electrochemistry 1999; 35:1380.
Nagaev E. L. (in Russian). Advances of Physical Science 1992; 162:49.
Takai Y., Calderwood J.H., Allen N.S. Makromol Chem Rapid Comm 1980;
Bl: 17-21.
Pomogailo A.D., Rosenberg A.S., Ufland U.E., Metal Nanoparticles in Polymers
(in Russian). Moscow, Chemistry, 2000.
Zavyalov S., Yablokov M., Pivkina A., Lushnikov A., and J.Schoonman.
Nanodispersed Metal Oxide-Polymer Films For Novel Solar Cells: Synthesis and
Characterisation. Proceedings ofAnnual ESF-NANO Meeting; 1999 10-11
December; Duisburg, Germany.
S.A. Zavyalov, A.N. Pivkina, and J. Schoonman. Formation and
Characterization of Metal-Polymer Nanostructured Composites. Proceedings of
Polymer Electrolytes Symposium (PES-2001); 2001 14-16 May;
Noordwijkerhout, The Netherlands.
Loeb W.E. Soc. Plast. Eng. J. 1971; 27:46.
NANOCRYSTALLINE LAYERS OF CdSe
PRODUCED BY MEANS OF A MULTILAYER
APPROACH
D. Nesheva1*
, Z. Levi1
, I. Bineva1
and H. Hofmeister2
1
Institute of Solid State Physics, Bulgarian Academy of Sciences, 72 Tzarigradsko Chaussee
Blvd., 1784 Sofia, Bulgaria
2
Max Planck Institute of Microstructure Physics, Weinberg 2, D-06120 Halle, Germany
Abstract
Multilayer structures are deposited by sequential physical vapor
deposition of CdSe and a second amorphous or polycrystalline
(ZnSe) semiconductor. The layer thickness of CdSe is varied between 1 and
10 nm; the layer thickness of the second material is either equal or twenty
times greater then that of CdSe. High-resolution electron microscopy results
prove the formation of CdSe nanocrystalline layers, which are continuous in
the structures with equal layer thickness of the constituent materials. In the
second kind of structures CdSe layers are discontinuous; particles with nearly
spherical form are observed, whose spatial distribution followed the surface
morphology of the previous film. The mechanism of nonepitaxial
semiconductor nanocrystal growth on ‘smooth’ and ‘rough’ surfaces is
discussed. Quantum-size increase of the optical band gap of CdSe
nanocrystals is observed in all multilayer structures prepared. It is ascribed to
one-dimensional carrier confinement in the continuous CdSe layers and quasi
three-dimensional one in the discontinuous layers.
1. INTRODUCTION
The development of various techniques for growing nanocrystals (NCs)
of a great variety of semiconductor materials has allowed intensive studies on
their unusual linear, non-linear and electro-optical properties (Klein 1994,
Bowen Katari, Cilvin and Alivisatos 1994, Murray, Norris and Bawendi
1993, Ekimov 1996), which result from both quantum confinement of charge
carriers and surface effects. Moreover, much attention has been addressed on
developing micro- and nanosized structures and devices using such NCs
(Grätzel 2000, Alivisatos 1998). Three-dimensional solid-state electronic
devices are giving way to new ones including multilayers with twodimensional
layers, nanowires, clusters, and porous nanocrystalline films.
Nanocrystalline elecronic junctions can be constituted by a network of
oxide or chalcogenide particles such as ZnO, CdSe, CdTe etc., which
are interconnected to allow carrier transport. Usually, a paste containing the
semiconductor NCs is applied on a glass coated with a transparent conducting
oxide; subsequent sintering produces a mesoporous film whose porosity varies
between 20% and 80%. Then the pores are filled with electrolyte or a solid
charge-transfer material. Important features of such mesoporous films are the
extremely large interface area which could be electronically addressable, and
appearance of carrier confinement effects for films constituted of quantum
dots. The illumination of nanocrystalline electronic junctions can be used to
generate electric current from light and, in particular, for solar energy
conversion (Grätzel 2000, Alivisatos 1998). It is generally accepted (Grätzel
2000 and references therein) that in mesoporous films charge carrier transport
occurs by diffusion, the rate ofwhich is controlled by traps. This distinguishes
the nanoparticle conversion devices from the conventional p-n or Schottky
junctions, in which the response to photoexcitation is governed by the electric
field in thejunction.
Electron-hole pair generation in photovoltaic devices occurs from the
absorption of photons whose energy exceeds the optical band gap of the
used semiconductor. This is among the reasons for the relatively low
effectiveness ofthe energy conversion in three-dimensional solar cells. Various
solutions of this problem have been suggested such as sets of consecutive
photoelements made of semiconductors with different optical band gaps
application of semiconductors with varying optical band
gap etc. In most semiconductor nanocrystals (CdSe, CdS, Si, Ge) the optical
band gap can be changed merely by changing their size between 1 and 10 nm.
Thus, nanocrystalline photovoltaic devices having photosensitivity in a wide
spectral region can be produced if one uses nanocrystals of various sizes in the
same device. Furthermore, for generation of electricity from light, it is
116
necessary that the minority charge carriers can diffuse to the interface before
recombination with majority charge carriers has occurred. In a nanocrystalline
device, it is possible to apply NCs whose size are smaller than the diffusion
length of minority charge carries.
In this chapter a technique for preparation of continuous and
discontinuous CdSe nanocrystalline layers in multilayer structures is
described.Amorphous or as well as polycrystalline ZnSe are used
as the second constituent material in these structures. High-resolution electron
microscopy measurements are carried out to prove the formation of CdSe
nanocrystalline layers. Results on optical absorption of the structures are also
presented.
2. SAMPLE PREPARATION
Multilayer structures of and ZnSe-CdSe were
produced in a vacuum chamber at residual pressure of Pa for the first
system and Pa for the others. A consecutive evaporation was carried
out from two independent tantalum crucibles mounted as far as possible apart
from each other. Powered CdSe and ZnSe (Merck, Suprapur) and preliminary
synthesized glassy or commercial SiO were used as source materials.
Two calibrated quartz microbalance systems were used in order to measure
nominal layer thickness and deposition rate of both materials. Their quartz
crystal heads were fixed above the respective source. Two kinds of multilayer
structures were deposited from each pair of materials: i) structures in which
the thickness of layers was approximately equal, now referred to as
multilayers, MLs with thicknesses varied between 2.5 and 10 run, and
ii) structures in which the nominal thickness of CdSe layers was varied
between 1 and 6 nm, while the layers of the second material were 20 times
thicker than the CdSe ones, now referred to as composite films, CFs. During
the deposition of each layer of MLs Corning 7059 glass and c-Si substrates,
respectively, kept at room temperature were rotated at a rate varying between
8 and 30 turns/min. Their residence time over the source is only 1/12 part of
the turn time and, hence, ten or more ‘layers’ with a nominal thickness of
~0.1-0.25 nm in each step form the respective layer in MLs. The substrates
were stopped at the greatest possible distance from both sources at the
moment the evaporation source was changed. During the preparation of the
composite films, CdSe layers were deposited in the same step-by-step manner,
while and ZnSe layers (20-120 nm thick) were deposited in a onestep
procedure. Some samples of each system were annealed for 60-90 min at
673 K; the composite films were annealed in air, while the
117
and ZnSe-CdSe ones were annealed in argon. More details on
preparation conditions of MLs and CFs can be found in our previous articles
(Nesheva and Levi, 1997, Nesheva, Raptis, and Levi, 1998, Nesheva, and
Hofmeister, 2000).
3. EXPERIMENTAL DETAILS
In order to study CdSe nanocrystallite size, structure and spatial
distribution, high-resolution electron microscopy (HREM) measurements were
performed using a JEM 1010 operating at 100kV and a JEM 4000 EX
operating at 400kV. Electron micrographs of and
three layer structures were recorded using optimum contrast
conditions (near Scherzer defocus).
All samples produced showed good photoconductivity, which allowed
us to carry out spectral photocurrent measurements in the 1.45-3.0 eV energy
region and thus to obtain useful information of the optical absorption of both
kinds of multilayer structures. Planar contacts (~10 mm long and ~1 mm
spaced) from In-Ga alloy or sputtered Au were used for the photoelectric
measurements on ZnSe-CdSe MLs and CFs. They showed linear currentvoltage
characteristics at fields of applied. Samples were
illuminated by a chopped (2 Hz) monochromatic light from a diffraction
grating monochromator MDR 2 at a spectral resolution of 4 nm. Intensity
dependencies of the photocurrent were measured in the entire spectral
region studied and the relation (Moddel, Anderson, and
Paul, 1980) was used in order to obtain the spectral dependence of the
absorption coefficient the subscript “o” refers to a reference energy. The
constant photocurrent method (CPM) (Vanecek, Kocka, Stuchlik, Ozisek,
Stika, and Triska, 1983) was also employed. Optical transmission of ZnSe
single layers and ZnSe-CdSe CFs was measured by a double-beam
spectrophotometer, Specord UV VIS.
4. RESULTS AND DISCUSSION
A HREM micrograph of a
three-layer structure annealed for 90 min at 673K is shown in Fig. 1. It is seen
from the micrograph that the CdSe layer is continuous; it consists ofrandomly
oriented nanocrystals. The size of CdSe nanocrystals along the ML axis is
equal to the layer thickness. The small angle X-ray diffraction measurement
118
119
carried out on an annealed MLs has shown (Popescu, Sava,
Lorinczi, Vateva, Nesheva, Tschaushev, Mihailescu, Koch, Obst, and
Bradaczeck, 1998) that these MLs exhibit a very good periodicity. Crosssectional
electron micrographs of and three layer
structures are shown in Figs.2a and 2b, respectively.‘Island’ type CdSe layers
are seen in both systems. It should be noticed that CdSe nanoparticles are not
disposed in a plane; their spatial distribution follows the surface morphology
of the and films, respectively. Most of CdSe nanoparticles are
interconnected. The higher magnification has shown (Nesheva and Hofmeister,
2000) that in as-deposited samples the particles have a nearly spherical shape.
They are crystalline with their lattice oriented randomly. The annealing
improves the spherical appearance and crystallinity of the nanoparticles. It has
also been obtained from the HREM results (Nesheva and Hofmeister, 2000)
that the average nanoparticle size is more than two times greater than the
respective nominal CdSe layer thickness. The annealing slightly increases both
average NC diameter and width of size distribution. A relatively narrow
nanocrystal size-distribution at half maximum
ofthe size distribution) has been obtained (Nesheva and Hofmeister, 2000) for
CdSe nominal thickness of 1 and 2 nm. Hence, the described deposition
technique allows fabrication of ‘island’ type CdSe nanocrystalline films with
various desired average nanocrystallite sizes by varying CdSe nominal
thickness.
Sequential vapor deposition of metals and insulating materials having
approximately equal thickness of several nanometers, has been applied for
non-epitaxial growth of nanosized metallic nanoparticles embedded in
insulating matrices. The partial wetting character of the metal with respect to
the insulator (Tonova, Patrini, Tognini, Stella, Cheyssac, and Kofman, 1999),
the existing valleys on the surface of the insulator (Drüsedau, Panckow, and
Klabunde, 1996) and the large difference of surface energies of the two
materials (Campbell, 1997) have been considered as responsible for the
formation of metal nanoparticles. The cross-sectional electron micrograph of
multilayers shows (Fig.1) that when in the MLs (Tonova,
Patrini, Tognini, Stella, Cheyssac, Kofman, 1999) the metal is replaced by a
semiconductor film, the latter is continuous rather than ‘island’ type.
However, when the surface is ‘rough’, the CdSe layers deposited are
discontinuous (Fig.2a). This implies that, probably, different mechanisms are
responsible for the metal and semiconductor nanoparticle formation; neither
the assumption for difference of surface energies, nor filling up of the existing
valleys could explain our observations. Besides, one could not expect the same
partial wetting character of CdSe with respect to and ZnSe.
It has been shown (Chopra 1972, Tersoff, Tu, and Grinstein, 1998) that
at the first stage ofthin film deposition on a not atomically smooth surface an
embryo formation takes place, which is affected by the surface curvature and
stress or even disorder of an amorphous surface. They create different
environments at different reaction sites and thus strongly affect reaction rates.
The HREM and X-ray diffraction results showed that the surface roughness of
layers in the MLs is relatively small. Therefore, we assume
120
that a homogeneous embryo formation could take place at the first stage of
CdSe deposition on the ‘smooth’ surface. A gradual increase of the
nanoparticle size and coalescence as well as new embryo formation can be
expected further; CdSe layers become continuous at a nominal thickness
The observed spatial distribution of the CdSe NCs in the composite
films (Fig.2) indicates that surface roughness plays an important role in the
CdSe nanoparticle formation. It is assumed that at the very beginning of CdSe
deposition on a ‘rough’ oxide or chalcogenide surface, embryos are formed at
those surface positions at which the curvature and lattice stress are the
greatest. The relatively narrow size distribution indicates that, most likely,
further CdSe deposition does not create new embryos but leads mainly to an
increase of the nanoparticle size.
Dark and photocurrent measurements have been carried out on all
multilayer structures prepared. The obtained results have shown that in both
MLs and CFs charge carrier transport involves networks of interconnected
CdSe nanocrystals. In the low-conductive CdSe nanocrystalline layers carrier
transport is controlled by potential barriers for electrons existing at the CdSeCdSe
interface (with a height <0.25 eV); in high-conductive CdSe layers their
effect is marginal (Nesheva, Levi, and Pamukchieva, 2000). Taking advantage
of the existence of current paths, optical absorption spectra of all kinds of
multilayer structures are obtained with a good accuracy by means of spectral
photocurrent and CPM measurements. Different size-induced increase of the
optical band gap has been observed in the continuous and discontinuous CdSe
layers of the multilayer structures, in which CdSe nanocrystal layers were
deposited on amorphous surfaces of and (Nesheva and Levi 1997,
Nesheva, Levi, Aneva, Nikolova, and Hofmeister, 2000). Thickness
dependencies of the optical absorption gap of CdSe nanocrystalline layers in
MLs and CFs are shown in Fig.3. It is seen from the figure that at
the same CdSe nominal thickness the increase in the discontinuous layer is
larger than that in the continuous one. This difference has been explained
assuming one-dimensional carrier confinement (along the ML axes) in the
MLs (Nesheva, Raptis, and Levi, 1998) and quasi three-dimensional one in
121
122
the CFs (Nesheva and Levi, 1997, Nesheva, Raptis, Levi, Popovic, and Hinic,
1999). The assumed difference in the carrier confinement regime appears at a
nominal CdSe layer thickness Obviously, at the
coalescence of nanoparticle leads to the formation of rather large CdSe
‘islands’.
It is known (De, Misra, and Ghosh, 1995) that ZnSe films deposited at
room temperature are polycrystalline with small crystallite size. In order to
investigate the effect of crystallinity of the second material on the electronic
properties of the nanocrystalline CdSe layers, ZnSe-CdSe multilayer
structures of both kinds were prepared as well as ZnSe single layers for
comparison. Optical transmission measurements were carried out on ZnSe
single layers having thickness of ~1 µm and applying Swanepoel’s method
(Swanepoel, 1983), the spectral dependence of the absorption coefficient
has been obtained. It is known that ZnSe is a direct gap semiconductor and for
direct band-to-band electron transitions in crystals the relation is
valid, while for amorphous semiconductors (Tauc dependence).
