Transcript
Resume
La nano-structuration topographique des surfaces est un concept prometteur,
notamment pour ameliorer la bio-integration des protheses orthopediques en ti-
tane. Dans cette optique, une nouvelle methode de nano-structuration des surfaces
de titane anodise a ete developpee.
Des asperites ordonnees de l’ordre de quelques dizaines de nanometres de
hauteur ont ete creees en anodisant du titane electropoli en presence de parti-
cules polymeriques deposees en mono-couches. Pour ce faire, un procede existant
d’electropolissage du titane a ete adapte et optimise, permettant la production
de surfaces de depart extremement lisses. Le contact entre les particules et le
substrat a ete mesure, modelise et finalement modifie par le biais de traitements
plasma, thermiques et par vapeurs chimiques, et son influence sur la topographie
de la couche d’oxyde anodique a ete examinee.
Differents types de structures ordonnees ont ainsi ete produits, puis carac-
terises par microscopie a force atomique (AFM) et par microscopie electronique
a balayage (SEM). L’effet des conditions d’anodisation sur la topographie et la
morphologie de la surface a ete etudie.
Afin d’acquerir une meilleure comprehension des phenomenes a l’œuvre, une
experience modele utilisant la lithographie electronique a ete concue. Des rangees
de masques polymeriques de diametre croissant ont ete deposees sur le titane
afin de simuler la presence de particules de tailles correspondantes. L’effet de ces
masques sur la topographie de la couche d’oxyde a ensuite ete mesure par AFM,
SEM et tomographie FIB. Certains aspects des phenomenes de structuration
precedemment observes avec les particules ont ainsi ete eclaircis, en particulier
l’etendue de la propagation de la couche anodisee sous les masques.
Enfin, la modelisation numerique par elements finis a ete utilisee pour simuler
la phase initiale de croissance de la couche anodique autour de masques isolants,
permettant ainsi une meilleure comprehension de la formation de certaines des
structures topographiques observees.
En parallele a ces travaux sur la structuration de surfaces, des particules mul-
tifonctionnelles de silice mesoporeuse ont ete produites. Elles etaient destinees a
l’origine a servir d’alternative aux particules polymeriques utilisees pour creer nos
structurations. Leur procede de synthese, fonde sur la formation de particules par
sol-gel dans une miniemulsion, a ete mis a l’echelle et optimise, permettant la pro-
duction de particules submicroniques etroitement distribuees avec un important
volume de micropores et une forte charge de nanoparticules fonctionnelles.
Ces particules se sont averees inadaptees pour la structuration de surfaces,
mais leur potentiel pour d’autres applications a ete sonde, en fonction du type de
nanoparticles incorpore dans la matrice de silice.
En combinaison avec des nanoparticlules d’oxyde de fer superparamagne-
tiques, elles ont ete evaluees en tant que vecteur magnetique de medicament.
Le chargement et le relarguage du medicament anti-cancer paclitaxel dans les
particules ont ete analyse sur ce systeme modele, par l’experimentation et la si-
mulation. Le relarguage du medicament s’est avere tres lent, restreint par une
grande difference de potentiel chimique entre la molecule en solution et celle ad-
sorbee dans les pores.
Des nanoparticules d’alumine dopee au chrome (rubis) ont ete synthetisees
et caracterisees pour une eventuelle application en tant que marqueurs fluores-
cents pour l’imagerie biomedicale. Leurs proprietes de photoluminescence dans le
proche infrarouge ont ete mesurees et se sont revelees suffisamment intenses pour
etre detectees en microscopie par fluorescence en champ large et en microscopie
confocale a balayage laser. Elle n’ont pas pu en revanche etre incorporees dans
les particules multifonctionnelles lors de la synthese par gelification en emulsion.
Des nanoparticules fluorescentes de sulfure de zinc dope au manganese ont
egalement ete incorporees dans de la silice mesoporeuse, afin de constituer le ma-
teriau de base de surfaces fluorescentes multicolores. Par le biais d’un traitement
thermique local par laser, la longueur d’onde d’emission du materiau a ete mo-
difiee avec succes. Une programmation adequate du laser a ensuite permis de
realiser des images fluorescentes multicolores.
Mots clefs : titane, nano-structuration, topographie, silice mesoporeuses,
particules multifonctionnelles.
iv
Abstract
Topographic nanostructuring of surfaces is a promising concept, in particular
to improve the bio-integration of titanium orthopedic implants. In this context, a
new method for the nanostructuring of anodised titanium surfaces was developed.
Ordered topographic features in the tens of nanometre in height were created
by anodising electropolished titanium in the presence of polymeric particles de-
posited as monolayers. To do so, an existing electropolishing method for titanium
was applied and optimised, allowing the production of extremely smooth starting
surfaces. The particle-substrate contact was mesured, modeled and finally mod-
ified by plasma, thermal and chemical vapour treatments and its effect on the
topography of the anodic oxide layer was investigated.
Different types of ordered structures were produced, and characterised by
atomic force microscopy (AFM) and scanning electron microscopy (SEM). The
influence of the anodisation conditions on the topography and morphology of the
surface was studied.
To gain deeper understanding of the mechanisms at play, a model experiment
using electron beam lithography was designed. Circular masks of increasing di-
ameters were deposited on the surface to simulate the presence of particles of
corresponding sizes, and the effect on the topography of the oxide layer was char-
acterised. Some aspects of the structuring phenomena observed with the particles
were thus cleared up, in particular the extent of oxide layer growth underneath
the masks.
Finally, numerical finite element modeling was applied to simulate the initial
stages of anodic oxide growth around masks, leading to a better understanding
of the formation of some of the topographic features observed.
In parallel to this work on surface structuring, mesoporous silica particles
containing different functional nanoparticles were produced. They were origi-
nally destined to serve as an alternative to the polymeric particles used to create
our structurations. Their synthesis process, based on the formation of particles
by sol-gel in a miniemulsion, was scaled-up and optimised to obtain narrowly dis-
tributed submicron particles with a high micropore volume and a high functional
nanoparticle loading.
These multifunctional particles were finally found to be unadapted for surface
structuring but were assessed for various other applications, depending on the
type of nanoparticles incorporated in the silica matrix.
In combination with superparamagnetic iron oxide nanoparticles (SPIONS),
they were evaluated as potential magnetic drug delivery vehicles. Drug loading
and release of the anticancer drug paclitaxel was studied as a model system by
simulation and experiment. The release kinetics was found to be very slow, re-
stricted by the large chemical potential difference between the molecule in solution
and the molecule adsorbed inside micropores.
Chromium doped alumina (ruby) nanoparticles were synthesised and char-
acterised for potential application as near-IR fluorescent markers for biomedical
imaging. Their photoluminescent properties were studied and were found to be
sufficiently intense for bright-field fluorescence microscopy and confocal laser scan-
ning microscopy. They could however not be incorporated in the multifunctional
particles during the gel emulsion synthesis.
Manganese doped zinc sulphide quantum dots were also encapsulated in a
mesoporous silica matrix and used as the basis material for the production of
multicolour fluorescent surfaces. Localised laser thermal treatment of the mate-
rial was successfully realised, causing a shift in its fluorescence emission wave-
length. Multicolour fluorescent images could thus be drawn by programing the
laser appropriately.
Keywords: titanium, nanostructuring, topography, anodisation, mesoporous
silica, multifunctional particles.
A Gabrielle et a Joseph.
Life, oh life, oh life. . .
Des’ree
Acknowledgements
I want to thank my thesis director, Pr. Hofmann for his support,
guidance and trust, and for the great freedom he left me. I hope it
was put to good use. I also thank Pr. Alke Fink for her moral support
throughout the years.
I would also like to thank the Entwicklungsfond fur Seltene Metalle
(ESM) for their financial support of this project.
Futhermore, I would like to thank the following people:
� The other members of my thesis jury, Pr. Marcus Textor, Pr.
Thomas Graule and Dr. Pierre-Francois Chauvy for their very
constructive review of the present work.
� Michael Stuer and Amelie Bazzoni for their commitment in their
semester theses, as well as Pr. Patrick Hoffmann, Bamdad Afra
for their expertise and their time.
� Pr. Stefano Mischler, Jean-Daniel Neuvecelle and Julien Perret of
LMCH for their help with electrochemistry and their equipment.
� The people at the atelier MX, Pierre-Andre Desponts, Werner
Bronnimann, Yves Ruschetta and Adrien Grisendi, for their help
and for their competence.
� Pr. Jacques Lemaitre and Dr. Paul Bowen for their help with
experimental methodology and particle size measurements
� All the people at LTP, for their friendship and for making these
three and a half years a very pleasant time.