We plotted the dependence of ZnSe single layers using coordinates
and A straight line appeared in the first case, which indicated that,
123
in agreement with the results of other authors, the present ZnSe films are
polycrystalline; a value of 2.61 eV has been determined for the optical band
gap. Spectral dependencies of the absorption coefficient of ZnSe-CdSe MLs
and CFs with different nominal thickness obtained from photocurrent
measurements are shown in Fig.4 and Fig.5, respectively. The optical band
gap of ZnSe is significantly higher than that of CdSe and the
photoconductivity of CdSe single layers is orders of magnitude higher than
that of ZnSe ones. Therefore, it is assumed that the part of absorption spectra
at energies <2.5 eV is related to absorption in CdSe layers only. A ‘blue’ shift
of absorption curves is seen in Fig.4 if the CdSe layer thickness decreases
from 10 to 2.5 nm; no other features are observed in the ML absorption
spectra. The spectra of ZnSe-CdSe composite films look quite different. They
show a maximum at the end of the region of steep absorption increase and
some other features at higher energies. The first absorption maximum also
shows a ‘blue’ shift with decreasing CdSe nominal thickness. At the same
CdSe layer thickness the shift observed in CFs is stronger then in MLs
(compare Fig.4 and Fig.5). The optical band gap of semiconductor quantum
dots is determined from the energy position of this maximum. It is related to
124
electron transitions from the highest occupied energy level in the valence band
to the lowest empty level in the conduction band. These results are very
similar to those obtained for multilayer structures, which implies
that crystallinity of the second constituent material does not strongly change
both CdSe nanoparticle growth process and carrier confinement regime in
MLs and CFs.
5. CONCLUSIONS
A technique for preparation of CdSe nanocrystalline layers in multilayer
structures has been described. Amorphous or as well as
polycrystalline ZnSe have been applied as second constituent material.
Depending on the thickness ratio of materials, continuous or discontinuous
CdSe layers have been produced. It has been shown that neither crystalline
structure nor composition of the second constituent material strongly affects
CdSe nanoparticle growth; the surface roughness has been considered as the
main factor to control the nanoparticle growth process. Quasi threedimensional
carrier confinement has been observed in the discontinuous CdSe
layers; in the continuous CdSe nanocrystalline layers one-dimensional carrier
confinement, along the ML axes, takes place. The last observation is different
from results obtained in continuous nanocrystalline Ge, Si and GaAs films, in
which three-dimensional confinement and tunneling carrier transport
mechanism have been observed due to the existence of high potential barriers
at interfaces between nanocrystals. The suggested mechanism of non-epitaxial
growth of continuous and discontinuous CdSe layers at ‘low’ and ‘high’
surface roughness, respectively, implies that it is possible to apply the
described multilayer approach for fabrication of nanocrystalline layers from
II-VI semiconductors surrounded by a suitable solid charge transfer material.
This, along with the low interface potential barriers in CdSe nanocrystalline
layers of multilayer structures, opens the door for fabrication of
nanocrystalline electronic junctions suitable for various applications.
Alivisatos A.P., MRS Bulletin, 23 (1998) 18.
Bowen Katari J.E., Cilvin V.L., and Alivisatos A.P., J. Phys. Chem., 98 (1994) 4109.
Campbell C.T., Surf. Sci Rep., 27 (1997) 1.
Chopra K.L., Electrical phenomena in thin films, Mir, Moskwa, 1972.
De C.K., Misra N.K., and Ghosh T.B., Indian J. Phys., 69A (1995) 261.
Drüsedau T.P., Panckow A.N., and Klabunde F., J. Non-Cryst.Solids, 198-200 (1996) 829.
Ekimov A., J. Luminescence, 70 (1996) 1.
Grätzel M., in Handbook on Nanostructured Materials and Nanotechnology, Ed. Nalwa
H.S., Academic Press, 2000, vol.3, 527.
Klein L.C., Sol-Gel Optics-Processing and Applications, Kluwer Academic, Boston, 1994 .
Moddel G., Anderson D., and Paul W., Phys. Rev. B, 22 (1980) 1918.
Murray C.B., Norris D.J., and Bawendi M.G., J .Am. Chem. Soc., 115 (1993) 8706.
Nesheva D. and Levi Z., Semicond. Sci. Technol., 12 (1997) 1319.
Nesheva D., Raptis C., and Levi Z., Phys. Rev. B, 58 (1998) 7913.
Nesheva D., Raptis C., Levi Z., Popovic Z., and Hinic I., J. Luminescence, 82 (1999) 233.
Nesheva D. and Hofmeister H., Solid State Commun., 114 (2000) 511.
Nesheva D., Levi Z., Aneva Z., Nikolova V., and Hofmeister H., J.Phys.:Condensed Matt.,
12 (2000) 751.
Nesheva D., Levi Z., and Pamukchieva V., J.Phys.:Condensed Matt., 12 (2000) 3967.
Popescu M., Sava F., E.Lorinczi F., Vateva E., Nesheva D., Tschaushev G., Mihailescu I.N.,
Koch P.J., Obst S., and Bradaczeck H., Proc. SPIE, 3409 (1998) 964.
Swanepoel R. J., Phys. E: Sci. Instrum, 16 (1983) 1214.
Tersoff J., Yu.Tu, and Grinstein G., Appl. Phys. Lett., 73 (1998) 2328.
Tonova D., Patrini M., Tognini P., Stella A., Cheyssac P., and Kofinan R., J.Phys.:Condens.
Matter, 11 (1999) 2211.
Vanecek M., Kocka J., Stuchlik J., Ozisek Z., Stika O., and Triska A., Solar Energy Mater.,
8 (1983) 411.
REFERENCES
125
X-RAY DIFFRACTION FROM
NANOSTRUCTURED MATERIALS
J.Pielaszek
Institute of Physical Chemistry, Polish Academy of Sciences,
01-224 Warszawa, Poland
1. INTRODUCTION
Analysis of X-Ray diffraction (XRD) patterns from polycrystalline
materials, very often referred to as powders, is routinely performed. There
are many basic and advanced textbooks and review papers dealing with the
problem [1-7 and references cited therein]. The development of the methods
for the last fifty years is really impressive. Analysis of XRD patterns of
nanocrystalline and nanostructure materials may be brought down to the
question: are they simply polycrystalline materials but only with sizes of the
constituent diffracting structural elements in the nanoscale range and can be
studied by XRD using the same procedures?. And what, if any, are the limits
of applicability of these standard methods to studies of this kind of
materials?. Are they related to the very nature of nanocrystallinity or to the
limits of the experimental and computational techniques currently used?.
Some information concerning in general the XRD studies of polycrystalline
materials with emphasis on problems connected with nanocrystallinity of the
samples are presented. The discussed examples illustrate, although rather
difficult for the XRD studies, cases of nanocrystalline materials which, as it
is shown, can nevertheless be treated.
2. NANOCRYSTALS AND NANOSTRUCTURED
MATERIALS
Due to their already found or expected unusual properties, as compared
to the bulk, nanosized materials attract attention of many researchers [see
e.g. 8-10 and references therein]. There is, however, some confusion in
respect to the definition of different terms. Depending on the context,
authors attribute somehow different meaning to them. Nevertheless it can be
assumed that there is general consensus that nanomaterials are: clusters,
nanoparticles and nanocrystals. This is to be distinguished from
nanostructured materials. The latter are bulk solids composed of nanoscale
(or partly nanoscale) materials.
It is also generally accepted that nanocrystals are entities having linear
dimensions (in at least one direction) not exceeding about 100 nm. Thus thin
films and nanowires are included, although it is often accepted that the term
nanocrystal is reserved to objects having all the three dimensions in the
nanoscale range, the others being named either nanocrystals in one or two
dimensions or nanoscale materials, although sometimes the nano-thin films
are classified as separate kind of nanomaterials.
As used mostly in physical chemistry, the very convenient measure of the
size of nanomaterials is dispersion. It is defined as the percentage of the
atoms exposed on the surface to the total number of atoms in an average
cluster (nanoparticle, nanocrystal). Thus an individual atom has dispersion of
100%, and the cluster of 13 atoms (first coordination sphere in close-packed
structure) has dispersion 92%, as presented in Table 1.
When modeling, however, the shape of the nanocrystals are very often
approximated by cubooctahedra of “magic” numbers of atoms. They
correspond to the number of atoms in consecutive closed shells of a
cubooctahedron. Thus the dispersion calculated for both shapes (i.e. closed
by coordination spheres and closed shells) slightly differ as the definition is
clearly shape dependent and the value calculated for infinite number of
atoms in plane is still 100%. The value of dispersion is usually
experimentally obtained from chemisorption experiment and can be
compared with particle size only assuming specific shape of the particles
involved.
3. X-RAY DIFFRACTION PATTERN
FROM POLYCRYSTALS
The XRD method is a volume weighted one, i.e. the diffracted intensity
from any sample brings about information averaged over the whole
irradiated volume which depends on the geometric arrangements (e.g. width
and divergence of the X-ray beam) and physical properties of the sample
(absorption coefficient). The reader will recall that Transmission Electron
128
Microscopy (TEM), very often used in studies of nanocrystals, is a surface
weighted method.
By common definition polycrystals are highly divided solids composed of
coherently diffracting units of sizes down to few hundreds of nanometers. If
they are well crystallized their XRD patterns exhibit well resolved
reflections even at high diffraction angles. In a classical X-Ray diffraction
experiment on polycrystalline samples there are standard procedures of
measurements and data treatment. They are supplied nowadays with most of
the software when purchasing an XRD diffractometer. First, the data are
collected. Then they can be smoothed using any of the algorithms
(polynomial smoothing, spline smoothing, Fourier filtering etc.). Very often
the component of the radiation is stripped off (if the data are collected
using classical sealed-off X-Ray sources) and the background subtracted.
The subsequent treatment ofthe XRD pattern depends on its quality and aim
of the analysis. When the XRD reflections are well defined and in a great
part of the pattern well resolved, the phase identification (if necessary) can
be done using again, in most cases, a commercial search - match programme
equipped with appropriate database. From the intensity distribution of
particular reflections and/or their integral intensity, crystallite size
distribution and/or average crystallite sizes can be calculated. It is worth
mentioning that the XRD supplies information about average crystallite
size(s) which are domains of coherently diffracting X-Rays. This should not
be confused (but often is) with particle size(s) which may be entities
composed of many crystallites separated by different kind of grain
boundaries.
4. X-RAY DIFFRACTION PATTERN
FROM NANOCRYSTALS
It is well known that with diminishing crystallite size the measured XRD
pattern exhibits broadened, and very often overlapping reflections. The
broadening of the reflections is inversely proportional to the crystallite size
(i.e. size of coherently diffracting domains). This relation is known as
Scherrer’s [2,11] equation:
It relates, for the given wavelength the average crystallite size D,
diffraction angle of a particular reflection and broadening of this
reflection due to crystallite size effect. The is commonly expressed as a
full width at half maximum (FWHM) of the reflection (background
subtracted), corrected for instrumental broadening. The constant k, close to
1, depends on the assumption on the crystallite shape and the reflection used
[12]. Since the XRD method is a volume weighted, the physical meaning of
129
the FWHM is not obvious and although in many instances its use gives good
results, nevertheless it is recommended to use the integral width of the
broadening defined as:
The integral is the integral intensity of the particular reflection (measured
above the background) and the is the intensity at the maximum of this
reflection. The value D calculated both ways is more precisely called
“apparent crystallite size”. The observed experimentally intensity
distribution of a particular reflection is a convolution of instrumental
broadening function (which normally is assumed to include spectral
distribution function) and crystallite – induced broadening (size, stress,
strain). It should be, however, pointed out that introduction of the
instrumental correction is much easier in calculation of then In the
former case as instrumental profile usually the XRD pattern of a standard
polycrystalline sample with enough big and strain/stress free crystallites is
recorded and FWHMs are measured for particular reflections. Then
assuming the same analytical shape (very often gaussian or lorentzian) of the
reflections for both standard and measured samples the corrections are
introduced [see e.g. 1]. If the FWHMs of measured and standard profiles are
B and b respectively, then for the simplest case of the assumed gaussian
profiles the is expressed as Other assumed analytical shapes of
the reflections can also be taken into account [see e.g. 13]. However, to
calculate the deconvolution of the observed and instrumental profiles (e.g.
using Stokes’ method [14] of Fourier analysis) is needed, making procedure
more troublesome. It has nevertheless the advantage that the resulting
broadening profile can directly be used for particle size and strain
determination using Warren and Averbach method [2]. This method, making
use of Fourier analysis of the intensity distribution in reflections of different
orders from the same set of diffraction planes allows separation of size and
strain effects and calculation of “crystallite size” distribution. Under some
assumptions the calculation of this distribution is also possible from profile
of one reflection [15] and also employing set of reflections from different
diffraction planes. At this point it should be remembered that one of the
main features of the diffraction pattern from nanocrystalline material, as
compared to polycrystalline one, is a quick decrease of the diffracted
intensity with diffracting angle, overlapping of the reflections and thus
practically making difficult to have well defined distribution of intensity for
higher order of reflections. The use of parenthesis above stresses the fact,
shown long time ago by Berthaud [ 16 ], that the total diffracted intensity for
a given Bragg reflection from a crystallite is the sum of independently
diffracted intensities by each of the unit-cell columns making-up the
crystallite. It means that the calculated size distribution using the Warren-
130
131
Averbach method is in fact a distribution of diffraction column lengths in a
given crystallographic direction perpendicular to the diffraction planes and
not of crystallite (coherently diffracting domains) sizes. Theoretical
considerations show that the interference function of a polycrystalline or
nanocrystalline solid is identical to that of an arrangement of isolated
particles with the same size or size distribution as those of the
polycrystalline or nanocrystalline solid [17]. Thus the values of D resulting
by using Scherrer’s formula are solely an estimate of a volume-weighted
average column length. This explains why the term “apparent crystallite
size” is often used. The average particle size or crystallite size distribution
can be calculated from the average column length or column length
distribution provided the shape of the crystallites is known [see e.g. 18]. The
definition ofthe “true size” of the crystallites is not unambiguous. When all
crystallites are assumed to have the same size and shape, the cube root ofthe
volume of one crystallite defines the size. Distribution of crystallite sizes
will affect this definition. In this case the “true size” can be defined either as
a mean value of the cube roots of the individual crystallites volumes or as a
cube root of the mean value of volumes of the individual crystallites.
Expressions for effective Scherrer’s constants when there is a distribution of
crystallite sizes for two definitions of “true crystallite size”, are given in
[12]. The general considerations on different definitions can be found
elsewhere [e.g. 18]. The problem of the influence of crystallite shape on the
intensity distribution ofreflections was considered by many authors.
Both the shape and size of nanocrystal affect the interference function. It
represents only the pure diffraction effects. Only when convoluted with
instrumental and spectral functions, it is what we are recording as an XRD
pattern. In Fig. 1 interference function is presented for different shapes of
randomly oriented nanocrystals of ideal FCC lattice (palladium) of about
400 to 800 atoms.
Whereas for regular shapes (cube, sphere) the interference functions are very
similar, they are significantly different for crystallites with flat or elongated
cylindrical shapes of approximately the same number of atoms. There were
many calculations ofthe intensity distribution in the reflections and value of
constant k in Scherrer’s equation depending on the crystallites shape. The
detailed analysis of the intensity distribution from nanocrystals in form of
cylindrical prism and hexagonal prism [19] shows that it is possible to
distinguish between the broadening from both kinds of shapes. In principle it
is only necessary to record two lines to determine the average dimensions of
the average cylinder (prism) provided that their orientation with respect to
the crystallographic axis is known or can be assumed. This can very often be
the case of materials which belong to hexagonal system. Analysis of an
annealed ZnO powder revealed that the crystallites are hexagonal prism with
an average height of 21.3 nm and edge length 8.7 nm and z axis parallel to
the axis of the prism and x or y parallel to an edge and to differentiate them
with high degree of probability from cylindrical ones. The measurements
performed on nanocrystalline Pd samples obtained by gas condensation and
compacted showed that average crystallite size in [h00] direction was about
8 nm whereas a larger average crystallite size of about 16 nm was obtained
from measurements of(hhh) reflections [20].
5. XRD PATTERN FROM
NANO-STRUCTURED MATERIALS
XRD pattern from nanostructured materials is, in principle, similar to
that from nanocrystalline materials. If the lattice orientations of particular
entities forming a solid, nanostructured material are distributed at random,
the interference functions are identical and they sum up. There are, however,
additional factors which should be considered. First these materials can be
multiphase. The phases can all be nanocrystalline or a nanocrystalline phase
can be embedded inside a polycrystalline and/or amorphous matrix. Thus
besides the influence of nanocrystal size, also the grain boundaries, defects
and microstrains (being the result of local plastic deformation) are to be
considered. Simple modeling of XRD pattern from material with different
density and arrangement of dislocations in the grain boundaries show that
the different methods of its analysis (Scherrer’s equation, Williamson - Hall
method [21], Warren-Averbach [2]), based on different assumptions can
132
bring about significantly different results [22]. In many instances the method
of material preparation and its history can suggest which factors can mostly
influence the observed diffraction pattern. The measurements of XRD
profiles on nanocrystalline palladium [17], depending on the history of the
preparation of the sample shows that there is a different structure of grain
boundaries and the number of atoms at the grain boundaries being not on
crystal lattice side, i.e. in disordered state. A particular example of
nanostructured materials can be supported metal catalysts in which the
prevailing part of a mass is a support, and interest is mostly in the structure
ofa small fraction ofthe highly dispersed (nanocrystalline) metal. Due to the
high specific surface of the support and usually low loading of the metal, the
nanocrystals of the metal phase are distanced each from other. The XRD
pattern ofthe catalyst is usually a superposition of two: from the metal phase
and, usually predominant, from the support. The contribution to the pattern
originating from the distances between the support and the metal in most
cases can be neglected. The simplest case can be when the background is
from amorphous (or quasi-amorphous) phase which can be e.g. for silica. In
such a case the background can easily be modeled with smooth analytical
function and subtracted from the measured pattern and the resulting profile
fitted to an analytical function [23]. This way it was possible to follow for
Pd-Au nanocrystalline palladium catalyst (on silica support) changes in the
133
positions of XRD reflections caused by series of reduction-oxidation cycles
[24]. It was shown that for nanocrystallites of about 10 nm in diameters and
nominal composition of 70 at.% Pd – 30 at.% Au there is a shift in the
position of reflections which could be interpreted via Vegard’s law as the
apparent average increase of the concentration of palladium in the alloy by
about 3 at.% due to the reduction ofthe sample in hydrogen. It means that in
air the surface layer of these bimetallic nanocrystals is enriched in
palladium. In addition to immediate application of these results in the
interpretation of catalyst behavior it is worth to indicate the more general
aspect; the XRD being, as mentioned before, the volume weighted method
can be, for the case of nanocrystalline materials, also the surface sensitive
method. This aspect of XRD is especially interesting when combined with
atomistic modeling (see below).
Another example of using fitting of analytical functions to the bi-phase
system XRD pattern is shown in Fig. 2. This is one of the XRD patterns
measured for a series of palladium on samples with concentration
of palladium varying from 0.3 wt.% to about 2.7 wt.%. For the presented
case of 1.46 wt.% of palladium the palladium phase is hardly visible on the
XRD pattern of this sample as compared with XRD pattern from the
However the proper scaling of both patterns and subtraction of
support [23] gives the XRD pattern originating from the metal alone.
Analytical function fitting (Pearson VII in this case) allowed calculation of
the average crystallite size using Scherrer’s equation employing integral
breadth and resulted in the estimated size of 3 nm.
6. MODELLING OF XRD PATTERN
There are, in general, several methods of modeling XRD pattern. The very
first step is to fit an analytical function to distribution of intensity of a
particular reflection and subsequently extract from the fitted values
(position, intensity, widths) the required physical information. As it was said
before there are different analytical functions which approximate well,
depending on the case, the experimentally observed intensity distribution of
a reflection. This method can be used even when there is overlapping of
individual reflections. In principle, this is a method of decomposition of a
measured XRD pattern into constituent individual reflections, fitted with
analytical function(s). Therefore it is often referred to as the decomposition
method. One of the important drawbacks of this method is that there are
arbitrarily defined truncation limits for given reflection(s). These are usually
chosen as points on both sides of the reflection position being apart by a
multiplicity of the FWHM value of the reflection on both sides of the
ordinate of the reflection position. In the case of overlapping and low-
134
intensity reflections (particularly at high diffraction angles) this can pose a
serious problem. Due to the different shape of the tails of different analytical
functions it can be shown [25] that to include 99% ofthe integral intensity it
is sufficient to have for Gaussian function the truncation at the distance of
3xFWHM on each side of the maximum, whereas for Lorentzian function
this should be 63xFWHM. Therefore acceptable (especially from the point
of view of overlapping reflections), intermediate, although arbitrary, values
are usually chosen. An example ofthe decomposition procedure is presented
in Fig. 3.
The nanocrystalline palladium (10 wt.%) on silica was subjected to reduction
in hydrogen at 873 K and it was concluded [26] that phase is formed.
Closer examination of the XRD profile as presented in Fig. 3 by open circles
(data after subtraction of silica background) showed that there are probably
more phases. So the pattern was decomposed into minimum number of
Pearson VII function to obtain the best fit and the result is presented in the
middle part of the figure. The lower part ofthe figure presents histograms of
possible Pd-Si phases. The solid line superimposed on XRD profile is a sum
of the decomposed profiles. As it can be seen the fitting is quite good,
supporting conclusion that in fact there is a mixture of different Pd-Si
phases. The estimated crystallite sizes resulting from the fitted parameters
are less than 10 nm.
6.1 Total pattern fitting
Fitting of superimposed individual analytical profile functions (as
described above) together with suitable background function model to the
observed data is one of the methods of the total profile fitting. The resulting
set of profile parameters can be used for calculation of physical parameters.
There are, however, substantial difficulties with application of this method to
nanocrystalline materials. In general the XRD pattern from these materials
becomes increasingly difficult to fit with increasing diffraction angle
(decreasing values of d spacing). This is due, besides the intrinsic crystallite
size effect, to the influence of thermal atomic vibrations and decreasing
values of atomic scattering factor. As a result the reflections become more
and more difficult to be distinguished from the statistical noise of the
background. The procedures of XRD pattern treatment and our
understanding of their shapes are all the time being developed. One of the
problems still to be solved is to obtain the real crystallite size distribution
from XRD. It is now commonly accepted that the particle size distribution of
materials obtained by vacuum deposition or gel techniques is close to
lognormal. Assuming this and the shape ofthe crystallites, it was shown [29]
that even for very low crystallite sizes (ceria samples with average apparent
crystallite sizes of 3-4 nm) it is possible to model the observed XRD
profiles.
135
136
6.2. Rietveld refinement
The method which in the last decade underwent a vivid development and
increasing applicability in solving different structural problems is the
Rietveld method. This is a total pattern fitting method introduced by
Rietveld for fitting structural parameters obtained from neutron diffraction
data [30], but soon applied to XRD data and now very popular and effective
in XRD data analysis [31,32]. The experimentally observed XRD pattern is
fitted to the calculated one. To calculate the XRD profile two models have to
be known (or assumed): the model of approximate positions of atoms in the
structure and the model describing the Bragg reflections (their shape) in
terms of analytical functions. Then the calculated XRD profile is compared
with the measured one. The total intensities of reflections and their (starting)
position are defined by structural models. Parameters describing the form of
Where are the coordinates of the I-th atom in the cell, expressed as
the fraction of the edges; h, k, 1, are the Miller indices; is the fractional
occupancy for the i-th atomic site; is the X-ray scattering factor and is
the temperature (Debye – Waller) factor. Function describes the shape of
reflections (its parameters being also angle and, in general, lattice direction
dependent). The summation is over all reflections which contribute to the
intensity at It means that the minimization of (2) involves directly the
assumed (starting) structure of the phase to be considered. The parameters
are fitted by a non-linear gradient minimization till the minimum values of
the function S (2) are obtained. The final results of the goodness of fit are
estimated on statistical ground (see e.g. Table 1 in [33]). Interpreting the data
only on the basis of these parameters can sometimes be misleading. As an
example, examination of a series of Li-Mn-O spinels [34 ] can be presented.
Some of the spinels, prepared by low-temperature sol-gel synthesis were
nanocrystalline with average crystallite sizes, as estimated from FWHM of
the (111) and (311) reflections, equal to about 17.0 nm. The quality of the
Rietveld fit of the (non-stoichiometric composition
was good when fitted to the structure of Fd3m symmetry
and lattice constant corresponding to the phase. Closer
examination of the shape of the resulting fitted profile at higher angles of
137
the Bragg reflections depend on the instrumental functions and on the
microstructure of the sample (e.g. crystallite size).
The quantity to be minimized is:
Where the and are the calculated from the model and
measured intensities at the i-th point of the pattern, respectively, the sum
being over all data points in the diffraction pattern. The is expressed
by:
where is the integrated intensity of the j-th reflection, expressed as:
With s being the scale factor, – the multiplicity, – Lorentzpolarization
factor and is a preferred-orientation function. is
background at the point
is the structure factor given by:
138
diffraction (see Fig. 4 ) showed however that there is noticeable discrepancy
between the observed and fitted profiles, especially for some reflections.
These were the low intensity reflections due to the intrinsic nature of the
XRD pattern from nanocrystalline substances (see above), but nevertheless
the discrepancies were noticeable indicating the possible presence of
additional weak reflections. Thus the fitting was repeated taking into account
this possibility of multiphase composition. The best agreement between the
shapes of measured and fitted profiles (see Fig.5) was obtained for three
phase system. The dominant phase was the one identified before (about
76%), but two others were also tentatively identified: stoichiometric, high
temperature spinel (about 17%) and Mn-O oxides. It should be pointed out
that the parameters describing goodness of fit were very similar for both
fittings (e.g. the R-Bragg factor was for one and multiphase fitting 4.6 and
4.82 respectively).
6.3. Atomistic modeling.
The interference function which reflects solely the effect of the positive
and negative interference of the incident and diffused X-Ray beams on
individual diffracting centers (atoms) is calculated using Debye formula [see
e.g. 5]. As it was shown in Fig. 1 the interference function from nanocrystals
of few nm in size exhibits well defined reflections of widths depending on
the linear dimension of the nanocrystal in a given direction. With increasing
crystallite size the XRD pattern more and more resembles that from
conventional coarse grain polycrystalline material. In most cases the material
under study is not a monoatomic one. In the case of metals very common are
bi- or multimetallic systems. For the simplest case of a solid solution of two
elements it is usually assumed that the change of lattice parameter (and thus
position of reflections) of the matrix element depends linearly on
concentration of the solute element randomly distributed in the matrix. This
relationship known as Vegard’s law is often observed and exploited in
practice for studying polycrystalline materials. It is however known that for
many multielement systems there is a preferential surface segregation of one
of the elements, often observed in surface studies of single crystals and in
gas adsorption probing of different systems. For large crystals this surface
enrichment does not influence noticeably the average bulk concentration.
For nanocrystalline materials the situation can be, however, different. Due to
a significant proportion of the surface atoms in respect to the bulk, the
surface segregation of one element can substantially disturb the average
concentration inside the nanocrystal. The calculations [35] of the XRD
interference function were performed for a nanocrystal of f.c.c. structure and
shape close to sphere, having about 2000 atoms (i.e. about 3 nm in diameter)
and consisting of 25 at.% Pd – 75 at.% Co. Different concentration profiles
in the clusters were considered. In the simplest case of an ideal Pd lattice
imbedded in an ideal Co lattice the resulting XRD pattern was a simple
superposition of the two, the interphase intensity contribution being not
noticeable. When, however, it was assumed that the outer layers of Co atoms
are arranged in such a way as to gradually accommodate their distances to
those ofthe underneath layer (the model of complete segregation or epitaxial
139
140
growth with distortions in Co-Co distances), the positions of reflections in
the resulting XRD pattern were almost identical with these of pure Pd
(without any presence of reflection from Co phase). Changing Co profile
concentration inside the crystallites resulted in changes in the shape and
position of the calculated reflections but nevertheless their positions were
very distant from what can be expected from Vegard’s law and indicated
much lower than actual concentration of the Co atoms. This example show
that for multi-element
(but also multi-phase) nanocrystalline materials, the positions and intensities
of the XRD reflections do not necessarily reflect the nominal composition
but strongly depend on other factors (e.g. surface segregation which in turn
can depend on the environment). Even in the case of one element
nanocrystalline materials XRD patterns, exhibiting distinct reflections, their
interpretation can be ambiguous. For the nanocrystals of sizes down to 2 – 3
nm we observe that the lattice parameters calculated from different
reflections (and corrected for errors) give different values. Thus the indexing
ofthe reflections makes a limited sense. As was pointed out by Kaszkur [36]
“we depart from the Bragg interpretation zone”. When modeling nanocrystal
(cluster) with minimum energy configuration, one can get more “realistic”
structure because the presence of the free surface (or interphase boundary)
can induce the displacement of atoms from their lattice sites. In energy
minimization procedure, of crucial importance is a choice of the interaction
potentials. It was shown [37] that modified Sutton-Chen [38] potentials work
well for some transition metals (palladium, cobalt). Thus the structure of the
energetically minimized clusters can be modeled and the resulting XRD
pattern calculated. To make however the simulated XRD pattern more close
to the real (experimental) one, which is measured over a definite period of
time, the Molecular Dynamics (MD) run on energy minimized nanocrystal
should be done and then the resulting XRD pattern (averaged over the runs)
presented. This way it was shown [36] that for Pd nanocrystals the overall,
average lattice constant decreases with decreasing crystallite size. Fig. 6
presents the 561 atom cubooctahedral cluster and calculated XRD pattern
(solid line) together with the XRD pattern calculated for the energy
minimized cluster (dashed line). The shifts in the position of the reflection,
increasing with the diffraction angle, are visible. It means that if the
influence of the environment on the positions of surface atoms is significant,
this should be reflected in the average position of the XRD reflection. It was
in fact shown [39] that there is a shift of Pd(111) reflection when palladium
nanocrystals (palladium catalyst on support) is subjected to change of the
atmosphere from air to argon. Moreover, the atomistic modeling ofthe XRD
pattern of palladium nanocrystals showed that to explain the change in
position and shape of the pattern when transforming palladium into its
hydride it should be assumed that part ofthe nanocrystals has the icosahedral
symmetry and these nanocrystal do not undergo transformation into hydride.
Thus the atomistic modeling when compared with experiments can bring
about the information otherwise not possible to retrieve.
7. CONCLUDING REMARKS
The presented above considerations and results show that for studying
nanocrystalline materials the methods usually used in XRD studies of
polycrystalline materials are applied. This include also the modeling of the
XRD profiles and whole pattern fitting. Nanocrystallinity of the materials
needs however some special attention in experimental procedures and data
treatment. In the former case it is especially important when high dispersion
and properties of the material make them sensitive to the environment. This
make the handling of the samples more difficult, but in return very minute
changes in the surface structure of nanocrystalls can be followed up,
especially when aided by atomistic modeling. It can be speculated that this
opens the way for studying by XRD the surface reactions on nanocrystalline
materials. It should be finally mentioned that the combination of the XRD
method with other physical methods of studying nanocrystalline systems,
especially EXAFS should push forward our understanding of their structure
and perhaps predict some oftheir properties.
141
Klug H.P. and Alexander L.E., X-Ray Diffraction Procedures for Polycrystalline and
Amorphous materials. New York: Wiley, 1974.
Warren B.E., X-Ray Diffraction. New York: Dover Publ. 1990.
Louër D., Advances in X-Ray Analysis, 37 (1994) 27.