Contents
List of Figures ix
List of Tables xv
1 Introduction 1
1.1 Context of the work . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Objectives of the project . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Structure of the thesis . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Multifunctional mesoporous silica particles 7
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 State of the art . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Functional nanoparticles used . . . . . . . . . . . . . . . . 9
2.1.2.1 Superparamagnetic iron oxide nanoparticles . . . 9
2.1.2.2 Chromium doped alumina nanoparticles . . . . . 9
2.1.2.3 Manganese doped zinc sulphide nanoparticles . . 9
2.2 Gel emulsion synthesis . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Previous work . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.3 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3.1 Emulsions . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3.2 Coarsening of emulsions . . . . . . . . . . . . . . 13
2.2.3.3 The AOT-octane-water system . . . . . . . . . . 14
2.2.3.4 Sol-gel process . . . . . . . . . . . . . . . . . . . 15
2.2.4 Materials and methods . . . . . . . . . . . . . . . . . . . . 16
iii
CONTENTS
2.2.5 Results and discussion . . . . . . . . . . . . . . . . . . . . 17
2.2.5.1 Scale-up of the emulsion . . . . . . . . . . . . . . 17
2.2.5.2 Kinetic evolution . . . . . . . . . . . . . . . . . . 19
2.2.5.3 Optimisation of emulsion parameters . . . . . . . 21
2.3 Materials characterisation . . . . . . . . . . . . . . . . . . . . . . 25
2.3.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.2 Results and discussion . . . . . . . . . . . . . . . . . . . . 25
2.4 Mesoporous silica spheres as drug delivery vehicles . . . . . . . . . 30
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.2 Materials and methods . . . . . . . . . . . . . . . . . . . . 31
2.4.2.1 Paclitaxel loading . . . . . . . . . . . . . . . . . . 31
2.4.2.2 Paclitaxel release . . . . . . . . . . . . . . . . . . 31
2.4.2.3 Instrumentation and measurements . . . . . . . . 31
2.4.2.4 Simulation . . . . . . . . . . . . . . . . . . . . . 32
2.4.3 Results and discussion . . . . . . . . . . . . . . . . . . . . 33
2.4.3.1 Paclitaxel adsorption . . . . . . . . . . . . . . . . 33
2.4.3.2 Simulation . . . . . . . . . . . . . . . . . . . . . 35
2.4.3.3 Paclitaxel release . . . . . . . . . . . . . . . . . . 37
2.4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.5 Chromium doped alumina . . . . . . . . . . . . . . . . . . . . . . 44
2.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.5.2 Materials and methods . . . . . . . . . . . . . . . . . . . . 44
2.5.3 Results and discussion . . . . . . . . . . . . . . . . . . . . 45
2.5.3.1 Materials characterisation . . . . . . . . . . . . . 45
2.5.3.2 Characterisation of fluorescence . . . . . . . . . . 49
2.5.3.3 Effect of chromium concentration and annealing
time on fluorescence . . . . . . . . . . . . . . . . 51
2.5.3.4 Fluorescence imaging . . . . . . . . . . . . . . . . 53
2.5.4 Application to multifunctional particles . . . . . . . . . . . 55
2.5.4.1 Effect of SPIONS on fluorescence . . . . . . . . . 55
2.5.4.2 Synthesis of MS-SPION-Ruby multifunctional par-
ticles . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.5.5 Conclusion and outlook . . . . . . . . . . . . . . . . . . . . 58
iv
CONTENTS
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles . . . . . . 59
2.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.6.2 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.6.2.1 Silica coated ZnS:Mn2+ nanoparticles . . . . . . . 59
2.6.2.2 Effect of annealing . . . . . . . . . . . . . . . . . 61
2.6.3 Materials and methods . . . . . . . . . . . . . . . . . . . . 61
2.6.4 Results and discussion . . . . . . . . . . . . . . . . . . . . 64
2.6.4.1 Oven annealing . . . . . . . . . . . . . . . . . . . 64
2.6.4.2 Laser annealing . . . . . . . . . . . . . . . . . . . 65
2.6.4.3 Cathodoluminescence . . . . . . . . . . . . . . . 70
2.6.5 Conclusion and outlook . . . . . . . . . . . . . . . . . . . . 72
3 Titanium surface structuring 77
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.1.1 Titanium . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.1.2 Electrochemical properties . . . . . . . . . . . . . . . . . . 78
3.1.2.1 Passivity . . . . . . . . . . . . . . . . . . . . . . 78
3.1.2.2 Anodic oxide film growth . . . . . . . . . . . . . 79
3.2 Structuring of titanium surface . . . . . . . . . . . . . . . . . . . 80
3.3 Electrochemical polishing of titanium . . . . . . . . . . . . . . . . 83
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.3.2 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.3.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . 86
3.3.3.1 Electrolyte . . . . . . . . . . . . . . . . . . . . . 86
3.3.3.2 Sample preparation . . . . . . . . . . . . . . . . . 87
3.3.3.3 Rotating disc electrode . . . . . . . . . . . . . . . 87
3.3.3.4 Multisample rotating electrode . . . . . . . . . . 87
3.3.3.5 Electrochemical cell . . . . . . . . . . . . . . . . 87
3.3.3.6 Multisample electrochemical cell . . . . . . . . . 89
3.3.3.7 Roughness measurements . . . . . . . . . . . . . 90
3.3.3.8 Roughness parameter . . . . . . . . . . . . . . . 90
3.3.4 Experimental optimisation of polishing . . . . . . . . . . . 91
3.3.4.1 Experiment plan . . . . . . . . . . . . . . . . . . 91
v
CONTENTS
3.3.4.2 Results and discussion . . . . . . . . . . . . . . . 91
3.3.5 Other factors affecting polishing. . . . . . . . . . . . . . . 95
3.3.5.1 Effect of water. . . . . . . . . . . . . . . . . . . . 95
3.3.5.2 Sources of water contamination. . . . . . . . . . . 96
3.3.5.3 Non-anodic current limitations. . . . . . . . . . . 97
3.3.6 Assessment of the different polishing setups. . . . . . . . . 98
3.3.6.1 Rotating electrode setups . . . . . . . . . . . . . 98
3.3.6.2 Fixed electrode setups . . . . . . . . . . . . . . . 98
3.4 Particle-substrate contact . . . . . . . . . . . . . . . . . . . . . . 101
3.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.4.2 Materials and methods . . . . . . . . . . . . . . . . . . . . 101
3.4.3 Results and discussion . . . . . . . . . . . . . . . . . . . . 102
3.4.3.1 Pristine particles . . . . . . . . . . . . . . . . . . 102
3.4.3.2 Effect of heat . . . . . . . . . . . . . . . . . . . . 104
3.4.3.3 Effect of acetone . . . . . . . . . . . . . . . . . . 109
3.5 Particle deposition . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.5.2 Materials and methods . . . . . . . . . . . . . . . . . . . . 112
3.5.3 Dip coating . . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.5.4 Drop drying . . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.6 Air plasma treatments . . . . . . . . . . . . . . . . . . . . . . . . 116
3.6.1 Air plasma for particle deposition . . . . . . . . . . . . . . 116
3.6.2 Air plasma for SAM preparation . . . . . . . . . . . . . . 116
3.6.3 Air plasma for cleaning . . . . . . . . . . . . . . . . . . . . 118
3.7 Surface structuring with polystyrene particles . . . . . . . . . . . 121
3.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.7.2 Materials and methods . . . . . . . . . . . . . . . . . . . . 121
3.7.3 Results and discussion . . . . . . . . . . . . . . . . . . . . 122
3.7.3.1 Influence of the contact radius . . . . . . . . . . . 122
3.7.3.2 Effect of particle deposition . . . . . . . . . . . . 127
3.7.3.3 Influence of the anodisation voltage . . . . . . . . 129
3.7.3.4 Influence of the anodisation profile . . . . . . . . 131
3.7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
vi
CONTENTS
3.8 Model experiment using e-beam lithography . . . . . . . . . . . . 133
3.8.1 Aim of the experiment . . . . . . . . . . . . . . . . . . . . 133
3.8.2 Materials and methods . . . . . . . . . . . . . . . . . . . . 135
3.8.3 Results and discussion . . . . . . . . . . . . . . . . . . . . 136
3.8.3.1 E-beam lithography . . . . . . . . . . . . . . . . 136
3.8.3.2 Effect of the masks on topography . . . . . . . . 138
3.8.3.3 Effect of anodisation parameters . . . . . . . . . 141
3.8.3.4 Cross-sections by focused ion beam . . . . . . . . 144
3.8.3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . 146
3.9 Finite element modeling . . . . . . . . . . . . . . . . . . . . . . . 149
3.9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 149
3.9.2 Single physic model . . . . . . . . . . . . . . . . . . . . . . 149
3.9.3 Two physics model . . . . . . . . . . . . . . . . . . . . . . 151
3.9.4 Results and discussion . . . . . . . . . . . . . . . . . . . . 153
3.9.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
4 Conclusion and outlook 157
Bibliography 159
vii
CONTENTS
viii
List of Figures
1.1 Envisaged process flows for surface structuring with sacrificial polystyrene
(PS) particles as template. . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Envisaged process flows for surface structuring with multifunc-
tional particles as inclusions in the oxide layer. . . . . . . . . . . . 4
2.1 Scheme of the emulsification process. . . . . . . . . . . . . . . . . 11
2.2 Phase diagram for the AOT-octane-water system. . . . . . . . . . 14
2.3 High shear rotor stator homogeniser. . . . . . . . . . . . . . . . . 18
2.4 Kinetic evolution the emulsion by dynamic light scattering. . . . . 19
2.5 Kinetic evolution of the emulsion by dynamic light scattering. . . 20
2.6 Experimental values and values predicted by the statistical model
(Adjusted values) for all the different treatments. . . . . . . . . . 22
2.7 Effect of the factors on the particles’ mean volume diameter. . . . 22
2.8 SEM micrographs of composite silica-SPION particles. . . . . . . 25
2.9 Particle size distribution of composite silica-SPION particles. . . . 27
2.10 Nitrogen adsorption and desorption isotherms. . . . . . . . . . . . 28
2.11 Incremental pore diameter distribution in volume. . . . . . . . . . 29
2.12 Paclitaxel molecule. . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.13 Adsorption isotherm of paclitaxel on silica-SPION particles. . . . 33
2.14 Paclitaxel release profile in water. . . . . . . . . . . . . . . . . . . 38
2.15 Configuration of paclitaxel in a 2 nm diameter pore . . . . . . . . 40
2.16 Schematic representation of the chemical potential gradient at the
pore opening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.17 Calculated paclitaxel concentration profile inside the pores. . . . . 42
ix
LIST OF FIGURES
2.18 Cumulative particle size distribution in volume for pristine HT-100
Condea powder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.19 Cumulative and frequency particle size distribution in volume for
annealed chromium doped alumina after attrition. . . . . . . . . . 47
2.20 X-ray diffraction spectrum of chromium doped alumina annealed
at 900 ◦C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.21 Excitation and emission spectra of chromium doped alumina par-
ticles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.22 Emission intensity decay over time. . . . . . . . . . . . . . . . . . 50
2.23 Bi-factorial plan for the fluorescence response of chromium doped
alumina. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.24 Fluorescence response surface for chromium doped alumina: influ-
ence of annealing time and chromium concentration. . . . . . . . . 52
2.25 Fluorescence microscopy images of chromium doped alumina ag-
glomerate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.26 Fluorescence microscopy images of chromium doped alumina. . . . 54
2.27 Effect of SPION concentration on the fluorescence emission inten-
sity of chromium doped alumina. . . . . . . . . . . . . . . . . . . 55
2.28 Relative fluorescence intensity for different SPION concentrations. 56
2.29 Phase diagram between iron oxide and chromium oxide. . . . . . . 57
2.30 Jablonski diagram for ZnS. . . . . . . . . . . . . . . . . . . . . . . 60
2.31 Different sample preparation methods for laser annealing. . . . . . 62
2.32 Colours obtained after annealing of ZnS:Mn2+@SiO2 at different
temperatures and in different atmospheres. . . . . . . . . . . . . . 64
2.33 Emission spectra of annealed ZnS:Mn2+@SiO2 samples. . . . . . . 65
2.34 Array of laser impact points. . . . . . . . . . . . . . . . . . . . . . 67
2.35 Laser annealing of ZnS:Mn2+@SiO2 powder in different conditions. 68
2.36 EPFL logo drawn by IR laser on a bed of ZnS:Mn2+@SiO2. . . . . 69
2.37 Emission spectra of the laser treated and non treated zones by
cathodoluminescence. . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.38 Cathodoluminescence of ZnS:Mn2+@SiO2 before and after laser