Louër D., J. de Phys. IV, (Suppl. J. de Phys. III), C4 6 (1996) 57.
Klug H.F., Alexander L.E , X-Ray Diffraction Procedures. John Wiley, 1974
Jenkins R., Snyder R.L., Introduction to X-Ray Powder Diffractometry. John Wiley, 1996
Delhez R., deKeijser Th.H., Mittemeijer E.J., Fresenius Z.Anal.Chem. 312 (1982) 1
Morris D.G, Mechanical behavior of nanostructure materials. Ueticon-Zuerich, Trans
TechPub., 1998
Nanomaterials: synthesis, properties andapplications. Ed. by Edelstein A.S and
Cammarata R.C, Bristol, Philadelphia: Institute of Phys. Pub., 1998
Metal clusters in chemistry. Volume 3, Nanomaterials andsolid-state chemistry. Ed. by
Shihe Yang and Ping Sheng, London, New York: Taylor & Francis, 2000.
Scherrer P., Nachr.Ges.Wiss. Goetingen, Math.-Phys. Kl. 2 (1918) 96.
Langford J.I. and Wilson A.J.C., J. Appl. Cryst. 11 (1987) 102.
de Keijser Th.H., Langford J.B., Mittemeijer E.T., Vogels A.P.B., J.Appl.Cryst. 15
(1992) 308.
Stokes A.R., Proc. Philos. Soc. London, Ser A, 61 382 (1948)
Nandi R.K., Kuo J., Schlosberg W., Wissler G., Cohen J.B., Crist B. Jr, J.Appl. Cryst. 17,
22(1984)
Berthaud E.F., Acta Cryst., 3 (1950) 14
Löffler J. and Weissmüller J., Phys.Rev B. 52 (1995) 7076.
Matyi R.L., Schwarc L.J., Butt J.B., Catal.Rev.Sci.Engn. 29 41 (1987)
Langford J.I., Louër D., J.Appl.Cryst. 15 (1982) 20.
Fitzsimmons M.R., Eastman J.A., Müller-Stach M., Wallner G., Phys. Rev. B 44
(1991) 2452.
Williamson D.K. and Hall W.H., Acta Metall. 1 (1953) 22.
Alexandrov I.V., Materials Sci. and Engn. A 286 (2000) 110
J.Pielaszek, X-Ray Diffractometry in Supported Catalyst Studies, Institute of Physical
Chemistry Monogr. Series, Warszawa 1995.
J.Pielaszek, in Applied Crystallography, Proceedings of the XVI Conf. On Appl. Cryst,
Cieszyn, Poland, 1995, H.Morawies, D.Stroz (Eds), World Scientific 1995, p.21
Toraya H., J.Appl.Cryst. 18 (1985) 351.
Juszczyk W., Pielaszek J., Karpinski Z., Sobczak J.W., New J. Chem. 17 (1993) 573.
Anderko K., Schubert K., Z.Metallkd. 44 (1953) 307.
Wysocki J.A., Duwez P.E., Metall. Trans. A12 (1981) 1455.
Langford J.I., Louër D., Scardi P., J.Appl.Cryst.33 (2000) 964.
Rietveld H.M., Acta Cryst., 20 (1966) 508, 21 (1966) A228
The Rietveld Method, Ed. By Young R.A., International Union of Cryst. Book Series,
Oxford Univ. Press, 1993.
McCusker L.B., Von Drelle R.B., Cox D., Louër D., Scardi, P., J.Appl. Cryst. 32 (1999)
36.
Langford J.I., Louër D., Rep. Progr. Phys. 59, 131 (1996).
142
REFERENCES
ACKNOWLEDGMENTS
Dr Z.Kaszkur is acknowledged for supplying Fig.6 and permitting its
publication. The work was supported in part by the Polish Committee for
Scientific Research (KBN) grant No PBZ-KBN-013/T08/12.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
Lisovytskiy D., Kaszkur Z., Baumer V.N., Pielaszek J., Marzec J., Molenda J., Dygas J.,
Kopec M., Krok F., Proceedings of the European Powder Diffraction Conference,
Uppsala 23-26 may 2002, Sweden , to be published in Materials Sci. Forum.
Pielaszek J., Barczynska J., Mat.Sci Forum, 79-82 (1991) 139
Kaszkur Z., J.Appl.Cryst. 33 (1999) 87.
Kaszkur Z., Mierzwa B., Phil.Mag. 77 (1998) 781.
Sutton A.P., Chen J., Phil. Mag. 61 (1990) 139
Kaszkur Z., J.Appl.Cryst. 33 (2000) 1262.
34.
35.
36.
37.
38.
39.
143
HIGH RESOLUTION ELECTRON MICROSCOPY OF
SURFACES AND INTERFACES
H. W. Zandbergen
National Centre for HREM, Laboratory of Materials Science, Delft
University of Technology, Rotterdamseweg 137, 2628 AL Delft, The
Netherlands
Introduction
Compared to other techniques used to obtain structural information
of materials, high resolution electron microscopy (HREM) has the great
advantage that it yields local information about the atomic arrangements,
projected along the direction of electron incidence at a resolution
comparable to the interatomic distances. This enables the study of
complicated structures, crystal defects, precipitates, etc. down to the
atomic level. Furthermore the electronic structure can be probed by
electron energy loss spectroscopy (EELS), yielding information on the
local environment of the atoms in the illuminated area.
Knowledge of the atomic arrangements of surfaces and interfaces is of
vital importance for the understanding of the properties. With the
increasing need of the accurate determination of the atomic arrangements
at non-periodic structures in materials design and control of
microstructures and nanostructures, techniques that provide quantitative
structural information at the sub-angstrom level will be indispensable.
High-resolution electron microscopy (HREM) is by far the best technique
to obtain sub-surface structural information on a local scale. X-ray and
neutron diffraction can only give structural information that is averaged
over a large area. This is in part because the interaction between electrons
and matter is much higher than for X-rays and neutrons (about to
times respectively) and in part because electrons can be focussed very
easily by magnetic lenses. The local 3-D information is essential for
understanding the properties of all types of materials, since the local
structure plays a paramount role in this respect. In particular for further
improvements of these materials based on materials design, accurate atom
positions are required, with a precision of the order of 0.01 nm. Scanning
tip microscopy can provide the resolution but only 2-D information is
obtained in this way. Although the resolution of HREM is about 0.1 nm,
the scattering potential of atoms is known which allows a position
accuracy of 0.003 nm provided the atom columns in the viewing
direction are sufficiently separated. HREM gives 2-D information about
146
the atom positions, but by combination of HREM information from
different directions a 3-D reconstruction can be made. Thus HREM can
provide the required accuracy. However, at the moment this accuracy
cannot be obtained because of the aberrations imposed by the optics of
the electron microscope. In this respect, important steps towards
quantitative HREM are the developments of through focus exit wave
reconstruction and off-axis electron holography. Both methods provide
the complex image wave in the image plane allowing proper aberration
correction.
A major advantage of the reconstructed exit wave (corrected for the
aberrations) is that it allows a more straightforward quantitative
interpretation than HREM images, in particular for non-period features
like grain boundaries. Whereas grain boundaries play a large role in most
materials, their role is even larger in nanophase materials, which consist
of mostly synthetic materials with an average grain size below 100 nm
[Siegel, 1993]. The various grains can be of different phases or structures,
but the basic idea behind nanophase materials is the generation of
materials with new atomic structural arrangements by introducing of a
very high density of defects such as grain boundaries, interphase
boundaries and dislocations resulting in new macroscopic properties. As
divers as the compounds and the applications, are also the actual reasons
for using electron microscopy techniques. Evidently, size and shape can be
easily determined by TEM, whereas small angle X-ray and neutron
diffraction allow determination of the size distribution. Still, in dense
nanophase material the grain boundary structure (Merkle (1995), Möbus
(1998), King (1998) and Kienzle (1998)) is the most important part.
Determination of the grain boundary structures is, however, a difficult
task because it requires a 3D determination of the whole grain boundary to
be done by making 2D images, which are in a first approximation
projections. Because of that, the 2D imaging requires that the grain
boundary is (almost) parallel to the electron beam, without steps in that
direction and being continuous throughout the specimen. Also the grains
on both sides of the boundary should be in such an orientation that the
structures of both grains are imaged. Finally the specimen should not be
covered with an amorphous layer. Since the boundary core structure
depends on the interatomic bonding forces and the boundary
crystallography and since many different grain orientations and grain
boundary orientations occur, a large variety of grain boundary core
structures will occur in a given specimen. Therefore a HREM study can
only provide fragmented information, describing only a part of the grain
boundary structures. Obviously, the higher the resolution the more grain
boundaries can be characterised and the higher the precision will be.
147
The imaging process
Due to the aberrations of the electron microscope, the information
coming from one point in the object is delocalised in the image. The
amount of delocalisation depends on the spatial frequency (and on the
microscope). One can reduce this delocalisation by defocusing of the
microscope, but only up to a certain spatial frequency. When we only
consider the phase contrast of perfect-periodic specimen, the image
reveals at Scherzer focus directly the projected potential, i.e. the structure
of the object, provided the object is very thin and the atom columns are
so far separated that they are not masking each other by the blurring. The
resolution which can be obtained in this way (point-point resolution) is
about 0.17 nm for the best microscopes at intermediate acceleration
voltage (100 to 200 kV). For non-periodic specimen like defects and grain
boundaries, the delocalisation (see Figure 1) due to aberrations prevents a
direct interpretation of images of these structures. The information
beyond the point resolution is transferred with a non-constant larger
phase shift and, as a consequence, is dispersed over a larger image area.
148
The information limit corresponds to the maximal diffraction angle
that is still transmitted with appreciable intensity, i.e. the transfer
function of the microscope is a spatial band filter such that all
information beyond the information limit adds to the background. For a
thin specimen, the damping envelope of chromatic aberration (temporal
incoherence) mainly determines this limit. A typical value of the
information limit for the best microscopes operating at 300 keV (with
field emission gun) is slightly better than 0.1 nm.
Since the information limit is mostly better than the point
resolution, a logical way (Kirkland (1984), Kirkland et al (1985), Van
Dyck (1990)) of increasing the resolution consists in restoring the
information that is present between the point to point resolution and the
information limit. For this purpose, electron holographic methods
(through focus holography and off-axis electron holography) have been
developed. It should be stressed that the improvement of the resolution
from 0.17 nm (point resolution) to 0.1 nm (information limit) is not just
a simple reduction, because this improvement crosses the boundary
beyond which atom columns separated by ‘normal’ interatomic distances
can be discriminated. Especially for non-periodic structures, the aberration
correction, which is included in the holographic reconstruction
algorithms, makes a direct imaging of atom columns within grain
boundaries and defects feasible.
The retrieval of the maximum amount of information
from HREM images
The information of the electron wave that has passed through the
specimen is scrambled in the image due to the aberrations of the
microscope. The unscrambling (deblurring) can be done in two steps. First,
one has to reconstruct the wavefunction in the image plane. Since by
recording the image only the amplitude information is collected, one has
to find a way to determine also the missing phase information. By
definition this can only be done using holographic methods. After
determining the image wave one has to reconstruct the exit wave of the
object. For the latter the microscope parameters have to be known very
precisely. This forces the microscopist to perform the experiments with
care and to decide beforehand which information he wants to achieve,
such that the experiments are optimised to obtain this information.
Two holographic methods exist to determine the phase in HREM
images: off-axis electron holography [Lichte (1986, 1992)] and focus
variation [Van Dyck (1996a)].
In the focus variation method, the focus is used as a controllable
parameter so as to yield a through focus series from which both amplitude
and phase information can be extracted [Schiske (1973), Saxton (1986)]
[Van Dyck (1990), Coene et al, (1992), Op de Beeck (1995)]. Images are
149
captured at very close focus values so as to collect all information in the
three-dimensional image space. Each image contains linear information
and non-linear information. By Fourier transforming the whole 3D-image
space, the linear information of all images is located on a parabola. By
selecting only the information on this paraboloid [Van Dyck (1990)] the
contribution to non-linear interference is strongly reduced. This
reconstruction also allows the retrieval of the phase and amplitude of the
wave function at the image plane. The aberrated electron wave can still be
refined by iteration comparing calculated images from the aberrated
electron wave with all HREM images of the focus series. The best
matching criterion for this purpose is the maximum likelihood criterion
(for details see Coene et al (1992)). In this way all information, both
linear and non-linear, that is present in the images is fully exploited, so
that the highest precision is achieved. In practice a maximum likelihood
fitting can suffer from a low convergence rate. Therefore the procedure
can be best started with the paraboloid method of Van Dyck after which
the maximum likelihood fitting can be performed straightforwardly. An
important limitation of the reconstruction method is that the non-linear
interaction is underestimated due to a smaller contrast in the experimental
images compared to that in the calculated one.
For off-axis electron holography, the electron optical set-up of an
electron microscope has to be changed only slightly. The object plane is
illuminated by a plane electron wave where the object covers only half of
the plane while the other half serves as a reference area. Behind the
object, one finds the exit wave and the unscattered reference wave. Both
waves are imaged by the objective lens. A Möllenstedt biprism
[Möllenstedt (1956)], which is located (behind the specimen) in the
selected area aperture holder, deflects both waves towards each other when
a positive voltage is applied to the biprism filament. Therefore, the
aberrated electron wave and reference wave interfere in the overlapping
region of the image plane, resulting in an interference pattern which is
highly magnified by projective lenses and recorded by means of a slowscan
CCD camera. The intensity distribution of this "off-axis electron
hologram" can be written as
where denotes the carrier frequency and µ the contrast of the
interference fringes. This intensity distribution consists of two parts: The
first part, represents the intensity of the corresponding
conventional electron micrograph, whereas the second part shows both
amplitude and phase of the aberrated electron wave encoded as
contrast modulation and bending of interference fringes, respectively. The
reconstruction of the aberrated electron wave is performed by means of a
computer programmed with the applicable laws of wave optics [Franke et
al. (1988), Rau et al. (1991)]. The first step is a Fourier transformation
yielding a complex spectrum. The central band represents the
150
diffractogram of a conventional electron micrograph, whereas each of the
two side-bands corresponds to the complex diffraction pattern found in
the back focal plane of the objective lens. In the second step, one of them
is numerically centred and isolated in order to exclude information from
the auto-correlation. Finally, an inverse Fourier transformation yields the
aberrated electron wave in amplitude and phase
151
The wavefunction at the exit face of the object can be calculated from
the wavefunction in the image plane by applying the inverse phase
transfer function of the microscope. This procedure is in principle
straightforward. However, the retrieval of information up to the
information limit requires that the transfer function must be known with
highest accuracy. Hence, this requires an accuracy of better than 1% for
and 1 nm for the absolute focus. The accuracy and hence the
knowledge of the aberration coefficients is most crucial for all highresolution
methods.
152
The exit wave can only be correctly interpreted if this is done at the
exact focus and with a proper correction of all the lens aberrations,
because otherwise phase and amplitude information are mixed. A robust
method for determining the exact focus is to exploit the entropy of the
exit wave as a function of focus [Tang et al (1996), Van Dyck et al
(1996c)]. An efficient determination for all microscope parameters is
possible with the optical fingerprint method reported by Thust et al
(1996). This method, however, requires a round illumination (not the
elliptical illumination used for acquisition of off-axis electron holograms)
and a thin amorphous part of the specimen which should not be present at
clean specimen. In off-axis electron holography, an algorithm for
determination and correction of the coherent wave aberration has been
developed which is based on a combination of Genetic Algorithm and
Downhill-Simplex-Algorithm [Lehmann (2000)]. The basic criterion is
the minimisation of the amplitude contrast, which is a well-established
correction guideline for weak phase objects.
Two examples of the strength of the reconstruction are given in
Figures 3 and 4. Figure 3 illustrate the use of through-focus exit wave
reconstructions for the imaging of slabs to improve the
interpretability of the data. to which cobalt sulphide or nickel
sulphide has been added on an atomic scale (CoMoS and NiMoS catalysts
respectively) is employed on a large scale as catalyst in the hydrotreating
of oil. Figure 4 shows the use of exit wave reconstructions in the analysis
of the structure of small precipitates in commercial Al-Mg-Si alloys, which
153
play a crucial role in increasing their mechanical strength. The
composition and structure of the phase, occurring as precipitates of
typically which are associated with a particularly strong
increase in mechanical strength, has been determined. Element analysis
using a small electron beam probe indicates the composition to be
[Zandbergen et al (1997a)]. Fig 4b shows a reconstructed exit wave of a
precipitate embedded in aluminium, whereas one of the HREM images of
the through focus series is shown in Figure 4a. A comparison of Figures 4a
and 4b shows that a major advantage of the exit wave is that the
information, which is delocalized in the HREM images resulting in a
blurred image (Fig. 4a), is restored to its origin (Fig. 4b).
From HREM images to the structure
Once the exit wave has been accurately determined a major final step
remains: the retrieval of the projected structure of the object from the
exit wave. This is only straightforward if the object is thin enough to act
as a phase object: in that case the phase is proportional to the
electrostatic potential of the structure, projected along the beam
direction. However, if the object is thicker as is mostly the case, the
problem is much more complicated. In principle one can retrieve the
projected structure of the object by an iterative refinement based on
fitting the calculated and the experimental exit wave. This is basically a
search procedure in a giant parameter space whereby it is possible to get
stuck in a local maximum [Thust and Urban (1992)]. It is possible,
however, to obtain an approximate structure model in a more direct way.
If the object is a crystal viewed along a zone axis, the incident beam is
parallel to the atom columns. It can be shown [Van Dyck (1996b)] that in
such a case, the electrons are trapped in the positive electrostatic
potential of the atom columns, which then act as channels.
If the distance between the columns is not too small, a one-to-one
correspondence between the wavefunction at the exit face and the column
structure of the crystal is maintained. Within the columns, the electrons
oscillate as a function of depth without however leaving the column. It is
important to note that channelling is not a property of a crystal, but
occurs even in an isolated column and is not much affected by the neighbouring
columns, provided the columns do not overlap. Hence the one-toone
relationship is still present in case of defects such as translation
interfaces or dislocations provided they are oriented with the atom
columns parallel to the incident beam. One can explicitly specify the
thickness dependency of the wavefunction at the exit face of a column as
[Van Dyck, 1996b, 1997]:
154
where z is the thickness, a constant expressing the electron matter
interaction and E the energy of the column. This result holds for each
isolated column. In a sense, the whole wavefunction is uniquely
determined by the eigenstate of the Hamiltonian of the projected
columns and its energy E which are both functions of the “density” of the
column and the crystal thickness. It is clear from the formula above that
the exit wave is peaked at the centre of the column and varies periodically
with depth. The periodicity is inversely related to the “density” of the
column. In this way the exit wave still retains a strong correspondence
with the projected structure, whereby the positions are still (almost) the
same but the contrast (in phase or amplitude) cannot be used to determine
the scattering potential in the projected structure unless the thickness is
accurately known. We used this to determine the structure of
[Zandbergen (1998a)]. From the exit wave we were able to determine the
positions of the columns. This model was the basis of a structure
refinement using electron diffraction data for several thicknesses, whereby
first the position of the columns was refined assuming all columns to be
equal Cu columns. Next the occupancies were refined, allowing one to
decide which atom columns contained, Ce, Cu or P. The final refinement
gave very accurate atomic positions.
155
HREM investigation of grain boundaries
In the HREM investigation of grain boundaries and surfaces three
aspects are very important restrictions: the shape and orientation of the
grain boundary, the crystal lattices of the adjacent crystals and the
resolution of the electron microscope. The shape and orientation of the
grain boundary is important, because HREM provides in first
approximation only projections of the structure, such that only lateral
information and no depth information is available. This implies that the
information of a grain boundary that is not parallel to the electron beam
156
will be smeared out. Concerning the determination of rough shapes, a lot
of development has been done in the last few years on the determination
of the shape of biological particles by means of electron tomography
[Waltz (1997), Bakker (1996)]. The present state of the art is that a 3-D
resolution of about 1 nm can be obtained. For this a tilt series from –70° to
+70° with increments of 2° are typically used. The 3D reconstruction of
these biological specimens is based on contrast changes due to the overlap
of scattering material in projection. In the study of inorganic solid state
materials a 1 nm resolution is insufficient and also more difficult to obtain.
The 1 nm resolution is insufficient because one will be interested in the
atomic arrangements at the grain boundary (e.g. even an amorphous grain
boundary layer of 0.5 nm can strongly influence the properties). The 3D
reconstruction of grain boundaries is more difficult when these materials
consist of (small) crystals. Crystallinity results in extra contrast as can be
seen in Figure 1, which depends very strongly on the orientation. In
orientations where the crystal is in a low-index orientation (a low index
orientation is an orientation [hk1] that has a low sum h+k+1.), it scatters
the electrons much stronger than in a high-index orientation. This cancels
the one-to-one correspondence between contrast and the projected average
scattering potential. This problem can be partly overcome by using a
conical illumination or by using a high angular dark field detector in
combination with scanning transmission electron microscopy [Thomas,
2001].
A crystalline specimen allows one to obtain a structure image. In this
case one has to orient the crystal such that the projection of the structure
along the electron beam results in image features that can be resolved given
the resolution of the electron microscope. This implies that the imaging
can only be done with the electron beam along a small selection of crystal
directions (see Figure 6), whereby the better the resolution of the electron
microscope the more directions can be used.
If the crystal is perfectly aligned, the atoms can be considered as
aligned in columns. In this case a column scattering potential [Van Dyck,
1996b, 1997] can replace the scattering potential of the atoms. The
column scattering potential depends on the weight of the atoms and their
distance along the column. If the crystal is tilted such that the column is
not exactly along the electron beam (mistilt less than 1°) this column
approach is still valid.
Summing up the requirements for obtaining HREM images of grain
boundaries that can be interpreted in a straightforward way: i) the grain
boundary is straight, ii) the grain boundary is parallel to the electron beam,
iii) one of the grains and preferable both are in such an orientation, that
lattice imaging is possible. Obviously most of these requirements become
the more stringent the thicker the specimen. On the other side, a too thin
specimen might not reflect the initial structure of the grain boundary due
to changes induced by the thinning of the specimen or lattice relaxation
due to the very limited thickness of the specimen. The interaction between
157
electrons and matter is very large (about times larger than for X-rays),
such that very small specimen thicknesses can be investigated. The
specimen thickness required for a good image contrast ranges from 1 to 10
nm depending on the scattering potential. This implies that the
requirements that the electron beam is parallel to the grain boundary and
parallel to atom columns are not very stringent.
158
Obviously, the better the resolution of the microscope, the more
directions one can get structural images from. In particular for the study of
grain boundaries this increase in suitable directions is important. Assuming
that a HREM allows a double tilt of ±30°, one out of three grain boundary
can be aligned along the electron beam. Assume next that by rotation of
the specimen about the interface normal with a range of 40°, one
orientation can be reached allowing structure imaging for the first or the
second adjacent grain. Then the chance of being able to have a grain
boundary being parallel to the electron beam with the first grain only (the
situation for a surface) is still one out of three. Assuming random
orientation of the grains and that the maximum misorientation is 2°, the
chance that an orientation can be found in which both grains are in a both
grains in an orientation allowing structure imaging is 1 over 30. A doubling
of the number of orientations allowing structure imaging increases this
chance to 2 over 15. Figure 2 illustrates the interpretation problems one
can encounter. Here an example of the study of grain boundaries in the
superconducting is shown.
Adding information of the compositional and electronic
structure
In fact one aims at a complete picture of the Atomic arrangement, the
Chemical composition and the Electronic structure on an atomic scale.
The atomic arrangement with a resolution of about 1.7 Å can be obtained
through HREM experiments as discussed above. The chemical composition
can be obtained through Energy Dispersive X-ray analysis (EDX-analysis)
combined with analysis of the energy loss of the electrons that are
transmitted through the specimen (Electron Energy Loss Spectroscopy
(EELS)). In both cases the minimum area that can be analysed depends on
the probe size, which can be as small as 2Å (see for instance Yan (1998)).
The electronic structure can in principle be determined from EELS data,
since the loss of energy of the inelastically scattered electrons will be
determined by the electron-specimen interactions, which are not only
atom dependent (coarse structure of EELS spectrum), but depend also on
the environment of the atom (fine structure of the EELS spectrum). EELS
is presently possible with a resolution of approximately 0.1eV. This will
create new and exciting possibilities in the investigation of the physics and
chemistry of materials.
References
Bakker T.S. and R.H. Cheng, J. Struct.Biol., 116 (1996) 120.
Bokel R.M.J., Jansen J.,Zandbergen H.W. and van Dyck D. (2001), Ultramicroscopy 87 89-96
Bokel R M J, J Jansen, D van Dyck, H W Zandbergen; Ultramicroscopy, 1999, p. 255-269.
Coene W., Janssen G., Op De Beeck M. and Van Dyck D. (1992a), Phys. Rev. Lett., 69, 3743
Franke F.J., Herrmann K.-H., Lichte H. (1988), Scanning Microscopy Supplement 2, 59 - 67
159
Fu Q., Lichte H., Völkl E. (1991), Physical Review Letters 67, 2319
Hÿtch M.J. and W.M. Stobbs (1994), Ultramicroscopy 53, 191.
Ishizuka K., Tanji T., Tonomura A., Ohno T., Murayama Y. (1994), Ultramicroscopy 53, 361
Jansen J., Tang D., Zandbergen H.W. and Schenk H. (1998), Acta Cryst A. Acta Cryst. A 54, 91
Kienzle O., F. Ernst and G. Möbus, Journal of Microscopy, 190 (1998) 144-158 144-158
King W.E., G.H. Cambell, S.M. Foiles, D. Cohen K. Hanson, J. Micr., 190 (1998) 131-143.
Kirkland E.J (1984), Ultramicroscopy 15, 151-172
Kirkland E.J., Siegel B.M., Uyeda N. and Fujiyoshi Y. (1985), Ultramicroscopy 17, 87-104
Kirkland E.J. (1982),. Ultramicroscopy 17, 87-104
Lehmann M., Lichte H. (1994) , 13th
Intern Congr on Electron Microsc, ICEM13, Paris, 293
Lehmann M. (2000), Ultramicroscopy 85, 165 - 182
Lichte H. (1986), Ultramicroscopy 20 293
Lichte H. (1992), Ultramicroscopy 47, 223
Malamidis D., Lichte H. (1998), Proceedings of the 14th
International Congress on Electron
Microscopy ICEM14, Cancun, Vol I, 561
Merkle K.L., Interface Science 2, (1995) Kluwer Academic Publishers, Boston, 311-345.
Meyer R.R. (2001), submitted to Ultramicroscopy
Möbus G., R. Schweinfest, T. Gemming, T. Wagner M. Rühle J. Micr., 190 (1998) 109-130.
Möllenstedt G., Düker H. (1956), Vol. 145, 377 - 397
Op de Beeck M., Van Dyck D., Coene W. (1995), (Ed. A. Tonomura et al.), North HollandElsevier,
ISBN 0-444-82051-5, p. 307-316.
Orchowski A., Rau W.D., Lichte H. (1995), Physical Review Letters 74, 399
Rau W.-D., Lichte H., Völkl E., Weierstall U. (1991), Journal of Computer-Assisted
Microscopy 3, 51
Saxton W.O. (1986), in: Proc. XIth International Congress on Electron Microscopy, Kyoto,
post deadline contributions page 1.
Schiske P. (1973), in: Image Processing of Computer-aided Design in Electron Optics (Ed. P.
Hawkes).
Siegel R.W., Mater. Sc. Eng, A168 (1993) 189
Stadelman P.A. (1987), Ultramicroscopy 21, 131-146
Steinecker A. and W. Mader (1998), Journal of Microscopy 190 281.
Tang D., Zandbergen H.W., Jansen J., Op de Beeck M. and van Dyck D. (1996),
Ultramicroscopy 64, 265-276
Thomas P.J., Midgley P.A. (2001), J. Phys. Chem. B, 105, 7882.
Thust A. and Urban K. (1992), Ultramicroscopy 45, 23-42
Thust A., Overwijk M.F.H., Coene W.M.J., and Lentzen M. (1996 ), Ultramicroscopy 64, 249.
Van Dyck D., M. Op de Beeck(1990), in: Proc. XIIth International Congress for Electron
Microscopy (Seattle), San Francisco Press Inc., p. 26-27
Van Dyck D. and De Jong A.F. (1992), Ultramicroscopy 47, 266
Van Dyck D., Lichte H. and K.D. van der Mast (1996a), Ultramicroscopy 64,1
Van Dyck D. and Op De Beeck M. (1996b), Ultramicroscopy 64, 99
Van Dyck D., op de Beeck M., Tang D., Jansen J. and Zandbergen H.W. (1996c), IEEE Intern.
Conf on image processing 1996 Los Alamitos 737-170
Van Dyck D., Handbook of Microscopy 1997.
J. Walz, et al, J. Struct.Biol., 120 (1997) 387-395.
Yan Y., M.F. Chisholm, G. Duscher, A. Matti, S.J. Pennycook and S.T. Pantelides, Phys. Rev.
Lett. 81 (1998) 3675-3678
Zandbergen H.W., Andersen S.L. and Jansen J. (1997), Science 227, 1221
Zandbergen H.W. and Jansen J. (1998), J. Microscopy 190, 222
Zandbergen H W, R M J Bokel, E Connolly and J Jansen 1999; Micron, 30, p. 395 - 416
NANOELECTRONICS
G. Allan, C. Delerue, C. Krzeminski, M. Lannoo*
Institut d’Electronique et de Microélectronique du Nord, Département Institut Supérieur
d’Electronique du Nord, 41 boulevard Vauban, 59046 Lille Cédex, France
* Laboratoire Matériaux et Microélectronique de Provence, ISEM, Place G. Pompidou, 8300
Toulon, France
1. INTRODUCTION
In this chapter we intend to discuss the major trends in the evolution of
microelectronics and its eventual transition to nanoelectronics. As it is well
known, there is a continuous exponential tendency of microelectronics
towards miniaturization summarized in G. Moore’s empirical law. There is
consensus that the corresponding decrease in size must end in 10 to 15 years
due to physical as well as economical limits. It is thus necessary to prepare
new solutions if one wants to pursue this trend further. One approach is to
start from the ultimate limit, i.e. the atomic level, and design new materials
and components which will replace the present day MOS (metal-oxide-semiconductor)
based technology. This is exactly the essence of nanotechnology,
i.e. the ability to work at the molecular level, atom by atom or molecule by
molecule, to create larger structures with fundamentally new molecular organization.
This should lead to novel materials with improved physical, chemical
and biological properties. These properties can be exploited in new
devices. Such a goal would have been thought out ofreach 15 years ago but
the advent ofnew tools and new fabrication methods have boosted the field.
We want to give here an overview of two different subfields of nanoelectronics.
The first part is centered on inorganic materials and describes two
aspects: i) the physical and economical limits ofthe tendency to miniaturization;
ii) some attempts which have already been made to realize devices with
nanometric size. The second part deals with molecular electronics, where the
basic quantities are now molecules, which might offer new and quite interesting
possibilities for the future ofnanoelectronics.
162
2. DEVICES BUILT FROM INORGANIC MATERIALS
These are mainly silicon based microelectronic devices which have
invaded our life. Integrated circuits are now found everywhere not only in
Personal Computers but also in a lot of equipment we use each minute as
cars, telephones, etc. We always need more memory as well as faster and
cheaper processors.