treatment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
x
LIST OF FIGURES
2.39 SE image of compacted ZnS:Mn2+@SiO2 surface and chemical map-
ping of the surface. . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.40 SE image of laser treated ZnS:Mn2+@SiO2 surface and chemical
mapping of the surface. . . . . . . . . . . . . . . . . . . . . . . . . 75
3.1 Pourbaix diagram for titanium in aqueous environment. . . . . . . 79
3.2 Levelling effect. . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.3 Titanium sample before polishing. . . . . . . . . . . . . . . . . . 86
3.4 Rotating electrode setups. . . . . . . . . . . . . . . . . . . . . . . 88
3.5 Multisample electrochemistry cell. . . . . . . . . . . . . . . . . . . 89
3.6 Results: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.7 Effect of the factors. . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.8 Response surface at 20 V. . . . . . . . . . . . . . . . . . . . . . . 94
3.9 AFM-Typical polishing defects caused by small amounts of water
in the electrolyte. . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.10 Multiscale roughness measurement from AFM profiles for samples
polished before and after optimisation. . . . . . . . . . . . . . . . 99
3.11 AFM-Polished titanium surface. . . . . . . . . . . . . . . . . . . . 100
3.12 Contact diameter measurement. . . . . . . . . . . . . . . . . . . . 102
3.13 Linear plot of the contact radius a0 against the square root of the
particle radius. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.14 Effect of heat on the contact diameter. . . . . . . . . . . . . . . . 105
3.15 Geometry of a deformed elastic sphere. . . . . . . . . . . . . . . 106
3.16 Elastic, surface and total energies for the spreading of 200 nm and
500 nm PS particles at 150 °C. . . . . . . . . . . . . . . . . . . . . 108
3.17 500 and 200 nm deformed particles (150 °C for 5 minutes). . . . . 109
3.18 Capillary condensation around a spherical particle in presence of
a vapour. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.19 Effect of acetone capillary condensation. . . . . . . . . . . . . . . 110
3.20 Schematic illustration of the phenomenon of particle diffusion to-
ward the meniscus encountered in processes such as dip coating
and drop-drying. . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.21 Different deposits obtained by dip coating of 200 nm particle. . . 114
xi
LIST OF FIGURES
3.22 Self assembled monolayers by drop drying. . . . . . . . . . . . . . 115
3.23 Dense monolayer of 500 nm particles before (A) and after (B) 2
minutes oxygen plasma treatment. . . . . . . . . . . . . . . . . . . 117
3.24 Ordered array of hemispherical PS particles. . . . . . . . . . . . . 117
3.25 200 nm diameter particles after 2 minutes plasma treatment. . . . 118
3.26 Deformed particle after different plasma treatment times. . . . . . 119
3.27 Deformed particle array after different plasma treatment times. . 120
3.28 Structuring with particles. . . . . . . . . . . . . . . . . . . . . . . 122
3.29 Sample coated with 200 nm particles. . . . . . . . . . . . . . . . . 123
3.30 AFM image of anodised samples. . . . . . . . . . . . . . . . . . . 124
3.31 Topography profile across protrusions. . . . . . . . . . . . . . . . 124
3.32 Effect of heat deformed particles on the topography. . . . . . . . . 125
3.33 Surface structuring with deformed particles. . . . . . . . . . . . . 125
3.34 Profile across the imprint of a deformed particle. . . . . . . . . . . 126
3.35 Surface structuring with self-assembled monolayers of deformed
particles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
3.36 Regular honeycomb structurations caused by SAM’s of deformed
particles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
3.37 Details of structurations by SAM’s. . . . . . . . . . . . . . . . . . 128
3.38 Imprint of deformed particles after anodisation at different voltages.129
3.39 Height of topographic features as a function of anodisation voltage. 130
3.40 Structure of the oxide layer underneath the particles. . . . . . . . 131
3.41 Scheme of the different anodisation treatments. . . . . . . . . . . 132
3.42 Process flow and pattern for e-beam lithography. . . . . . . . . . . 134
3.43 Overview of the e-beam patterns for different electron dosages. . . 136
3.44 Arrays of masks by e-beam lithography. . . . . . . . . . . . . . . . 137
3.45 Tilt view of the 4 µm, 800 nm, 400 nm and 100 nm masks. . . . . 137
3.46 2 µm mask at different stages of the process. . . . . . . . . . . . . 138
3.47 SEM image of the e-beam sample after anodisation and removal of
the masks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
3.48 Tilt SEM image of the e-beam sample after anodisation and re-
moval of the masks. . . . . . . . . . . . . . . . . . . . . . . . . . . 139
xii
LIST OF FIGURES
3.49 AFM topography of the e-beam sample after anodisation and clean-
ing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3.50 AFM topography of the e-beam sample after anodisation and clean-
ing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3.51 Height profiles across the different masks. . . . . . . . . . . . . . . 141
3.52 Effect of mask diameter on the edge structure. . . . . . . . . . . . 142
3.53 Effect of the anodisation parameters. . . . . . . . . . . . . . . . . 142
3.54 Profiles across the different masks for different anodisation condi-
tions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
3.55 Damage through a 100 µm mask after anodisation at 55 V. . . . . 144
3.56 Different steps of the FIB technique. . . . . . . . . . . . . . . . . 145
3.57 FIB cross-sections across different e-beam imprint. . . . . . . . . . 148
3.58 Scheme of the model system. . . . . . . . . . . . . . . . . . . . . . 150
3.59 Current density around a mask. . . . . . . . . . . . . . . . . . . . 151
3.60 Oxide layer with mask. . . . . . . . . . . . . . . . . . . . . . . . . 152
3.61 Oxide layer growth around a mask. . . . . . . . . . . . . . . . . . 153
3.62 Detail of the edges at different anodisation times. . . . . . . . . . 154
3.63 Oxide growth between two adjacent masks. . . . . . . . . . . . . . 154
xiii
LIST OF FIGURES
xiv
List of Tables
2.1 Structure of the emulsion experiment plan. . . . . . . . . . . . . . 21
2.2 List of emulsification experiments. . . . . . . . . . . . . . . . . . . 21
2.3 Typical material characteristics. . . . . . . . . . . . . . . . . . . . 27
2.4 Fitting parameters for Langmuir’s isotherm. . . . . . . . . . . . . 34
2.5 Calculated differential adsorption enthalpies. . . . . . . . . . . . . 35
2.6 Diffusion coefficients for different cases. . . . . . . . . . . . . . . . 36
2.7 Material properties of chromium doped alumina. . . . . . . . . . . 48
2.8 List of parameters tested. . . . . . . . . . . . . . . . . . . . . . . 68
3.1 Structure of the experiment plan. . . . . . . . . . . . . . . . . . . 91
3.2 List of experiments. . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.3 Properties of polystyrene. . . . . . . . . . . . . . . . . . . . . . . 103
3.4 Contact radii predicted by the JKR model vs. measured contact
radii. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.5 Calculated Young’s modulus according to the Lau et al. model. . 108
3.6 Electrical conductivities. . . . . . . . . . . . . . . . . . . . . . . . 150
3.7 Boundary conditions. . . . . . . . . . . . . . . . . . . . . . . . . . 152
xv
LIST OF TABLES
xvi
1
Introduction
1
1. INTRODUCTION
1.1 Context of the work
Surface structuring of titanium oxide surfaces is a promising concept that finds
applications in various fields, such as biomedical engineering [64, 77, 128] and
micro-systems applications [10, 55].
In this context, The aim of this work is to create a versatile technology plat-
form making it possible to create precise and reproducible surface topographic
structures in the nanometer range (10 to a few hundred nanometer in height) on
titanium surfaces by an electrochemical route. Such a platform, together with
the understanding of the underlying science, could eventually open new fields of
application.
Most attempts at nanostructuring titanium surfaces were done either by top-
down methods such as microcontact printing and lithography, or by anodisation
of titanium in different electrolytes, yielding characteristic structures such as tita-
nium oxide nanotubes for anodisation in the presence of fluorine ions, or more or
less controlled pore structures in other types of electrolytes [42, 50, 109, 121, 128].
These features are difficult to control and to order satisfactorily and such pro-
cesses often result in a more or less random topography.
Another possibility is to create topographical structures by anodisation of ti-
tanium in the presence of particles. This approach can potentially present several
advantages. Provided one achieves a good control of the particle size distribution
on the one hand, and a good understanding of particle deposition processes on
surfaces on the other, one can create arrays of adsorbed particles with a high
degree of order and well defined characteristic length scales.
It is a bottom-up approach (self-assembly of particles on surfaces) and there-
fore promises be comparatively cheap, quick and easy compared to expensive
top-down processes such as traditional lithography techniques. These adsorbed
particles then act as masks, allowing a selective exposure of the surface, to anodic
oxidation in our case.
In a related theme, the functionalisation of surfaces by adsorption of functional
microparticles has been realised, for example to confer antiseptic properties to
implant surfaces in order to prevent septic implant loosening [25]. Alternatively,
2
1.2 Objectives of the project
Figure 1.1: Envisaged process flows for surface structuring with sacrifi-
cial polystyrene (PS) particles as template. Anodisation and removal of the
particles.
encoded fluorescent microparticles have been immobilised on surfaces for bio-
sensor applications [22, 60].
An interesting concept would be to link these two ideas, and to perform parti-
cle lithography using adsorbed multifunctional microparticles, so as to make use
of their structural properties in the lithography process, and then keep them ad-
sorbed on the surface, and make use of their other properties, such as for example
drug elution or fluorescence.
1.2 Objectives of the project
The original aim of this project was to explore the feasibility of achieving ordered
structuring of titanium surfaces by anodisation in the presence of particles. The
purpose was not so much to optimise a process of a particular application as to
find creative ways to achieve new types of surface structures on the nanometer
scale, and attempt to understand the phenomena observed.
Two concepts were originally considered:
3
1. INTRODUCTION
The first one was to perform the anodisation of a titanium surface covered
with polymeric submicron particle in the hope that the particles would act as
masks, locally preventing the growth of the oxide film, and that the film would
then grow around the particles, partially or completely embedding them inside
the newly formed oxide layer. This would create, after thermal removal of the
particles, a more or less structured topography on the titanium surface, with a
typical features height of the order of the layer thickness, as illustrated on figure
1.1
The second concept considered was to anodise the titanium surface in the
presence of mesoporous silica (MS) composite multifunctional particles in the
hope that they could be embedded inside the oxide layer during anodisation, and
remain there. These particles are composed of a mesoporous silica matrix in which
different functional nanoparticles (superparamagnetic iron oxide nanoparticles,
fluorescent quantum dots) are encapsulated. Thus, in addition to the topographic
effect, the inclusion of these particles in the surface could confer it a whole new
range of desirable properties (figure 1.2).
Figure 1.2: Envisaged process flows for surface structuring with mul-
tifunctional particles as inclusions in the oxide layer. Embedded particles
retain their functionalities such as drug elution, fluorescence or magnetic properties
in addition to topographic structuring.
These two directions were developed in parallel, on one side attempting to
4
1.2 Objectives of the project
understand, by characterisation and modeling, the phenomena a play during an-
odisation in the presence of model commercial polymeric particle, and on the
other synthesising multifunctional particles with different properties that could
eventually be applied to surface structuring.