The race for miniaturization began just after Kahng and Atalla [1]
demonstrated in 1960 the first metal-oxide semiconductor field effect transistor
(MOSFET) (Fig. 1). It turned out to be a success because a large number
of transistors and their interconnections could be easily built on the surface of
a single silicon chip. Ten years later, the first 1 kilobyte memory chip was on
the market and this trend has been followed until now (a 64 megabit one contains
more than one hundred millions electronic components). In 1965, Gordon
Moore predicted what is known as the Moore’s law: for each new
generation of memory chip on the market, the number of components on a
chip would quadruple every three years. Miniaturization not only decreases
the average current cost per function (historically, ~25%/year) but also
improves the cost-performance ratio. In the same time, the market growth
was close to 15%/year.
Most of the improvement trends are exponential and are resumed in
the scaling theory [2]. It shows that a MOSFET operates at a higher speed
without any degradation of reliability ifthe device size is scaled by a factor 1/
k and at the same time the operating voltage is scaled by the same factor. The
speed of the circuits has increased up to one Gigahertz in today’s personal
computers. Continued improvements in lithography and processing have
made possible the industry’s ability to decrease the minimum feature sizes
used to fabricate integrated circuits. For four decades, the semiconductor
technology has distinguished itself by the rapid pace of improvement in its
products and many predicted technological limitations have been overcome.
In the same time it is difficult for any single company to support the progressively
increasing R&D investments necessary to evolve the technology. Many
forms of cooperation have been established, as the «International Technology
Roadmap for Semiconductors" [3] which is the result of a large world consensus
among leading semiconductor manufacturers. It looks at the 10-15
years in the future showing that most of the known technological capabilities
will be approaching or have reached their limits. Continued gains in making
VLSI (Very Large Scale Integration) seem to be first severely limited by technological
considerations [4].
The first one which is known as a «power crisis» is an heat dissipation
problem. While the size of the components is reduced by a factor k, the scaling
theory shows that the power density increases like A today’s single
chip processor in production requires 100 W and this value will rise to
150 W during the next decade. Low-power design of VLSIs is also necessary
because they are more and more used in mobile electronic systems which
need long-lasting batteries. The reduction of the supply voltage (~0.37 V in
2014) will be accompanied by an increase of the current needed to operate
VLSIs to huge values (500 A). This contributes to the second limitation: the
«interconnect crisis».
A large operating current gives rise to voltage drop problems due to
the resistance of the interconnections and to reliability degradation due to
electro-migration ofdefects in the conducting wires. On the other hand, to be
attractive, a scaling of the components must be accompanied by a scaling of
the interconnect line thickness, width and separation. Then signal integrity is
also becoming a major design issue. A high crosstalk noise is due to larger
capacitive couplings between interconnects. A smaller geometry also
increases the RC (resistance-capacitance) delay (it increases as If
this delay increases, the signal cannot propagate anywhere within the chip
within a clock cycle.
According to T. Sakurai [2], a «complexity crisis» will appear. Design
complexity is increasing superexponentially. The first obvious reason is the
increased density and number of transistors. The complexity is also growing
due to designs with a diversity of design styles, integrated passive components
and the need to increase incorporate embedded software. The integrated
circuit is built on several interconnected levels which interact. Verification
163
complexity grows with the need to test and validate the designs. Finally the
tests must be done at higher speed, higher levels of integration and greater
design heterogeneity.
There are potential solutions to solve some of these problems during
the 15 next years [3]. In some other cases the solution is still unknown and
this will have a price that people could not continue to afford. Moreover until
now, we have considered more technological limitations but new fundamental
quantum phenomena will also appear. The first one which has already been
investigated is the tunnel effect through the gate insulator which is made of
the silicon native oxide One important factor of the MOSFET is the
gate capacitance of the parallel plate capacitor made by the gate and the conducting
channel in the semiconductor (Fig. 1). This capacitor is filled by an
insulator which is at present silicon dioxide. The charge ofthis capacitor controls
the current between the drain and the source. When the gate (which is
one ofthe capacitor plates) is scaled by 1/k, the gate thickness scales as 1/k to
maintain the same capacitance. It will be reduced to 0.7 nm in 2014. at the
same time the gate voltage should be reduced to maintain the electric field
across the oxide below an undesirable value. This is equivalent to 3 layers of
oxygen and 2 of silicon in the oxide. When the insulator thickness is reduced,
the overlap of the electronic wavefunctions on both barrier sides increases
exponentially and this gives rise to a non-zero probability for the electrons to
tunnel across the barrier. An other formidable challenge is to replace the silicon
oxide by a dielectric material with a higher dielectric constant. A review
of current work and literature in the area of alternate gate dielectrics is given
in reference [5].
Doping the semiconductors with impurities, donors (n type) or acceptors
(p type), is an essential feature to get free carriers and thus get a substantial
conductivity. With current impurity concentration one can show that
below a 0.1 * 0.05 mm gate, there are about 100 free carriers only. This
means that a fluctuation ofplus or minus one charged impurity in the channel
gives a 1% error, which represents another type of limitation.
Some other new quantum effects appear due to the size reduction:
- as in a molecule or an atom, the electron energy can only take discrete
value as opposed to the classical value or to the existence of bands of
allowed energies for bulk materials.
- when the wave function associated with an electron takes several distinct
channels, interference effects occur which give rise to conductance fluctuations
with a root mean square deviation equal to , Such effects are only
observed when the phase is not destroyed by inelastic collisions, i.e. when the
distance covered is lower than the inelastic mean free path. This is close to
164
165
0.1 at room temperature for GaAlAs-GaAs heterojunctions or in Si.
- a one-dimensional quantum wire is analogous to a wave guide connected
to electron reservoirs. When a voltage is applied between these reservoirs,
electrons are injected in the wave guide. The number of onedimensional
channels for an electron depends on the number of energy levels
below the Fermi energy. This number N of discrete levels due to the confinement
perpendicular to the wire is fixed by the width of the quantum wire.
Within these conditions the wire conductance is quantized and equal to
- discrete energy levels in different wells can interact through a tunnel
effect. When a polarization is applied between the outer reservoirs, the conductance
is small except when the quantum wells energy levels are aligned.
Then the current is maximum and decreases for a further increase of the
applied voltage. This gives rise to a negative differential resistance (see Fig-
166
ure 2).
These effects which limit the feasibility to reduce the device size
below a certain value can also be used to invent new architecture. One electron
devices are certainly the most surprising and promising effect. The simplest
one is the tunneling junction shown on Figure 3 [6]. It is a metalinsulator-metal
junction between two electron reservoirs. When the barrier
width or height is large, the probability for an electron to cross the barrier
when a polarization is applied is small and the tunneling is quantized: the
electrons cross the barriers one by one. Then the system is equivalent to a
capacitor C and an electron leak due to tunneling across the insulating layer
(Figure 3b). When a voltage V is applied, the tunneling of an electron is possible
if the energy difference
for the capacitor with a charge equal to Q (=CV) and to Q-e becomes
167
positive. When no electron can cross the barrier
and we have a "Coulomb blockade". To maintain a constant current I, one
must applied a saw tooth potential with a frequency I/e. It is difficult to realize
such a system and the following electron box is much easier to make.
Let us take a metallic quantum box separated from two electron reservoirs
on one side by an ideal capacitor and on the other one by a tunnel
junction with a capacitance C (Figure 4). Electrons can tunnel into the quantum
box one by one until a charge –Ne, N depending on the applied voltage
V. The simplest way to calculate N(V) is to define the ionization level
For the box, is equal to the total energy differ-
168
ence which is roughly equal to Applying
the Koopmans theorem, this quantity is equal to the lowest one-electron
energy level which can accommodate an extra electron when the box is
already charged with N electrons. A classical electrostatic calculation gives
the potential of the charged box one must add to the HOMO energy level of
the isolated box to get the ionization potential:
As shown on Figure 4b, a stable -Ne box charge occurs when the metal
Fermi level is located between and A step in
N(V) occurs each time the metal Fermi level is aligned with an ionization
potential and we can easily control the box charge.
The next step is to realize a one-electron transistor [6] shown on Figure
5 with two tunneling junctions and an ideal capacitor. The gate potential
controls the charge of the dot and the current. More complicated devices
allow to control the electrons one by one like the electron pump [7] and the
single electron turnstile [8]. Experimental results and a good introduction can
be found in Ref [9-11].
To observe a Coulomb blockade, the energy difference between two
ionizations levels as calculated from Equation (2) must be larger
than kT. The first experiments were done at low temperature but one must
reduce the size to work at room temperature. Single electron electronics has
been proposed by Tucker in 1992 [12]. This would allow a further reduction
of size and electrical power, both conditions being necessary to increase the
speed of a device. Some components have already been realized [13-19] but
remain difficulties due notably to non-reproducibility.
3. MOLECULAR ELECTRONICS
If the reduction in size of electronic devices continues at its present
exponential pace, the size of entire devices will approach that of molecules
within few decades. However, major limitations will occur well before this
happens, as discussed previously. For example, whereas in current devices
electrons behave classically, at the scale of molecules, they behave as quantum
mechanical objects. Also, due to the increasing cost of microelectronic
factories, there is an important need for much less expensive manufacturing
process. Thus, an important area of research in nanotechnology and nanoscience
is molecular electronics, in which molecules with electronics functionality
are designed, synthesized and then assembled into circuits through
the processes of self-organization and self-alignment. This could lead to new
electronics with a very high density of integration and with a lower cost than
present technologies. In this section, we review recent progress in molecular
electronics and we describe the main concepts at the origin of its development.
We show that the latter is strongly connected to the invention of new
tools to identify, to characterize, to manipulate and to design materials at the
molecular scale. We also stress the importance ofthe basic knowledge which
has still to be obtained in the way towards the integration of molecules into
working architectures.
3.1 Concepts and origins of molecular electronics
Quite surprisingly, the first ideas of using specific molecules as electronic
devices and of assembling molecules into circuits were proposed more
than 30 years ago. At this time, the electronic processors were only in their
infancy, and adequate tools to perform experiments on single molecules were
not available. However, the concept ofmolecular electronics is appealing, and
two components were proposed, the molecular diode and the molecular wire,
169
which can be seen as elementary bricks to build more complex devices or circuits.
In spite of these early proposals, practical realizations came only
recently due to limitations in chemistry, physics and technology. In the following,
we briefly describe the basic principles ofthe two components.
3.1.1 Molecular diode
In 1974, Aviram and Ratner [20] proposed to make an electrical rectifier
based on molecules between two metallic electrodes, where D and
A are, respectively, an electron donor and an electron acceptor, and is a
covalent bridge. Figure 6 describes the physical mechanism for the rectification.
The electronic states are supposed to be totally localized either on the D
side or on the A side. The HOMO(D) and LUMO(D) are high in energy, compared
to, respectively, the HOMO(A) and LUMO(A). Therefore, a current
can be established at relatively small positive bias, such that the Fermi level at
the A side is higher than LUMO(A), and the Fermi level at the D side is lower
than HOMO(D), provided that the electrons can tunnel inelastically through
the bridge. Thus an asymmetric current-voltage I(V) curve is expected like
in a conventional electronic diode. In the prototype proposed by Aviram and
Ratner, the donor group is made by a tetrathiofulvalene molecule (TTF), the
acceptor group by a tetracyanoquinodimethane molecule (TCNQ), and the
bridge by three metylene bonds. These molecules are seen as the analogs of n
and p-type semiconductors separated by a space-charge region. Twenty years
have been necessary before the first experimental demonstration of rectifying
effects in diodes based on molecular layers [21], and studies have been amplified
in this direction recently [22-24]. Nevertheless, the origin ofthe rectification
is still matter of debate [25]. In addition, the Aviram-Ratner principle at
the level of a single molecule has not been demonstrated yet.
170
171
3.1.2 Molecular wires
The discovery ofthe first conductive polymers based on acetylene [26-
27] suggested that molecules could have interesting properties to make electrical
wires at the molecular scale. These wires are necessary to connect
molecular devices into circuits. In 1982, Carter [28] suggested to address the
acceptor and donor groups of an Aviram-Ratner diode with polyacetylene
chains, and even to control several diodes in the same time. Using this technique,
the expected density of components on a chip could be of the order of
which is far beyond the possibilities of present technologies. One
difficulty raised by Carter is that the transport in the molecular chains takes
place in the form of solitons whose motion is slow. But this disadvantage
would be largely compensated by the higher density.
3.1.3 Molecular circuits
Conventional microelectronic technologies presently follow a topdown
approach, processing bulk semiconductor materials to make devices
with smaller and smaller sizes. In molecular electronics, the basic components
like molecular diodes and wires are used to build more complex devices
and circuits. This is the bottom-up approach which starts from the molecular
level to build a complete chip. For example, it has been recently proposed to
create AND and XOR logical functions using assemblies of molecular films,
molecular diodes and nanotubes (wires) [29]. These functions could be used
to design additioners and other operations which are currently done in CMOS
technologies. However, the assembling ofthe elementary units will not necessarily
lead to the expected device as the interaction between individual molecules
may perturb their own function [30]. Thus, other proposals suggest to
consider the system as an ensemble and to use directly the chemistry to synthesize
molecules with the required functions (e.g. additioners) [31].
3.2 Transport experiments on ensembles of molecules
A great challenge of molecular electronics is to be able to transfer
information at the molecular scale in a controlled manner. In the devices
described above, it consists of a charged carrier (electron or hole) which is
transmitted through a single molecule. During several decades, the impossibility
to work at the level of a single molecule has been a major difficulty.
However, the characterization of molecular ensembles has been undertaken
by different means which we describe now.
3.2.1 Molecules in a solvent
The first approach consists to study the molecules in a solvent, and to
probe the electronic transfer using a combination of chemical and physical
methods. It is possible to characterize the electronic transport in solution
through the measurement of the oxido-reduction by voltametry and of the
electronic excitation by optical absorption. If the chemical reaction takes
place inside a molecule, then reaction and intra-molecular transfer become
equivalent.
Experiments have been made on donor-ligand-acceptor molecules, like
mixed valence compounds recently synthesized [32]. These organo-metallic
complexes contain two metallic atoms with different oxidation degrees. The
experiments allow the measurement of the electronic transfer between the
two sites. For example, Taube has synthesized a stable molecule containing
two ruthenium complexes which act as donor and acceptor groups [33].
When the molecule is partially oxidized, a charge transfer is observed
between the two ions. Other molecules have been made with a large distance
of 24 Å between the two ions, which is obtained by the intercalation of five
phenyl groups [34]. An electronic transfer is also observed in this system in
spite of the long inter-site distance. This effect is attributed to the electronic
coupling induced by the phenyl groups which play here the role of a molecular
wire between the two ruthenium ions [35].
Interesting results have been also obtained on biological molecules. In
the case of proteins [36], an increase of 20 Å in the distance between the
acceptor and donor sites leads to a decrease by a factor of the transfer
rate. Thus, in this process, proteins can be seen as a uniform barrier which
limits the electronic tunneling. The importance oftunneling effects in biological
molecules is presently a well studied topic.
3.2.2 Molecular layers
The second approach is the study of molecules which are self-organized
in two-dimensional monolayers on the surface of a conductive substrate.
Self-assembled monolayers (SAMs) are obtained by the LangmuirBlodgett
(LB) technique or by chemical grafting on a surface. The SAMs
have to be well organized to avoid artifacts due to disorder. Electrical measurements
require a second electrode which is made by evaporation ofa metal
on top of the SAM. This is usually the most difficult task because the metal
must not diffuse into the SAM where it could make short circuits [37]. In the
172
173
same time, the contact resistance must be small which requires a good control
of the metal/SAM interface.