The combination of these two research directions was however found to be
unrealistic. The advances and experience gained in surface structuring by anodi-
sation made it gradually clear that both the growth mechanism of the oxide layer
and the properties of the synthesised multifunctional particles rendered the idea
irrelevant.
Our results showed that when anodisation was performed in the presence
of adsorbed particles of the order of size of the oxide layer thickness (<5 µm),
the oxide layer did not grow around the particles leaving an hole, as illustrated
on figure 1.1 and as was originally conjectured. Instead, the oxide layer grew
unhindered underneath the particles, lifting them up with the rest of the surface.
As a result, the idea of embedding multifunctional particles inside the oxide layer
was not achievable by anodisation.
Furthermore, the effect of adsorbed particles on the topography of the anodic
layer was shown to be dependent on the particle-substrate contact area. In the
case of the model polymeric particles used, this contact area could be increased
by a heat treatment, causing a softening of the particles and their spreading
on the surface. This spreading was however nor feasible with the silica based
multifunctional particles, due to their mechanical properties, and the particle-
substrate contact area could not be increased in that case.
Consequently, both research directions were carried on independently and no
combination of the two technologies was attempted in the end.
SPION-loaded MS particles were studied for potential application in drug
delivery. In this perspective, an existing gel-emulsion synthesis method was stud-
ied and optimised to improve the production yield as well as the particle size
distribution. In addition, two other types of functional nanoparticles were stud-
ied in combination with mesoporous silica. Near IR fluorescent chromium-doped
alumina nanoparticles were investigated as potential fluorescent labels in bio-
imaging, and manganese doped zinc sulphide nanoparticles were applied in the
production of multicolour fluorescent displays.
5
1. INTRODUCTION
1.3 Structure of the thesis
The structure of the thesis reflects what is explained above, and is divided into
two main chapters. All aspects related to the synthesis, characterisation and
application of the multifunctional particles are treated in chapter two, and all the
work done on surface structuring of titanium is presented in chapter three.
6
2
Multifunctional mesoporous silica
particles
7
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
2.1 Introduction
2.1.1 State of the art
Mesoporous silica (MS), thanks to its high pore volume, tunable pore size, high
specific surface area and chemical inertness has been widely applied as a support
material in various areas such as enzyme immobilization, drug and gene delivery,
chromatography, adsorbents and catalysis [3, 38, 96, 111]. A notable trend is
the combination of the porous properties of MS with functional nanoparticles
such as superparamagnetic iron oxide nanoparticles (SPIONS) and/or fluorescent
semi-conductor nanoparticles (Quantum dots) to form multifunctional particles
[53, 125].
The challenge for any synthesis and envisaged application is to achieve simul-
taneous control of the pore morphology, of the macroscopic morphology (particle
size and size distribution) and of the functional nanoparticle loading. While pore
morphology can be controlled by appropriate synthesis conditions and the use
of self assembled surfactant templates, control of the macroscopic morphology is
difficult and has been an important hurdle in the development of these materi-
als [5]. Methods such as modified Stober processes, or microemulsion synthesis
can yield very narrow submicron size distributions and homogeneous, tunable
pore diameter and are often used to produce well characterized core shell type
nanoparticles [5, 62, 76, 106, 119, 125]. They are however less adapted to achieve
high nanoparticle loading inside the MS sphere, which is important to enhance
the desired functionalities.
Other methods, such as the one presented here, rely on metastable morphol-
ogy templates to form the desired product. In aerosol–assisted synthesis and
inverse miniemulsion synthesis, the droplets of liquid in air or of liquid in liquid
determine the final morphology of the particles [58, 90]. The main advantage
of these methods is that they allow the encapsulation of large amounts of func-
tional nanoparticles by dispersing them in the silica precursor, whilst retaining
the porous properties of MS silica. Their main disadvantage is that they make
it difficult to control the particle size and size distribution. They typically pro-
duce particles diameters in the tens of micrometer range with broad distributions.
8
2.1 Introduction
The challenge is therefore to produce well defined, narrowly distributed submi-
cron mesoporous particles with well defined pore diameter and high functional
nanoparticle loading.
2.1.2 Functional nanoparticles used
2.1.2.1 Superparamagnetic iron oxide nanoparticles
SPIONS have been used for separation techniques, site specific drug delivery,
hyperthermia and are a powerful contrast agent for magnetic resonance imaging
(MRI) [28, 75, 93, 114].
Magnetic Carrier Systems (MCS) combining their magnetic properties with
those of MS have been increasingly studied and used in industry, particularly
in diagnostics and bioseparation [78, 96]. A variety of synthesis approaches are
found in the literature. Hematite nanoparticles were provided with a mesoporous
silica shell, again using a sol-gel approach [76, 124]. Studies exist where the
iron oxide nanoparticles have been precipitated in situ in the pore volume of
the pre-formed silica particles, making it possible to achieve a high magnetic
content (19 wt%) inside monodisperse particles, at the expense however of the
pore volume [74]. Shao et al. [96] have formed magnetic silica particles through a
direct miniemulsion of tetraethyl orthosilicate (TEOS) in water, with hydrophobic
SPIONS dispersed in the TEOS phase, achieving high magnetic loading (10 to 30
wt%) and narrow size distribution, without however paying particular attention
to the porosity.
2.1.2.2 Chromium doped alumina nanoparticles
Chromium doped alumina nanoparticles were investigated for potential applica-
tion as a near IR fluorescent markers for bio-imaging applications.
2.1.2.3 Manganese doped zinc sulphide nanoparticles
ZnS:Mn2+ nanoparticles were extensively studied as fluorescent markers by B.
Steitz at LTP, who attempted to combine them to SPION containing systems
9
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
[100]. Their optical interaction with SPIONS however quenches most of their
fluorescence, which limits the interest of such an association.
In this study, laser thermal annealing of these fluorescent nanoparticles in
silica was attempted for the fabrication of fluorescent displays.
.
10
2.2 Gel emulsion synthesis
2.2 Gel emulsion synthesis
2.2.1 Principle
The synthesis of multifunctionnal MS particles by a gel emulsion process consists
in dispersing an aqueous silica precursor into a non-miscible oil phase in order
to form a homogeneous miniemulsion. Functional nanoparticles are present as
colloids in the dispersed phase.
Figure 2.1: Scheme of the emulsification process.
The silica precusor is then caused to undergo a polycondensation reaction and
form amorphous mesoporous silica. The miniemulsion droplets act as a morphol-
ogy template for the formation of solid spherical MS particles. The functional
nanoparticles suspended in the silica precursor are thus encapsulated inside the
particles.
2.2.2 Previous work
The formulation of the emulsion system used was developed by F. Krauss and B.
Steitz [51] in the framework of a master’s thesis at EPFL. The emulsification was
11
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
performed in their case using an ultrasound horn on small volumes of emulsion
(up to 5 ml).
2.2.3 Theory
2.2.3.1 Emulsions
Emulsions are a type of colloid consisting in the mixture of two immiscible liq-
uids such as water and oil [11]. They are macroscopically homogeneous but
microscopically heterogeneous as they are composed of two separate phases, one
phase generally dispersed in the other in the form of spherical droplets. There
are both oil-in-water emulsions (oil droplets in water, also called direct emulsions)
and water-in-oil emulsions, droplets of water dispersed in oil, also called reverse
emulsions).
The mixture can be stabilised by a third component class called surfactants.
A surfactant is a chemical compound which is preferentially adsorbed at a
surface or at an interface and which modifies its physico-chemical properties. It
is a so-called amphiphilic molecule, composed of a hydrophobic and a hydrophilic
group. The presence of surfactants in an emulsion tends to decrease the inter-
facial energy between the two phases, and slow down its separating back into a
macroscopic two-phase system.
Surfactants tend to for self-assembled aggregates in solution [29]. The simplest
and best understood type of aggregate is the micelle, a spherical aggregate of
surfactant molecules (see figure 2.1).
As one increases the surfactant concentration in an emulsion, the oil-water
interface is gradually populated by surfactant molecules, decreasing the interfa-
cial energy, until at last it becomes saturated. At this concentration, known as
the critical micelle concentration (CMC), the interfacial energy does not change
significantly anymore and the extra surfactant molecules start forming micelles
in solution.
Microemulsions are a specific subset of emulsion that have droplet with diam-
eters in the range of 1 to 20 nm. The main particularity of this class of emulsion
is that they are thermodynamically stable systems, as opposed to classical emul-
sions, which are metastable and have a tendency to be destroyed in time. They
12
2.2 Gel emulsion synthesis
are formed without extensive mechanical work. The surfactant micelles present
in the continuous phase can swell up to a certain extent and accommodate a small
amount of the other phase in their center in a process called micellar solubilisa-
tion. These micellar solutions are usually clear homogeneous liquids.
Miniemulsions are metastable emulsions with a droplet size usually comprised
between 50 and 500 nm. They are formed by applying high shear to the mixture
using various techniques such as ultrasonication, high pressure homogenisers or
high shear rotor stator turbines.
2.2.3.2 Coarsening of emulsions
There are four main mechanism for the coarsening and eventual destruction of
metastable emulsions:
Creaming and sedimentation is caused by a mismatch in the densities of the
two phases, which causes the dispersed phase to either float (creaming) or
sink (sedimentation). This phenomenon is faster with increasing droplets
or droplet flocs diameter.
Coalescence is the rupture of the interfacial film between two droplets causing
them to merge into a single large droplet. Proximity of the droplets due to
sedimentation, creaming or flocculation increases the risk of coalescence.
Ostwald’s ripening is the gradual coarsening of larger droplets to the expense
of smaller ones. The driving force behind it is the difference in free surface
energy between droplets of different sizes. It consists in a transfer of matter
from small droplets to larger ones through the continuous phase. This is
aggravated by the small residual solubility of the dispersed phase in the
continuous phase.
Flocculation is the aggregation of droplets due to insufficient colloidal sta-
bilisation forces. Flocs and single droplets can coexist at equilibrium in
suspension [57].
13
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
2.2.3.3 The AOT-octane-water system
Sodium-bis(2 ethylhexyl) sulfosuccinate (or aerosol OT, AOT) is a surfactant
with a polar head and two apolar tails. Phase diagrams could not be found in
literature for the exact AOT-octane-water system, but several publications have
studied similar systems in particular AOT-isooctane-water [104, 122] or AOT-
decane-water [14] systems.
The general aspect of these phase diagrams being quite alike, an approximate
ternary phase diagram can be drawn for our system (figure 2.2)
Figure 2.2: Phase diagram for the AOT-octane-water system.
The location of our formulation is near the top corner of the triangle, at low
water and AOT contents, in the two phases zone.
The critical micelle concentration in octane is of the order of 1 · 10−3 mol/l
[72].
14
2.2 Gel emulsion synthesis
2.2.3.4 Sol-gel process
Sol-gel processes are a type of wet chemical process by which a suspension of col-
loidal particles, or a solution of monomer, assemble to create a solid loose 3D net-
work (gel). Typical sol-gel systems involve metal alkoxides, such as tetramethyl
orthosilicate (TMOS), a widely used silica precursor, and the starting chemical
used throughout this study.
The sol -gel reaction comprises two successive steps, namely a hydrolysis re-
action followed by a polycondensation reaction. The hydrolysis reaction consists
in the replacement of a methoxide group of the TMOS1 by a hydroxyl group with
the production of methanol.
≡ Si−O− CH3 + H2O → ≡ Si−O− H + CH3OH (2.1)
The second reaction consists in a polycondensation between either two hydrol-
ysed TMOS molecules with the production of water, or alternatively between one
hydrolysed and one non-hydrolysed molecule, with the production of methanol.
≡ Si−OH + HO− Si ≡ → ≡ Si−O− Si ≡ + H2O (2.2)
≡ Si−O− CH3 + HO− Si ≡ → ≡ Si−O− Si ≡ + CH3OH (2.3)
The gelation speed of the solution depends on both hydrolysis and condensa-
tion. In acidic conditions, the rate determining step is condensation, hydrolysis
being quite fast. In basic conditions, the opposite is true. As a result, the mor-
phology of the gel produced in each condition is quite different. Chain formation
is favoured in acidic conditions and the gel obtained is softer and its porosity
tends to collapse upon drying [105], whereas denser spherical particles will rather
be formed at high pH.
To obtain a fast gelation inside the emulsion droplets, the TMOS can be pre-
hydrolysed in acidic conditions, and then, though the addition of a base, fast
polycondensation of the gel can be induced [117]. Gels formed in such conditions
have a larger SSA and a stiffer structure that prevents collapse of the porosity.
1The triple bond symbol in (2.1) represents the three remaining methoxide groups.
15
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
TMOS is not readily water soluble, which presents a problem for the pre-
hydrolysation step. Methanol is therefore used as a solvent.
In the conditions used, the gelation time is around 10 to 15 minutes.
2.2.4 Materials and methods
Based on the formulation established by F. Krauss.
Aqueous suspensions of polyvinyl alcohol (PVA) coated iron oxide nanoparti-
cles containing 10.75 mg iron/ml and 2 mg PVA/ml were prepared as described
previously [15]. Octane fraction (purum), tetramethyl orthosilicate (TMOS, p.a.),
methanol (puriss.), hydrochloric acid (HCl, p.a.), sodium-bis(2 ethylhexyl) sul-
fosuccinate (AOT, 98 %) and tetramethyl ammonium hydroxide (puriss) were
purchased from Fluka.
4 ml of methanol, 545 µl of 0.2 M HCl, and 680 µl of TMOS were mixed
together for 10 minutes under continuous stirring to pre-hydrolyze the TMOS
(molar ratio TMOS:methanol:H20=1:22:6). 2.5 ml of this pre-hydrolysed TMOS
and 2.5 ml of PVA coated iron oxide dispersion were added to 514 ml of a 0.0054 M
solution of AOT in octane fraction in a thermostatic reactor (20 ◦C) and emulsified
vigorously using a high shear rotor-stator homogenizer (Polytron PT DA 3030,
see figure 2.3) for 30 minutes at 16000 rpm. 250 µl of tetramethyl ammonium
hydroxide at 2.5% v/v in DI water were finally added into the reactor and the
mixture was emulsified for 5 more minutes. The emulsion droplets were left to
harden into solid particles for 30 minutes and were then sedimented magnetically
using a permanent FeNdB magnet (Maurer Magnets, 5x5x2.5 cm, BR 0.5 T),
the AOT-octane supernatant was discarded, and the particles re-suspended in
dichloromethane (DCM). They were washed three times by successive magnetic
sedimentation, discarding of the supernatant and re-suspension in fresh DCM.
Dynamic light scattering was on a Malvern ®mastersizer with a 633 nm He-Ne
laser on emulsions without addition of base.
16
2.2 Gel emulsion synthesis
2.2.5 Results and discussion
2.2.5.1 Scale-up of the emulsion
Three strategies were pursued, sometimes in combination, to increase the amount
of particles produced:
1. to increase the volume of the reactor whilst keeping the same emulsion
composition
2. to increasing the water-to-oil ratio
3. to set up a continuous emulsification system
Increasing the volume of the reactor has been attempted by F. Krauss [51]
using the ultrasound horn without much success as the power of the device is
not sufficient to homogenize volumes larger than 5 ml. To address this issue, a
more powerful emulsification setup was adopted, namely a high shear rotor-stator
homogeniser [71].
It consists in a turbine rotating in a toothed stator, resulting in a turbulent
flow. This system allows the emulsification of larger volumes (up to 1 l) with
comparable power per volume unit as the ultrasound horn. This system was used
for all emulsification experiments.
The second strategy consisting in increasing the volume of the dispersed phase
is not successful, as the emulsion is destroyed before the particles harden. Good
results are however obtained when emulsions with up to 4% dispersed phase
are diluted immediately after the emulsification step by addition of AOT-octane
solution at the CMC. The composition of this diluted emulsion can then be set
to be equal to the original formulation (around 1%) and the colloidal stability is
preserved sufficiently for the droplets to harden.
A continuous emulsification method was not achieved because of the difficulty
of matching the average residence time in the reactor with the hardening time of
the droplets. The emulsification process by turbine requires at least a few minutes,
and the gelation time should not be inferior to the emulsification time. In such
a case, the hardening gel starts sticking to the reactor’s walls and to the turbine
shaft and does not form particles. This is a major difference between ultrasound
17
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.3: High shear rotor stator homogeniser. The reactor is connected
to a thermostatic circulator.
18
2.2 Gel emulsion synthesis
treatment and high shear turbine. In the former case, the emulsification time
can exceed the gelation time without catastrophic consequence on the product as
the hardening droplets remain in suspension. This caused some difficulty when
the change was made between the two methods. Continuous emulsification could
be attempted with a high pressure homogeniser (HPH), which would yield more
reproducible results and could be set up in line.
2.2.5.2 Kinetic evolution
In order to understand the phenomena involved in the colloidal stability of the
emulsion, the kinetic evolution of the emulsion was investigated.
Depending on the synthesis parameters (AOT concentration, rotation speed of
the turbine, or water/oil ratio) the destruction of the emulsion can occur within a
few minutes, before complete hardening of the droplets. To monitor the evolution
of the droplet size, dynamic light scattering was implemented on an emulsion in
which no base was added, so as to prevent the gelation of the droplets.
Figure 2.4: Kinetic evolution the emulsion by dynamic light scattering.
The main mechanism at play is the flocculation and subsequent sedimentation of
the droplets.
The difficulty of the DLS method in the case of emulsions, is that the con-
centration of droplets is a very influent parameter of the kinetic behaviour of
the system. It is however also an important parameter for the light scattering
19
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
experiment, as the scattered intensity is directly dependant on it and has to be
sufficient for the measurement.
On figure 2.4, the main peak corresponding to single droplets is well visible
around 500 nm. The top of the peak hardly shifts from this position during the
time of the experiment, but it becomes broader towards the larger sizes, and new
broad peaks appear in the tens of µm range. These results are a clear indication
that the single droplets do not undergo significant Oswald ripening in the first four
hours after the end of emulsification, but are very prone to flocculation, followed
necessarily by sedimentation. The resistance of the droplets against coalescence
once they are agglomerated in flocks is low, and the emulsion is quickly destroyed.
Figure 2.5: Kinetic evolution of the emulsion by dynamic light scatter-
ing. Single droplet evolution.
Colloidal stability in inverse emulsions is difficult to achieve. The absence of
electrostatic charges to act as repulsive forces leave only steric hindrance as a
possible mechanism for colloidal stabilisation. In this case two steps can be taken
to improve the colloidal stability of the droplets. The water to oil ratio must
be kept low (in our case ≤ 1 % volume) and the surfactant concentration must
be kept to a minimum. At surfactant concentrations above the critical micelle
concentration, micelles are present, and cause an additional attractive depletion
interaction between the droplets [70].
20
2.2 Gel emulsion synthesis
2.2.5.3 Optimisation of emulsion parameters
The influence of emulsification parameters were studied in an experiment plan
including three parameters on two levels with two repetitions. The composition
of the emulsion was kept constant as described in section 2.2.4, but on a total
emulsion volume of 30 ml.
Factor Definition Levels
(-1) (+1)
A Temperature [ ◦C] 15 20
B Time [min] 10 20
C Rotation speed [rpm] 12000 16000
Table 2.1: Structure of the emulsion experiment plan.
Treatment Experiment n◦ Temperature [ ◦C] Time [min.] Rot. speed [rpm]
(1) 1 15 10 12000
A 2 20 10 12000
B 3 15 20 12000
AB 4 20 20 12000
C 5 15 10 16000
AC 6 20 10 16000
BC 7 15 20 16000
ABC 8 20 20 16000
Table 2.2: List of emulsification experiments.
The particles’ mean volume diameter was measured by centrifuge particle sizer
(see section 2.3.1 for more detail) after hardening of the particles and redispersion
in water. This represents a total of 16 different experiments, realised in random
order so as to rule out block or sequence effects.
The result of the experiments can be seen on figure 2.6.
The reproducibility of the experiments is poor due to the inherent thermo-
dynamic instability of the system. As can be seen on figure 2.7, the size of the
droplets depends primarily on the emulsification conditions. The rotation speed,
21
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.6: Experimental values and values predicted by the statistical
model (Adjusted values) for all the different treatments.
Figure 2.7: Effect of the factors on the particles’ mean volume diameter.
These three effects account for 83.4 % of the experimental variation.
22
2.2 Gel emulsion synthesis
which determines the shear rate, is the most important factor influencing the frac-
tionation process. Slight differences in the positioning of the turbine can affect
the state of the vortex created, or modify the circulation of the emulsion inside
the reactor, thus inducing variance in the droplet sizes.
Efforts in methodology were made during this study to understand and elimi-
nate the sources of variance, but although progress was made to an extent allowing
the conduct of a pertinent experiment plan, it still remains too large to deduce
physical models. Only three factors were retained as significant (time B, rotation
speed C and the interaction ABC), which alone account for 83.4 % of the total
experimental variability, the remaining being considered as experimental error.
This rather important experimental error is, as explained above, quite credible
for such a system and this interpretation of the results seems therefore safe and
consistent.
We can see that higher shear rates produce smaller droplets in the range
investigated, which perhaps leaves the possibility of further increasing the rotation
speed (the maximum is 24000 rpm) to diminish the mean diameter.
Emulsification time has a significant influence on droplet diameter in the range
measured, and even at 12000 rpm, the system does not reach a steady state after
only ten minutes of treatment. This means that destructive processes (coales-
cence, Ostwald ripening) which lead to a coarsening of the emulsion are not fast
enough to entirely counteract the effect of the mechanical energy brought to the
system, and the droplet size continues to decrease after this time.
The span was used as a measure of the width of the particle size distribution.