The first results based on this technique have been published by Mann
and Kuhn on LB films ofalkane chains [38]. Studies ofthe tunneling through
the molecular films show that the molecules are good insulators. Similar
results have been obtained on SAMs of n-alkyltrichlorosilane chemically
grafted on a Si substrate [39]. These chains realize a barrier ofthe order of4.5
eV for the tunneling of electrons or holes [40] and lead to a better insulating
layer than at the nanometer scale.
Other works on SAMs concern molecular diodes in the sense of Aviram-Ratner:
they are described in previous sections and in ref. [21-25]. Also,
nice results have been obtained by Fischer et al [41] on organic heterostructures
made with palladium phtalocyanines and compounds based on perylene.
Using gold electrodes at 4.2 K, they obtain I(V) curves with clear steps (Figure
7) which are attributed to the resonant tunneling through the molecular
levels. The I(V) characteristics is symmetric when the heterostructure is symmetric,
but it becomes rectifying when perylene layers are inserted on one
side of the structure.
3.3 STM measurements on single molecules
The invention ofthe Scanning Tunneling Microscope (STM) [42,43] is
a main step for the development of molecular electronics. The STM is based
on a sharp tip (curvature radius in the nanometer range) which is placed at a
small distance from a conductive surface in a way that electrons can tunnel
from one electrode to the other. Using piezoelectric tubes, the position of the
STM tip is fixed with an accuracy better than 1 Å, horizontally and vertically.
The STM is an efficient tool to study single molecules adsorbed on metallic
or semiconductor surfaces [44,45]. It can be used to image, to displace and to
characterize single molecules.
3.3.1 STM imaging of single molecules
A natural application ofthe STM is the imaging ofmolecules adsorbed
on a surface. In this mode, a constant bias is applied between the tip and the
substrate. When the tip is displaced laterally, one measures the height of the
tip which is adjusted in order to keep a constant tunneling current. The map
of the height versus the lateral position is roughly representative of the topography
of the surface, and thus gives information on the adsorbates. The
adsorption of small molecules like CO and benzene on Rh(111) has been
studied in detail [46,47], showing in some cases that self-organization can
take place at the surface. Larger molecules like naphtalene or phtalocyanines
have been also imaged [48,49]. The STM allows to study the adsorption sites.
Nevertheless, the interpretation of the image is not straightforward, as it does
not give information on the atomic positions but on the electronic structure.
Most of the work on STM imaging of adsorbed molecules concerns
metallic surfaces, and only few semiconductor surfaces. However, as microelectronics
technology is based on silicon substrates, there is an increasing
need to study organic molecules linked to a silicon surface [50,51]. For example,
Figure 8 shows high resolution STM images of a Si(l00) surface after
deposition of thienylenevinylene tetramers, which belong to a new class of
oligomers of particular interest as molecular wires [52]. The silicon
dimers, typical of a Si(l00) (2x1) surface, are visible, forming rows of
grey bean shaped. On the top of these rows, bright features can be seen, with
different shapes corresponding to different adsorption configurations [51].
These results show that the molecules are quite conductive. Detailed studies
allow a better understanding of the nature ofthe bonds between the molecules
and the surface. Spectroscopic measurements are also possible, as detailed in
the next sections.
174
175
3.3.2 STM as a tool to manipulate and to fabricate molecular objects
The STM tip is also a very interesting tool to manipulate the matter at
the atomic scale. The forces (Van der Waals, electrostatic, chemical) between
the atoms at the tip apex and the imaged object are used to displace atoms or
molecules on a surface [53]. The first controlled atomic manipulation was
presented by Eigler et al [54]. Xenon atoms were displaced on a nickel surface,
by moving the tip which was kept close to the atoms. Artificial structures
containing a small number of atoms were made using this technique.
Xenon atoms were also transferred vertically from the surface to the tip in a
reversible process [55,56].
The manipulation of molecules only came recently. Stipe et al have
shown that it is possible to induce the rotation of acetylene molecules by
excitation of C-H vibration modes with the STM [57]. This work shows the
importance of inelastic processes in the tunneling. This effect can be
exploited to perform vibrational spectroscopy on individual molecules [58]
and is therefore a powerful technique to identify adsorbates, their chemical
bonding and their local chemical environment. Due to the high spatial resolution
of the STM, correlations between the electronic structure and vibrational
excitation of adsorbed molecules can be determined. Moreover by using inelastic
tunneling, molecules can be dissociated, desorbed or even synthesized,
which is of essential importance in studies of molecular manipulation. Hla et
al have shown that the assembling of two molecules is possible using a STM
[59]. Starting from two molecules, they first removed the iodine atoms.
Then they approached the two phenyl groups and they observed the formation
of a biphenyl molecule. This reaction, entirely realized using the STM, was
made at 20 K whereas the conventional synthesis is impossible below 180 K.
Recently, larger molecules have been manipulated, showing that their adsorption
on a surface may induce important atomic reconstructions [60]. The
manipulation of specific parts of molecules may also lead to important
changes in their I(V) curve [61,62].
3.3.3 STM spectroscopy of molecules
In some conditions, the STM allows to study the electronic structure of
molecules in connection with their interactions with the surface. In the spectroscopic
mode of the STM, the tip is placed above an adsorbed molecule,
and the current is measured as a function of the applied voltage. Thus, using
this approach, a direct measurement of the transport properties of a single
molecule is possible.
Several studies have been applied to molecules because they are
quite stable and easy to manipulate due to their size. Joachim et al [63] have
measured the current through molecules on a gold substrate, at room temperature
and with a small applied bias (50 mV). As the tip-molecule distance is
reduced, the current increases at a very high rate which is interpreted by a distortion
of the electronic levels due to the pressure induced by the tip. Other
measurements at 4.2 K have been recently presented by Porath et al [64,65].
molecules are adsorbed on gold substrate covered by a thin amorphous
carbon film. The I(V) curves present clear steps which are interpreted by Coulomb
blockade effects and resonant tunneling through the discrete states of
the molecule. States close to the HOMO-LUMO gap are completely resolved
in spectroscopy. The degeneracy of the HOMO and LUMO levels is broken,
probably due to the tip-induced electric field or due to a Jahn-Teller effect.
Another way to probe the conductivity of a single molecule is to use
slightly defective SAMs. A particular system has been mainly studied, with
molecules terminated by a thiol end group. Self-assembly is routinely
obtained on gold surfaces (and others) using chemical grafting based on sulfur-gold
bonds [66]. In a well-known experiment [67], dodecanethiol SAMs
have been made with a small number of defects consisting of conjugated
wires which are slightly longer than the dodecanethiol molecules. The dodecanethiol
SAM forms a quite insulating layer [38]. STM images of the surface
show a smooth surface with only few bright spots at the position of the
molecular wires. This result demonstrates that the molecular wires have a
176
higher conductivity than the surrounding molecules. Other works have been
performed in this direction [68].
The spectroscopy of single molecules may be also evidenced by imaging
the surface at different tip-sample bias. In the example presented in a preceding
section where thienylenevinylene tetramers are adsorbed on Si(l00),
the images ofthe molecules are highly voltage dependent (Figure 8). At sufficiently
high negative sample voltages, the molecules are visible, whereas at
lower voltages, most of the molecules disappear. The interpretation is the following:
at high voltages, electrons can tunnel through the HOMO ofthe molecule,
whereas at lower voltages, only states close to the Fermi level of the
semiconductor and associated with the Si dimers can contribute to the tunneling
current.
Thus, interesting developments with a STM are taking place in various
areas. It remains that the experiments which have been done up to now are at
the state-of-the-art level, and their transfer to create new technologies is not
straightforward. Ifthe STM is a very good tool to study the transport through
a single molecule, it is clear that other approaches have to be developed to
make molecular electronic devices.
3.4 Molecules connected to nanoelectrodes
Here we describe other approaches to determine the I(V) characteristics
of a single molecule using nanoelectrodes. Compared to STM, these
methods lead to stable, permanent and symmetric junctions. Applications of
these techniques are already foreseen in the field of chemical and biological
sensors.
3.4.1 Co-planar electrodes
This approach is based on lithographic methods to make metallic electrodes
separated by a small gap, of the order of 5-10 nm for the smallest ones.
This system is used to characterize long molecules which are placed in such
manner that they are connected to the two electrodes (Figure 9). The conductivity
of carbon nanotubes has been studied using this approach [69,70]. It has
been verified that the nanotubes are metallic or semiconducting depending on
their geometry. Carbon nanotubes are presently considered as the best candidates
to make molecular wires, as the current can flow efficiently in long nanotubes
(~1-0.1 ). Co-planar electrodes are also used to measure the
conductivity of DNA [71], and even to explore superconductivity effects in
177
178
these systems [72,73].
3.4.2 Nanopores
In this approach, small holes are made in a thin silicon nitride film
deposited on a metallic electrode. The pores are filled with the molecules. A
second metallic electrode is evaporated on the top of the system to realize an
electrical contact. Compared to techniques involving molecular films, a
smaller number of molecules are probed in this approach (~1000). The number
of junctions which are short circuited is also reduced [74]. Measurements
have been realized on 1-4 phenylene diisocyanide [75] and [1: 2’-amino-4,4’di(ethynylphenyl)-5’-nitro-l-benzenedithiol]
[76,77]. In the case of the first
molecule, a symmetric I(V) curve is obtained. In the case of the second one, it
exhibits negative differential resistance with a high peak-to-valley ratio at low
temperature (Figure 10). The origin of this effect could be related to the electronic
structure of the molecule [78]. In this case, the molecule could be used
to realize oscillators.
3.4.3 Breaking junctions
In this third approach, a metallic wire (usually gold) is made by lithography
on a flexible substrate. Then the wire is broken using a small mechanical
perturbation. At the place where the wire is broken, there is a small gap of
nanometer size where molecules can be deposited from a solution. Using
piezoelectric tubes, the distance between the two electrodes is adjusted in
179
such a way that some molecules make a bridge between the electrodes. One
difficulty with this technique is that the number of molecules which are connected
is unknown. Reed et al have studied the conductance of benzene-1,4dithiol
grafted on gold electrodes [79]. Steps in the I(V) curve are obtained,
maybe due to Coulomb blockade effects. Other molecules have been studied
[80,81], confirming for example that dodecanethiol is good insulator at small
bias. Park et al have inserted a molecule in a breaking junction [82].
Using a gate voltage applied to the substrate, this system works as a transistor
at the molecular level. In addition, a new effect is observed as the electronic
transport is coupled to the motion of the molecule. The results are
explained by the oscillation of the molecule in the electrostatic field between
two electrodes. Coupling to internal modes of vibration of the molecule is
also suggested.
3.5 Molecular circuits
The fabrication of circuits based on molecules is obviously one main
objective of molecular electronics. Even if much work is needed to develop
the corresponding technologies, recent progress has been made in this direction.
Collier et al [83] have presented a molecular circuit based on the Teramac
architecture [84]. This computer architecture is defect-tolerant, meaning
that it continues to work up to three percent of defected components. This
kind of approach is ideal for molecular electronics since it will necessarily
contain defective components and connections. Logic gates are fabricated
from an array of configurable molecular switches, each consisting of a monolayer
of electrochemically active rotaxane molecules sandwiched between
metal electrodes. AND and OR functions have been realized. The circuit is
quite equivalent to Programmable Read Only Memory (PROM). Its fabrication
would be much cheaper than in CMOS technologies.
This work illustrates an important orientation of molecular electronics.
It could be used to make logic devices at a relatively low cost. It is foreseen
that expensive fabrication procedures in the electronic industry could be progressively
replaced by techniques derived from molecular electronics to
reduce fabrication costs. Thus, the main aim is no longer to realize nanometer
scale devices to follow the first Moore’s law, but to reduce costs which are
presently continuously increasing (second Moore’s law).
180
D. Kahng and M. Atalla, US patents 3206670 and 3102230 (1960).
T. Sakurai, Jap. Soc. Appl. Phys. 3, 15 (2001).
The International Technology Roadmap for Semiuconductors, Semiconductor Industry
Association (Sematech Inc., Austin, Texas 2000); see also http://www.sematech.org
for the most recent updates.
M. Schulz, Nature 399, 729 (1999).
G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89, 5243 (2001).
H. Grabert and M.H. Devoret "Single charge tunneling : Coulomb blockade phenomena
in nanostructures" (Plenum Press, New York, 1992)
L. J. Geerligs, V. F. Anderegg, P. A. M. Holweg, J. E. Mooij, H. Pothier, D. esteve, C.
Urbina, and M. H. Devoret, Phys. Rev. Lett. 64, 2691 (1990).
H. Pothier, P. Lafarge, P. F. Orfila, C. Urbina, D. Esteve, and M. H. Devoret, Physica B
169, 573 (1991); H. Pothier, P. Lafarge, C. Urbina, D. Esteve, and M. H. Devoret,
Europhysics Lett. 17, 249 (1992).
Special Issue on Single Charge Tunneling, Z. Phys. B 85,317 (1991).
D. V. Averin and K. K. Likharev in Mesoscopic Phenomena in Solids, B. L. Al’tshuler,
P. A. Lee and R. A. Wenn eds (Elsevier, Amsterdam, 1991).
D. K. Ferry and S. M. Goodnick in Transport in nanostructures (Cambridge University
Press, 1997).
J. R. Tucker, J. Appl. Phys. 72, 4399 (1992).
K. Yano, I. Tomoyuki, T. Hashimoto, T. Kobayashi, F. Murai, and S. Moichi, IEDM
Tech. Dig. 1993, p. 541.; K. Yano, T. Ishii, T. Hashimoto, T. Kobayashi, F. Murai, and
K. Seki, IEEE Trans. Electron Devices 41, 1628 (1994).41
L. Guo, E. Leobandung, and S. Chou, Science 275, 649 (1997); Appl. Phys. Lett. 70,
850 (1997).
A. Nakajima, T. Futatsugi, K. Kosemura, T. Fukano, and N. Yokoyama, Appl. Phys.
Lett. 70, 1742 (1997); Appl. Phys. Lett. 71, 353 (1997).
Y. Takahashi, M. Nagase, H. Namatsu, K. Kurihara, K. Iwadate, Y. Nakajima, S.
Horiguchi, K. Murase, and M. Tabe, IEDM Tech. Dig. 1994, p. 938; Electron. Lett. 31,
136 (1995); Y. Yakahashi, H. Namatsu, K. Iwadate, M. Nagase, and K. Murase, IEEE
Trans. Electron. Devices 43, 1213 (1996).
H. Ishikuro, T. Fujii, T. Saraya, G. Hashiguchi, T. Hiramoto, and T. Ikoma, Appl. Phys.
Lett. 68, 3585 (1996).
H. Ishikuro and T. Hiramoto, Appl. Phys. Lett. 71, 3691 (1997); Appl. Phys. Lett. 74,
1126 (1999).
L. Zhuang, L. Guo, and S. Y. Chou, Appl. Phys. Lett. 72, 1205 (1998); E. leobandung,
L. Guo, and S.Y. Chou, Appl. Phys. Lett. 67, 2338 (1995).
A. Aviram and M. A. Ratner, Chem. Phys. Letters 29, 277 (1974).
G.J. Ashwell, J.R. Sambles, A.S. Martin, W.G. Parker, and M. Szablewski, J. Chem.
Soc., Chem. Commun. 1374 (1990); A.S. Martin, J.R. Sambles, G.J. Ashwell, Phys.
Rev. Lett. 70, 218 (1993).
R.M. Metzger, B. Chen, U. Höpfner, M.V. Lakshmikantham, D. Vuillaume, T. Kawai,
X. Wu, H. Tachibana, T.V. Hughes, H. Sakurai, J.W. Baldwin, C. Hosh, M.P. Cava, L.
Brehmer, and C.J. Ashwell, J. Am. Chem. Soc. 119, 10455 (1997).