It is defined as follows:
Span =DV 90 −DV 10
DV 50
(2.4)
The effect of the parameters on the span of the particles size distribution is
small with respect to experimental error, as only one parameter, the temperature,
has a statistically significant effect. A 5 ◦C temperature increase reduces the span
by approximately 7 % the width of the distribution. This is probably due to a
faster gelation of the TMOS after the end of the emulsification step. Optimal
23
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
conditions for the production of small particles with a narrow size distribution
would therefore be those of experiment 8 (table 2.2).
24
2.3 Materials characterisation
2.3 Materials characterisation
2.3.1 Methods
Specific surface area measurements and porosimetry were realized by nitrogen ad-
sorption, using a Micromeritics ASAP 2010 machine at 77K. The specific surface
areas were calculated by the Brunauer-Emmet-Teller (BET) method, and the pore
diameter distributions and total pore volume were derived from the adsorption
branch by the Barrett-Joyner-Halenda (BJH) method. Density measurements
were done on a Micromeritics Acupyc 1330 helium pycnometer.
The particles were dried and degassed for 24 h at 200 ◦C in vacuum prior to
the measurements. Scanning electron microscopy (SEM) images were acquired
on a FEI XLF30-FEG. Samples were prepared by drying a drop of the suspension
on a polished titanium plate.
Particle size measurements were done on a disc photocentrifuge particle sizer,
CPS Model DC24000, using 470 nm light in a gradient of 8% to 24% sucrose
solution. The particles were re-suspended in DI water. The results obtained were
compared to the individual measurement of 536 particles from SEM pictures using
ImageJ image analysis program.
2.3.2 Results and discussion
The morphology of the particles is homogeneously spherical (figure 2.8) and their
size is in the sub-micron range.
Figure 2.8: SEM micrographs of composite silica-SPION particles.
25
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Their surface is rough, due to the presence of SPIONS and open porosity. The
particles are neither coalesced nor agglomerated, which shows that the colloidal
stability of the emulsion is good in the time range needed for the hardening of
the particles.
Light centrifuge techniques require a precise measurement of the hydrody-
namic density of the material, as well as a correct estimate of its refractive index.
The hydrodynamic density is determined mainly in our case by the pycnometric
density of the material and the total pore volume. It was calculated by combin-
ing helium pycnometry with nitrogen adsorption porosimetry, assuming that the
pore volume was fully filled with the dispersion medium (water).
To calculate the refractive index of the composite material constituting our
particles, the refractive index of each component, silica, iron oxide, PVA and the
water in the pores was multiplied by its volume fraction in the particles and added
up to yield a composite refractive index of 1.55.
The CPS (figure 2.9) shows a normal size distribution with a mean volume
diameter of 479 nm and a standard deviation of 106 nm, which corresponds to a
dispersion of 22%.
These results concur with SEM images, indicating that the method is valid,
and that the hydrodynamic density was measured accurately. The size distribu-
tion is narrow compared to other systems using metastable morphology templates,
such as aerosol assisted synthesis and other mini-emulsions systems [2, 90]. The
helium pycnometer density value of 2.28 g/cm3 (table 2.3) is in good accordance
with what can be calculated on the basis of the composition of the particles. The
hydrodynamic density was calculated to be 1.77 g/cm3 with a measured pore
volume of 0.291 cm3/g.
Mesoporous silica particles obtained by the Stober method or microemulsion
synthesis can be virtually monodisperse and present homogeneous, tunable pore
diameters [45, 120]. Our particles have a broader size distribution but can contain
up to 20 %wt of functional nanoparticles, thus enhancing the desired functional-
ity whilst retaining the porous properties of the silica. Such a high nanoparticle
content cannot to our knowledge be obtained with the abovementioned meth-
ods, which are more adapted for producing well characterized core shell type
nanoparticles-silica composites with low nanoparticle loads.
26
2.3 Materials characterisation
Figure 2.9: Particle size distribution of composite silica-SPION parti-
cles. Volume distribution measured by centrifuge particle sizer (CPS).
Specific surface area (by BET): 720 m2/g
Average pore diameter (4V/A by BET): 2.6 nm
Total pore volume 0.291 cm3/g
Porosity 0.4
He Pycnometer density 2.28 g/cm3
Hydrodynamic density 1.74 g/cm3
SiO2 content 86.6 %wt
SPION content 11.1 %wt
PVA content 2.3 %wt
Table 2.3: Typical material characteristics.
27
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
The particles have a high specific surface area (SSA) of 720 m2/g, which is close
to values reported in literature for mesoporous silica prepared from pre-hydrolyzed
TMOS in basic conditions [105], bearing in mind that they are made up of only 86
wt% silica. The BJH adsorption-desorption isotherm (figure 2.10) has shape of a
Brunnauer’s type I/type II combination, with a steep rise a low partial pressure,
due to the filling of micropores, a second regime corresponding to the filling
of small mesoporosity, and a plateau at higher partial pressure (P/P0 = 0.45)
[91, 116, 120].
Figure 2.10: Nitrogen adsorption and desorption isotherms. Before and
after paclitaxel adsorption (see section 2.4).
It shows another steep rise around P/P0 = 1 due to capillary condensation in
the inter-particle spaces [2, 66]. The nitrogen volume adsorbed due to this effect
was not considered in the calculation of the total pore volume. The isotherm also
shows no hysteresis, which indicates the absence of an ink bottle effect : all the
pores have access to the surface of the particle either directly, or through a larger
pore [79]. The pore diameter distribution (figure 2.11) shows that 87 % of the
total pore volume is constituted of pores smaller than 5 nm in diameter, narrowly
distributed around an average pore diameter of 2.62 nm, which is realistic for
TMOS based gels condensed in basic conditions [105].
28
2.3 Materials characterisation
Figure 2.11: Incremental pore diameter distribution in volume. Before
and after paclitaxel adsorption (see section 2.4).
29
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
2.4 Mesoporous silica spheres as drug delivery
vehicles
2.4.1 Introduction
The applicability of MS multifunctional particles for drug delivery applications
was assessed using the anti-cancer drug paclitaxel as a model compound.
Paclitaxel (or Taxol) is an anti-tumor agent that specifically binds to the
microtubules of cells during mitosis, thus blocking cellular division. Because
cancer cells divide more frequently than healthy cells, the tumors are affected in
priority. Paclitaxel has attracted much interest for the treatment of ovarian and
breast cancers. The dimensions of the molecules (see figure 2.12) compared to
our porosity, as well as the expertise developed at LTP by K. Dittmar [27] for
the handling and measurement of paclitaxel solutions by HPLC, made it a good
candidate for this study. It is however a very dangerous substance because of
its strong antimitotic properties at very low concentrations, and requires to be
handled with care.
Figure 2.12: Paclitaxel molecule.
The loading capacity of the drug in the particles was measured, as well as the
release kinetics in water. Finally, simulations of the molecule in silica micropores
were run using molecular dynamics.
30
2.4 Mesoporous silica spheres as drug delivery vehicles
2.4.2 Materials and methods
2.4.2.1 Paclitaxel loading
Known masses of MS-SPION particles were suspended in different concentrations
of paclitaxel in dichloromethane (DCM). Corresponding reference solutions with-
out particles were prepared at the same time. The paclitaxel was left to adsorb
on the particles for 48h at 4 ◦C in the dark. The particles were then magnetically
sedimented, the supernatant was pipetted out and its paclitaxel concentration
was measured by high performance liquid chromatography (HPLC). The concen-
tration in the references was measured at the same time. The amount of paclitaxel
adsorbed was calculated on the basis of the concentration difference between the
samples and the references solutions. The experiment was done in triplicates.
Sample preparation for nitrogen adsorption measurement on paclitaxel loaded
samples was adapted from Hata et al. [35].
2.4.2.2 Paclitaxel release
3.4 mg of MS-SPION particles were suspended for 48h in 10 ml of 100 µg/ml solu-
tion of paclitaxel in DCM. They were then magnetically sedimented, supernatant
was discarded, and the particles were left 1 minute in air for the remaining DCM
to dry. The particles were re-suspended in 10 ml DI water under stirring. At
different times, the particles were magnetically sedimented, and the supernatant
was entirely sampled out for measurement and replaced by 10 ml of fresh DI
water. At all times, the supernatant concentration of paclitaxel was kept well
below the saturation concentration. The paclitaxel concentration in the samples
was measured by HPLC.
2.4.2.3 Instrumentation and measurements
The HPLC system consisted of an Alliance 2690 pump module (Waters) con-
nected to the analytical column ODS-3 (Inertsil ®, particle size 5 µm; 4.6 x 150
mm), which was protected with an ODS-3 conventional guard column (Inertsil®,
particle size 5 µm; 4.6 x 50 mm). The analysis procedure applied in this study
was adapted from literature [4, 110]. The mobile phase consisted of UP-water
31
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
and acetonitrile in a gradient elution. The flow rate was set to 1 ml/min. Detec-
tion was performed at 227 nm with a Photodiode Array Detector 996 (Waters)
with connection to the Millenium 3.05.01 software data station (Waters). 100
µl of each sample were injected for analysis via autosampler. The absorption of
paclitaxel as a function of retention time was recorded in the chromatograms.
2.4.2.4 Simulation
All simulations were carried out with Dr. Ulrich Aschauer using the Discover
module within the Material Studio [1] modelling software package using the COM-
PASS forcefield. Amorphous SiO2 glass structures provided with this package
have been used for surface calculations and as the starting point for pores. For
pore generation, all atoms within a given radius of the center of the simulation
box were removed, while retaining the stoichiometry of the simulation box. After
the nano-pore atoms were fixed to retain periodicity, the molecule was placed in
the pore and DCM solvent molecules were allowed to diffuse into the pore during
an NPT (constant pressure) simulation from a reservoir until equilibrium was
reached [65]. The reservoir was then disconnected, all atoms allowed to move in
subsequent NVT (constant volume) simulations.
The simulation timestep was chosen as 1 ps and Nose-Hoover and Andersen
thermo- and barostats were applied as needed. Information about the total simu-
lation time are given along with the results. Adsorption enthalpies were calculated
using (2.5), where the letters S, A and L denote the substrate, the adsorbent and
the liquid respectively.
∆Hads = HSAL − (HSL +HAL −HL) (2.5)
This equation compares the state (SAL) where the paclitaxel is adsorbed to
the substrate (surface or pore), with the states of just the substrate (SL) or the
paclitaxel (AL) in contact with the solvent. The self interaction energy within
the solvent (L) has to be subtracted due to double counting in both SL and AL
terms.
Diffusion coefficients were calculated from the mean square displacement (MSD)
as given by equation 2.6 and Einstein’s equation as given by (2.7).
32
2.4 Mesoporous silica spheres as drug delivery vehicles
MSD(t) = 〈|~r(t = 0)− ~r(t)|2〉 (2.6)
D = limt→∞
MSD(t)
Mt(2.7)
Where the vectors ~r are the atom positions at time t and the average being
taken over all atoms of the diffusing molecule and M is a dimensionality number,
which is 6 for a 3D situation and 2 for a 1D situation.
2.4.3 Results and discussion
2.4.3.1 Paclitaxel adsorption
The amount of paclitaxel adsorbed on the particles increases with the supernatant
concentration until a saturation plateau is reached (figure 2.13).
Figure 2.13: Adsorption isotherm of paclitaxel on silica-SPION parti-
cles. The data is fitted with Langmuir’s adsorption isotherm.
Nitrogen adsorption porosimetry (figure 2.11) realized on the particles after
saturation with paclitaxel shows a decrease in specific pore volume of 0.19 cm3/g,
principally in the pores with diameters smaller than 5 nm. The average pore
diameter is decreased from 2.62 nm to 2.3 nm, indicating that the paclitaxel is
33
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
a 8.03 · 10−5 mol/g
b 2.63 · 104 l/mol
Table 2.4: Fitting parameters for Langmuir’s isotherm.
indeed adsorbed as a dense layer inside the larger micropores. The decrease in
pore volume is too large to be entirely accounted for by the presence of paclitaxel.
The total volume of paclitaxel adsorbed can be calculated to be 0.091 cm3/g using
using the value calculated by Huynh et al. for the density of paclitaxel [39]. Alter-
natively, is can be estimated geometrically from the decrease in the average pore
diameter to be 0.15 cm3/g. Between these two estimates, the adsorbed paclitaxel
volume represents only between 45 and 60 % of the total pore volume decrease,
which can be explained by assuming that the molecules partially or completely
block the opening of pores, leaving large volumes inaccessible to the adsorbing
nitrogen. This is supported by the fact that the nitrogen adsorption-desorption
isotherm on paclitaxel-filled particles presents an hysteresis which was not seen for
the pristine particles: the partial obstruction of pore openings by the molecules
results in an ink bottle structure and could account for this difference. Park et
al. [80] measured the adsorption isotherm of paclitaxel on mesoporous silica in a
HPLC column and found Langmuir’s adsorption isotherm to be a suitable model.
The paclitaxel adsorption data was fitted with a Langmuir isotherm of the form:
Q =a · b · C1 + b · C
(2.8)
with Q the quantity adsorbed, a the quantity adsorbed at saturation, b the
gradient of the curve at the origin and C the equilibrium concentration of pacli-
taxel in the supernatant. Our data points are also well described by this model.
Table 2.4 shows the values of the fitting parameters obtained.
The fit is satisfactory, with a goodness of fit close to 1 and less than 5%
standard error for the quantity adsorbed at saturation, and 18% for the slope at
the origin. The differential free adsorption energy can be calculated from b the
slope of Langmuir’s isotherm at the origin using the following relation [63]:
∆GΘads = −RT ln(b · C0) (2.9)
34
2.4 Mesoporous silica spheres as drug delivery vehicles
Structure ∆Hads[kJ/mol]
Surface -779
2.0 nm pore diameter -2633
1.5 nm pore diameter -460
Table 2.5: Calculated differential adsorption enthalpies.
Where R is the universal gas constant, T is the absolute temperature and C0
is the standard concentration defined as 1 mol/l.
The ∆GΘads obtained from our measurements is -23.5 kJ/mol ± 0.4. This is
a reasonable order of magnitude for a free differential adsorption enthalpy, and
indicates that the adsorption of paclitaxel on silica is spontaneous and stable. For
comparison, in a different yet similar system, the value that can be derived from
Langmuir’s isotherm measured by Park et al. [80] for paclitaxel on mesoporous
silica with an average pore diameter of 6 nm in hexane-methanol mixture is equal
to -9.31 kJ/mol. The systems and the methods are different, but it is interesting
to see that both values lie in the same order of magnitude.
2.4.3.2 Simulation
Adsorption of paclitaxel was calculated onto a flat surface and in pores approx-
imately 1.5 nm and 2 nm in diameter, larger pores not being realisable with
available computer resources at the present time.
All structures were allowed to evolve freely for 1ns and the average energy to
be used in the adsorption energy calculations was evaluated over the last 100 ps.
The obtained adsorption enthalpies are given in table 2.5.
As simulated values are very large compared to those measured experimentally
direct comparision is impossible and only relative values should be considered.
The reasons for the discrepancy between simulation and experiment are com-
plex and not fully explained, but several factors are worth noting. The simulation
assumes an ideal situation, without impurities or contamination, simple pore and
surface geometries and computes a finite number of conformations which are not
necessarily statistically representative. The real pore morphology and topology of
the material is different and cannot be considered as an interconnected network
35
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Case D[m2/s]
Pore total 7.4 · 10−11
Pore in axis 5.0 · 10−11
Free total 4.9 · 10−10
Table 2.6: Diffusion coefficients for different cases. In axis: M=2, others:
M=6 (see (2.7).
of perfect cylindrical pore. It was found by Coasne et al. [20] that the simulated
adsorption isotherm of molecular fluids in porous media differed substantially
when using simple geometry pores models versus realistic pore models.
Another important point is that standard molecular dynamics simulations
do not take into account entropic effects. The result is given as an interaction
enthalpy, and not as a free energy, as is the case for the results calculated from
Langmuir’s adsorption isotherm, although this is not enough in itself to account
for such a difference.
The results obtained by simulation, tough not directly applicable quantita-
tively, do however show the general tendencies that can be expected for this
system. Adsorption is more favourable in the 2.0 nm pore than on a flat surface.
Santhong et al. [92] found in their molecular dynamics simulations that the ad-
sorption enthalpy of hexane in mesoporous silica increased with decreasing pore
diameter.
The adsorption of paclitaxel is however less favourable in the 1.5 nm pore than
in the 2 nm pore. Hata et al. [35] found that paclitaxel was not adsorbed inside
silica pores smaller than 1.6 nm. This is most likely due to molecular distortions
as a result of confinement, the size of the paclitaxel molecule being approximately
1.8 nm.
Diffusion simulations in the pure solvent and the solvent filled pore were run
for 1 nanosecond. The diffusion coefficients (table 2.6) for the different cases
result from a linear fit of the mean squared displacement curves as a function of
time.
The free diffusion coefficient calculated by simulation can be compared with
the one that can be obtained from the Stokes-Einstein equation [24].
36
2.4 Mesoporous silica spheres as drug delivery vehicles
Dfree =kBT
6πµrm(2.10)
Where kB is the Bolzmann constant, µ the dynamic viscosity of the solvent
and rm the radius of the solute. The Dfree obtained for paclitaxel (rm = 8·10−10m)
in DCM (µ = 4.1 · 10−4 Pas) is 6.3 · 10−10 m2/s, which is consistent with the
Dfree of 4.9 · 10−10 m2/s calculated in our simulation.
Veith et al. [108] have studied the diffusion of small aroma molecules inside
mesoporous silica. Their phenomenological model can be applied to evaluate our
simulation results. The effective pore diffusion coefficient Dpore can be related to
the free diffusion coefficient using the following relation [61]:
Dpore
Dfree
=εpτpF (λ) (2.11)
Where εp is the porosity of the particle, τp is the tortuosity of the pore network,
F (λ) is a restrictive factor accounting for steric hindrance and λ is the ratio
between the molecule radius and the pore radius. The medium through which
diffusion takes place in our simulation is a single infinite cylindrical pore. The
porosity as well as the tortuosity can therefore both be considered as equal to 1,
and so only the steric hindrance factor F (λ) is taken into account. It is in our
case for mesoporous silica [47]:
F (λ) = (1− 1.83λ+ 4.18λ2) exp(−6.52λ) (2.12)
With λ = 0.6, Dpore in this case is equal to 1.77 · 10−11 m2/s, which also
compares well with the simulated value of 5 · 10−11 m2/s.
2.4.3.3 Paclitaxel release
The release profile in water (figure 2.14) shows a burst release in the first few
minutes which decreases to zero after approximately 100 minutes.
The experiment was carried on for 72 h and no further release could be de-
tected during that time. The total paclitaxel amount released corresponds to
36% of the total quantity originally adsorbed, indicating that the majority of the
37
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.14: Paclitaxel release profile in water.
substance in not eluted and remains inside the pores. The eluted paclitaxel prob-
ably comes from the largest pores and the outer surface of the particles. This is
consistent with elution results in water obtained by Hata et al. [35].
The release rate determining step could be determined by the diffusion rate out
of the porous structure, the dissolution kinetics of paclitaxel in water, the speed
of liquid penetration inside the pores, the desorption kinetics of paclitaxel from
the silica surface or finally the existence of a chemical driving force counteracting
diffusion. Using the following empirical relations established for mesoporous silica
[103]:
τp = εp + 1.5(1− εp) (2.13)
With εp = 0.4, we can calculate from (2.11) Dpore to be 5.45 · 10−12 m2/s.
The average time thus needed for a paclitaxel molecule to travel the distance from
the center of a MS-SPION particle to its surface (240 nm) can be calculated to
be of the order of one hundredth of a second. The release experiment having run
for over 72h, diffusion limitations cannot explain this retention effect.
The dissolution kinetics of paclitaxel in water limits the burst release rate as
is the case in most paclitaxel eluting devices in water [21, 46, 86]. This initial
burst release is caused by the dissolution of the drug not directly adsorbed on
38
2.4 Mesoporous silica spheres as drug delivery vehicles
the silica, but as multilayers either in larger pores or at the surface. It does not
however explain why most of the drug remains inside the particles.
The desorption kinetics of the drug on the silica could be a rate limiting factor,
but our diffusion simulations inside the pore do not show any stable adsorption
but diffusion limited transport (figure 2.15).
The substantial difference in adsorption enthalpy between pores and the flat
surface, and between the flat surface and the solution means that a molecule
exiting a small pore into a larger pore or into the solution goes through a strong
chemical potential gradient, and as a result experiences a force in the opposite
direction to the diffusion. The simulations only show a diffusion driven transport
mechanism because it is conducted in an infinite cylindrical pore. There is no
widening of the pore radius and thus no influence of the chemical potential on
the transport of the molecule. –
Rough calculations of this effect in DCM show that the majority of paclitaxel
is retained inside the pores. Assuming cylindrical pores, the chemical potential
profile inside the particles is constant and constitutes a potential well with respect
to that of free paclitaxel in solution. The chemical potential increases steeply at
the opening of the pore. As a first approximation, the length scale over which
this occurs is in the order of magnitude of the pore diameter, as illustrated in
figure 2.16. The force on the molecule is thus:
F =∆Gads
dA(2.14)
Where ∆Gads is the differential free enthalpy of adsorption obtained from
Langmuir’s isotherm, d is the average pore diameter and A is Avogadro’s constant.
Using the Nernst-Einstein equation we can calculate the flux JF due to this force
[83]:〈v〉Dpore
=F
kT(2.15)
JF = 〈v〉c(x) (2.16)
Where 〈v〉 is the time-averaged speed of the paclitaxel molecules and c(x) is the
concentration.