D. Vuillaume, B. Chen and R.M. Metzger, Langmuir 15, 4011 (1999). B. Chen and
REFERENCES
181
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
R.M. Metzger, J. Phys. Chem. B 103, 4447 (1999).
R.M. Metzger, J. Mater. Chem. 9, 2027 (1999); ibid 10, 55 (2000).
C. Krzeminski, C. Delerue, G. Allan, D. Vuillaume and R. Metzger, Phys. Rev. B, to be
published.
A. G. MacDiarmid and A. J. Heeger, Synthetic Metals 1, 101. (1980).
W. P. Su, J. R. Schrieffer, and A. J. Heeger, J. Chem. Phys, 73, 946 (1980).
F. L. Carter, Molecular electronic devices, Dekker, 51 (1982).
J. C. Ellenbogen and J. C. Love, Proceeedings ofthe IEEE 70, 386 (2000).
P. Ball, Nature 406, 118 (2000).
C. Joachim, J. K. Gimzewski and A.Aviram, Nature 408, 541 (2000).
H. Taube, Science 226, 1036 (1984).
C. Creutz and H. Taube, J. Am. Chem. Soc. 91, 3988 (1969).
C. Patoux, J-P. Launay, M. Beley, S. Chodorowski-Kimmes, J-P. Collin, S. James and
J-P Sauvage, J. Am. Chem. Soc. 120, 3717 (1998).
S. Larsson, Chemical Physics Letters 90, 136 (1982).
C. C. Moser, J. M. Keske, K.Warncke, R. S.Farid and P. L. Dutton, Nature, 355, 796
(1992).
R. M. Metzger and C. A. Panetta, New. J. Chem., 15, 209 (1991).
B. Mann and H. Kuhn, J. Appl. Phys. 42, 4398 (1971).
C. Boulas, J. V. Davidovits, F. Rondelez, and D. Vuillaume, Phys. Rev. Lett. 76, 4797
(1996).
D. Vuillaume, C. Boulas, J. Collet, G. Allan and C. Delerue, Phys. Rev. B 58, 16491
(1998).
C. M. Fischer, M. Burghard, S. Roth and K. V. Klitzing, Europhys. Lett. 28, 129
(1994).
G. Binning, H. Rohrer, Ch. Gerber and E. Weibel, Phys. Rev. Lett. 49, 57 (1982).
G. Binning and H. Rohrer, Rev. Mod. Phys. 59, 57 (1987).
J. Gimzewski, Physics World, 29 (1998).
J. Gimzewski and C. Joachim, Science 283, 1683 (1999).
H.Ohtani, R. J. Wilson, S. Chiang and C. M. Mate, Phys. Rev. Lett 60, 2398 (1988).
P. S. Weiss and D. M. Eigler, Phys. Rev. Lett 71, 3139 (1993).
V. M. Hallmark, S. Chiang, J. K. Brown and Ch. Woll, Phys. Rev. Lett. 66, 48 (1991).
J. Gimzewski and R. Moller, Phys. Rev. B, 36, 1284 (1987).
J.S. Hovis, H. Liu, R.J. Hamers, Surf. Sci. 402, 1 (1998).
B. Grandidier, J.P. Nys, D. Stiévenard, C. Krzeminski, C. Delerue, P. Frère, P. Blanchard,
J. Roncali, Surf. Sci. 473, 1 (2001).
C. Krzeminski, C. Delerue, G. Allan, V. Haghet, D. Stiévenard, E. Levillain, and J.
Roncali, J. Chem. Phys. 111, 6643 (1999).
S. Gauthier, Applied Surface Science 164, 84 (2000).
D. M. Eigler and E. K. Schweizer, Nature 344, 524 (1990).
D. M. Eigler, C. P. Lutz and W. E Rudge, Nature 352, 600 (1991).
M. F. Crommie, C. P. Lutz and D. M. Eigler, Science 262, 219 (1993).
B. C. Stipe, M. A. Rezaei, and W. Ho, Phys. Rev. Lett 81, 1263 (1998).
B. C. Stipe, M. A. Rezaei, and W. Ho, Science 280, 1732 (1998).
S-W. Hla, L. Bartels, G. Meyer and K-H. Rieder, Phys. Rev. Lett 85, 277 (2000).
M. Schunack, L. Petersen, A. Kuhnle, E. Laegsgaard, I. Stensgaard, I. Johannsen and
F. Besenbacher, Phys. Rev. Lett 86, 456 (2001).
182
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
F. Moresco, G. Meyer, K-H. Rieder, H. Tang, A. Gourdon, and C. Joachim, Appl.
Phys. Lett. 78, 307 (2001).
F. Moresco, G. Meyer, K-H. Rieder, H. Tang, A. Gourdon, and C. Joachim, Phys. Rev.
Lett 86, 672 (2001).
C. Joachim, J. K. Gimzewski, R. R. Schlittler, C. Chavy, Phys. Rev. Lett. 74, 2102
(1995).
D. Porath and O. Millo, J. Apply. Phys. Lett. 81, 2241 (1997).
D. Porath, Y. Levi, M. Tarabiah, and O.Millo, Phys. Rev. B 112, 558 (1990).
R. G. Nuzzo, L. H. Dubois and D. L. Allara, J. Am. Chem. Soc. 56, 9829 (1997).
L. A. Bumm, J. J. Arnold, M. T. Cygan, T. D. Dunbar, T. P Burgin, L. Jones, D. L.
Allara, J. M. Tour, P. S. Weiss, Science 271, 1705 (1996).
L. A. Bumm, J. J. Arnold, T. D. Dunbar, D. L. Allara, and P. S. Weiss, J. Phys. Chem.
B 103, 8122 (1999).
S. J. Tans, M. H. Devoret, H. Dal, A. Thess, R. E. Smalley, L. J. Geerligs and C. Dekker,
Nature 386, 475 (1997).
C. Dekker, Physics Today, 22 (May 1999).
D. Porath, A. Bezryadin, S. de Vries and C. Dekker, Nature 403, 635 (2000).
M. Kociak, A. Yu. Kasumov, S. Guéron, B. Reulet, I. I. Khodos, Yu. B. Gorbatov, V. T.
Volkov, L. Vaccarini, and H. Bouchiat, Phys. Rev. Lett. 86, 2416 (2001).
A. Yu. Kasumov, M. Kociak, S. Guéron, B. Reulet, V. T. Volkov, D.V. Klinov and H.
Bouchiat, Science 291, 280 (2001).
C. Zhou, M. R. Deshpande, M. A. Reed, L. Jones and J. M. Tour, Appl. Phys. Lett. 71,
611 (1997).
J. Chen, L. C. Calvet, M. A. Reed, D. W. Carr, D. S. Grubisha, D. W. Bennett, Chem.
Phys. Lett. 313, 741 (1999).
J. Chen, M. A. Reed, A. M. Rawlett, J. M. Tour, Science 286,1550 (1999).
J. Chen, W. Wang, M. A. Reed, A. M. Rawlett, D.W. Price, and J. M. Tour, Appl. Phys.
Lett. 77, 1224 (2000).
J. M. Seminario, A. G. Zacarias and J. M. Tour, J. Am. Chem. Soc. 122, 3015 (2000).
M. A. Reed, C. Zhou, C. J. Muller, T. P. Burgin, and J. M. Tour, Science 278, 252
(1997).
C. Kergueris, thesis from Orsay university (1998).
C. Kergueris, J.-P. Bourgoin, S. Palacin, D. Esteve, C. Urbina, M. Magoga and C.
Joachim, Phys. Rev. B, 59, 12505 (1999).
H. Park, J. Park, A. K. L. Lim, E. H. Anderson, A. P. Alivisatos, and P. L. McEuen,
Nature, 407, 57 (2000).
C. P. Collier, E. W. Wong, M. Belohradsky, F. M. Raymo, J. F. Stoddart, P. J.Kuekes, R.
S. Williams, J. R. Heath, Science 285, 391 (1999).
J. R. Heath, P. J. Kuekes, G. S. Snider, R. S. Williams, Science 280, 1716 (1998).
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
183
INDEX
A
Aberration, 162-164
Ag, 3, 4, 6, 7, 13
3,6,7
AgI, 7
Al, 97, 98, 104, 105,108, 110, 112
Al-Mg-Si alloys, 169
104, 105, 110, 112
63, 64, 133
Alloys, 9
A1OOH, 37
Atomic absorption analysis, 100
Atomic Force Microscopy (AFM), 101-104
Attrition, 56
Au, 3, 13
Autogeneous pressure, 23
B
38
Ball milling, 2
38
Bipolar coagulation, 49
Breaking junctions, 178
Broadening, 130
C
175, 179
177
Carrier confinement, 124
Catalyst, 132
Cathode material, 73
5
5
5
CdS, 3, 6, 7, 9
CdSe, 115-124
CdSe, 3
CdTe, 5
155
37, 135
Chip, D2
Co, 3, 4, 6, 8, 140
37
Co-condensation, 97, 99
Co-deposition, 98
Colloidal solutions, 1, 2
Colloids, 25, 35
Complexity, 163
Composite films, 117-124
Composites, 68
CoO, 3
Coprecipitation, 2
37
CrOOH, 37, 38
Crystal growth, 29, 39
Crystallite size, 129, 130
6, 8, 9, 15, 16
Cu, 3, 6, 8-11, 13, 14, 16
Cu, 97, 108
14, 15
Cubooctahedron, 128, 140
D
Debye-Waller factor, 137
Deintercalation, 74
Delocalisation, 147
Diamond, 69
Differential scanning calorimetry, 83, 85, 87-
89
Diffusion coefficient, 92
Distribution, 135
Dynamic cleaning, 56
E
Electrical conductivity, 83, 85, 87-92
Electrical resistance, 108, 110
Electro-Hydro-Dynamic Atomization, 43-53
Electron Energy Loss Spectroscopy (EELS),
159
Electron holography, 148, 149
Electrospray, 43-53
Energy Dispersive X-ray analysis, 159
Ethanol, 51
Ethylene glycol, 47, 51
Evaporation, 117
EXAFS, 141
Experimental set-up for hydrothermal
synthesis, 30, 31
Explosive compaction, 55, 58-60, 62, 66
186
F
27-28, 37
28
Ferrite, 2-4
Ferrofluids, 9
Field effect transistor (FET), 162
G
Gas sensors, 52
Gate dielectrics, 164
Gaussian, 135
27
115, 117, 121, 124
Glassy precursors, 32, 33
Grain boundary, 146, 156
Grain growth, 55
Graphite, 69
H
Hematite, 68
Heulandite 33, 34
Hexacyanometallates, 35
37
High resolution electron microscopy (HREM),
117, 118-120,145-160
High-pressure, 24
Holography, 148
Hydrazine, 7
Hydrothermal synthesis, 23-40
I
Intercalation, 74
Interconnections, 163
Interface potential, 121,125
Ionic conductivity, 68
Ionic template, 32
Iron oxide, 53
Isooctane, 8-11
K
KC1, 10, 13, 14
36
KI, 27
Kinetics and thermodynamics in hydrothermal
synthesis, 25
L
38
Langmuir-Blodgett technique, 172, 173, 177
35
66-68
73
73-95
Li-Mn-O spinel, 137-139
Lithium battery, 73
Lorentzian function, 135
M
Magnetic particles, 53
Magnetic Pulse Compaction, 55, 60-62
Mesoporous films, 116
158
153
Modeling, 45, 52, 134
Molecular circuits, 171
Molecular circuits, 179
Molecular diode, 170
Molecular electronics, 161, 169-180
Molecular manipulation by STM, 176, 177
Molecular wire, 171
MOS (Metal-Oxide-Semiconductor), 162
1539
Multilayer, 115-124
N
NaBr, 10, 12
NaCl, 10,12,13, 17
NaCl, 100
NaCl, 51
10, 12, 17
Nanocomposites, 97-113
Nanoelectrodes, 177
Nanoelectronics, 161-183
Nanoparticle, 51
Nanopores, 178
Nanoreactors, 1,4
Nanotubes, 177, 178
Natural formation conditions, 32
Navier-Stokes equation, 46
n-butanol, 47
Neutron diffraction, 145
28
NiO, A6, 37
Normal micelles, 9-17
187
Nucleation, 28, 29, 39
O
111
One-electron transistor, 168
Optical absorption, 121, 122
Optical absorption, 109
Optical band gap, 122-124
Organic materials, 27
Oxidation, 97, 110-112
P
Paracyclophane, 97, 99
Para-xylylene, 99
Particle size, 77, 129
PbS, 6, 7
PbTiO3, 38
Pd, 97, 98, 102, 103, 131, 132, 134, 135, 140,
141
Pd-Au, 133
Pd-Co, 139
135
Percolation, 97, 98, 100, 108, 109, 112
Phase diagram ofwater, 24
Phase diagram, 8, 10
Phase transition, 74, 81, 84
Photocurrent, 118
Photovoltaic devices, 116
Phtalocyanines, 173, 174
Physical Vapour Deposition, 115
Plastic deformation, 56, 57
Polaron, 74, 84, 90
Polyacetylene, 171
Polymer precursor, 99
Polymer, 97-113
Polymer-metal connections, 69
Poly-para-xylylene (PPX), 97-113
Porosity, 109
Precursor transformation, 37
Precursor-product correlation, 32
Precursors in hydrothermal synthesis, 23,28,
32
Pressure wave, 55, 57, 58
Prussian Blue, 35
Pt,3
Q
Quantum confinement, 115, 116, 122
Quantum dots, 123, 124
Quantum dots, 52
Quantum effects, 164
Quantum wire, 165
Quartz, 100
R
Rapid expansion ofsupercritical fluid, 27
Rapid thermal decomposition, 28
Recrystallization, 35
Reverse micelles, 2, 5-8
Rietveld, 136
Rotaxane, 180
S
Scanning Electron Microscopy, 76, 77
Scanning Tunneling Microscope (STM), 174-
178
Scherrer’s equation, 130, 132
Seebeck coefficient, 88
Segregation, 140
Self-assembled monolayer (SAM), 172, 177
Silicon, 162
Single-electron electronics, 169
Sintering, 55
27, 134, 135
115,117,119,122,124
Sn, 97, 101-104, 108, 109
48, 52
Soft Chemistry 1-21
Solar energy conversion, 116
Sol-gel, 74, 75
Spinel, 73
Spray pyrolysis, 52
Spraying modes, 44
38
Static compaction, 55, 56
STM spectroscopy, 176
Sulfosuccinate, 8, 9
Superconductors, 53
Supercritical fluid, 24
Supersaturation, 26, 29
Surface energies, 120
Surface roughness, 120, 121, 124
Surfacetension, 46
Surfactant, 3, 5, 8, 9
T
Template, 4
Tetracyanoquinodimethane (TCNQ), 170
188
Tetrathiofulvalene (TTF), 170
Thermoelectric power, 83, 85, 87, 88, 89
Thermogravimetry, 78
Thin-film, 52
Thiol, 176, 178, 179
28, 37
50, 53
65
Transitional restacking, 56
Transmission Electron Microscopy (TEM), 6,
8, 11-15, 97, 100, 105-108, 129
Tunneling effects, 164-166, 172, 173
Tunneling junction, 166-168
U
Uniaxial pressing, 58, 66
V
Varicose, 45, 47, 51
Vegard’s law, 134, 139
Viscosity, 45
VLSI (Very Large Scale Integration), 163
W
Warren-Averbach method, 131
WC-Co,62
Williamson-Hall method, 131
X
X-ray diffraction, 78, 79, 127-142
Y
39
Z
Zeolites, 32-34
ZnO, 28, 131
ZnS, 5-7, 43
ZnSe, 115, 117, 123, 124
ZnTe, 5
27, 28, 37, 38
68