39
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.15: Configuration of paclitaxel in a 2 nm diameter pore at
various equilibration times.
40
2.4 Mesoporous silica spheres as drug delivery vehicles
Figure 2.16: Schematic representation of the chemical potential gradient
at the pore opening.
We can compare JF to the diffusion flux Jdiff :
Jdiff = −Dpore ·dc
dx(2.17)
When a steady state is reached, the two fluxes equilibrate and the total net
flux is zero. In that case, the following is true:
Dpore ·dc
dx= 〈v〉c(x) (2.18)
The solution to this differential equation is a Bolzmann distribution of the
form:
c(x) = const. exp−Φ(x)/kT (2.19)
Where Φ(x) is the chemical potential profile, taken as linear in this case. We can
thus plot the concentration profile inside the pore.
The y-intercept on figure 2.17 is the minimal concentration in the solution
for which a steady state is reached. This equilibrium solution concentration is
determined by :
cs = cpe∆GΘ
adsRT (2.20)
Where cs is the equilibrium concentration in the solution, cp is the equilib-
rium concentration in the pores. In the case of release in water, given the poor
solubility of paclitaxel, ∆GΘads is larger than in DCM, and cs is probably under
the detection limit of HPLC which explains why no release can be detected after
41
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.17: Calculated paclitaxel concentration profile inside the pores.
x = 0 represents the pore opening
the burst release. With this argument, and knowing our HPLC’s detection limit
for paclitaxel in water, we can estimate using (2.20) a minimum ∆GΘads in water:
with ∆GΘads < −29 kJ/mol, any paclitaxel release is undetectable with our setup.
This value seems reasonable compared to the one measured in DCM.
The paclitaxel is thus released from the particles, but comes to equilibrium
with the solution at low concentrations (cs < 2.0 · 10−7 mol/l) and then remains
trapped inside the particles.
Finally, the penetration speed of water into the pore could be a limiting factor,
and can be calculated using the following relation [34]:
v =rpγLV cos Θc
4η· 1
l(2.21)
With v the meniscus speed, rp the pore diameter, γLV (0.0728 N/m) the sur-
face tension of water, and Θc (70◦) [27] is the contact angle between paclitaxel
and water and l is the distance between the meniscus and the opening of the
pore. The meniscus speed is low in such small pores, especially as the surface
is covered with a hydrophobic drug. It decreases to 7.6 · 10−2 m/s near the
center of the particle. The particles are however so small that the time needed
42
2.4 Mesoporous silica spheres as drug delivery vehicles
for water penetration is negligible. Equation (2.21) is however valid for pores
as narrow as 27 nm in diameter, and it is quite possible that the penetration of
water in smaller pores could be hindered not merely by a kinetic problem, but
be thermodynamically unfavourable under a certain diameter. In this case, our
main porosity could be too small for water to penetrate, especially when coated
with an hydrophobic drug, which would explain the retention of the drug in the
particles, as approximately 65 % of the drug is situated inside these pores.
2.4.4 Conclusion
Our particles have successfully been loaded with the anti-cancer drug paclitaxel
and their loading capacity was measured. An credible estimate for the free dif-
ferential adsorption enthalpy of paclitaxel on our mesoporous silica particles was
obtained from the adsorption isotherm. The drug release properties were found
to be extremely slow after an initial burst release, which could either be caused
by a high affinity between the paclitaxel and the silica, or by the inability for
water to enter the micropores. Such a slow release is a concern for drug delivery
applications, but could be an asset for purification applications using magneti-
cally separable adsorbent particles. The pore size distribution of the silica matrix
could be optimised to improve and control the release kinetics by pore templating
using surfactants.
43
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
2.5 Chromium doped alumina
2.5.1 Introduction
IR radiations penetrate living tissues much better than UV or visible light, and
are therefore indicated for use in medical imaging. The necessity of finding an
inorganic fluorescent marker emitting in the IR has been partially addressed by
the discovery of heavy metal quantum dots. These are however inherently toxic
and little is known about their long term effects on health [13]. In order to address
that issue, chromium doped alumina, also known as ruby, in the form of colloidal
nanoparticles was investigated. The main advantage of this material is its low
toxicity, being based on a biologically inert material and containing heavy metal
only in low concentrations (< 3% w/w).
The fluorescence response and the particle size distribution of the materials
were characterised and optimised. Conventional and confocal fluorescence mi-
croscopy were performed to assess the quality of the signal to noise ratio and
contrast that could be obtained for bioimaging applications. Finally, the par-
ticles were used in the gel-emulsion synthesis in order to combine them to MS
particles, alone or in the presence of SPIONS.
2.5.2 Materials and methods
HT-100 boehmite powder (Condea, Germany) was suspended as a 13% w/w
suspension into a 0.014 M solution of chromium nitrate (Merck, p.a.) in 2 % w/w
acetic acid (AA) to yield in the end product a chromium concentration (Cr2O3 to
Al2O3) of 0.7 % w/w. This suspension was then freeze dried −50 ◦C @ 0.08-0.1
mbar, Alpha 1-4, Christ, Germany).
Thermal treatments were done on the freeze-dried powder in a Naber 1400
oven (Nabertherm, Germany).
The annealed powder was crushed in a mortar before attrition milling to break
large agglomerates and resuspended at 30% w/w powder in 2% AA.
Attrition milling was performed on a 30%w/w suspension of the annealed
powder in an attrition mill with 400 g of 1.25 mm diameter 3YSZ grinding spheres
(Tosoh, Japan) at 1500 rpm for 3 hours.
44
2.5 Chromium doped alumina
The attrited suspension was retrieved by 10 fold dilution in 2% AA. To remove
larger agglomerates, centrifugation was performed on this diluted suspension in a
GR 20 22 Jouan ultracentrifuge. The concentration of chromium doped alumina
particles is estimated to be (after dilution centrifugation) maximum 2% w/w.
Particle size distributions were acquired on a CPS 24000 (see 2.3.1) centrifuge
particles sizer at 11000 rpm using n=1.624 [99] for the refractive index and ρ=3.82
g/cm3 for the density.
X-ray diffraction was realised on a Philips X’pert diffractometer with a Cu
Kα radiation.
Fluorescence spectroscopy was done on a Tecan® Infinite M200 well plate
reader. The powders were packed into the 5 mm wells of polystyrene 96-well
plates. The fluorescence was measured from the bottom of the well.
Fluorescence microscopy was done on a ZEISS LSM 700 upright microscope
(objective plan-apochromat 20x/0.80 in air) using a 639 nm solid state excitation
laser and a 655 to 715 nm band pass filter. The fluorescent particle suspension
was dried on a glass sample holder and imaged directly.
Confocal fluorescence imaging was done on a LEICA SP2 AOBS inverted
confocal laser scanning microscope (objective HCX PL APO 63x/1.40-0.60 in
oil) with an excitation laser at 633 nm and a 662 to 737 nm band pass filter. A
drop of non-attrited suspension was placed between two glass plates, which were
sealed with nail polish.
2.5.3 Results and discussion
2.5.3.1 Materials characterisation
The starting material (HT-100, Condea, Germany) is a high surface area pseu-
doboehmite. Pseudoboehmite, or gelatinous boehmite, is a highly hydrated,
poorly crystallised form of “mineral” boehmite (AlOOH), a major constituent
of many naturally occuring bauxites [112].
It is supplied as a spray dried powder with 5 to 15 µm granules, but is easily
dispersed in 2 % acetic acid. It has a very small particle size, which requires it to
45
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
be handled with special safety precautions1, a narrow size distribution and few
agglomerates.
Table 2.7 shows some materials characteristics of the pristine powder and the
chromium doped alumina.
The BET equivalent spherical particle diameter dBET is calculated from the
specific surface area (SSA) of the powder and its pycnometric density (ρ) accord-
ing to:
dBET =6
SSA · ρ(2.22)
It gives an indication on the average size of the primary particles of a powder
irrespective of its level of agglomeration. The agglomeration factor FAG is given
as the ratio of the mean volume diameter dV to dBET.
There is no substantial primary particle growth during annealing, as can be
seen from the nitrogen adsorption data. The density of the material increases
substantially during the process, while the SSA decreases accordingly, due to the
thermal dehydroxilation of boehmite into alumina (Al2O3) starting at around
300 ◦C.
The powder becomes yellowish during annealing and undergoes severe ag-
glomeration, which creates the necessity of an attrition treatment. After attri-
tion milling, the agglomeration factor remains over three time greater than before
the treatment but the mean volume diameter is now low enough to envisage the
inclusion of the particles inside MS multifunctional beads. The particles size dis-
tributions of the pristine powder and of our material after annealing and attrition
are shown on figures 2.18 and 2.19.
The polymorph of alumina obtained from pseudoboehmite after calcination at
900 ◦C is δ-alumina, a transition alumina closely related to tetragonal γ-alumina
[112]. This is confirmed by X-ray diffraction analysis (figure 2.20), although
significant peak broadening is observed, caused by the small size of the crystallites
(see table 2.7) and poor long range order of the material which makes it difficult
to differentiate these two quite similar forms.
1Because of its small size, and its potential adverse health effects, all manipulations of the
dry powder were done in a specially dedicated underpressurised room wearing protective gear
and a breathing mask.
46
2.5 Chromium doped alumina
Figure 2.18: Cumulative particle size distribution in volume for pristine
HT-100 Condea powder. From the work of Staiger, Bowen et al. [99]. Different
PSD characterisation methods are compared: X-ray disc centrifuge (XRD), photon
correlation spectroscopy (PCS) and cuvette photocentrifuge (CAPA)
Figure 2.19: Cumulative and frequency particle size distribution in vol-
ume for annealed chromium doped alumina after attrition. The particles
are strongly agglomerated compared to the pristine powder.
47
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Properties Pristine powder [99] After annealing and attrition
SSA [m2/g] 214 156
ρ [g/cm3] 3.01 3.82
dBET [nm] 9.3 10
dV [nm] 17 58
σV [nm] 4 28
FAG 1.7 5.8
Table 2.7: Material properties of chromium doped alumina. Annealing at
900 ◦C.
Figure 2.20: X-ray diffraction spectrum of chromium doped alumina
annealed at 900 ◦C. The crystal structure is that of δ- or γ-alumina.
48
2.5 Chromium doped alumina
2.5.3.2 Characterisation of fluorescence
Figure 2.21: Excitation and emission spectra of chromium doped alu-
mina particles. A and C: Excitation and emission spectra for λex = 580 nm,
λem = 700 nm, integration time = 2000 µs; B and D: Excitation and emission
spectra for λex = 610 nm, λem = 800 nm, integration time = 40 µs.
At 2000 µs integration time our material shows an excitation peak at 580 nm
and an emission peak at 700 nm. When acquired with a shorter (40 µs) integration
time however, the excitation and the emission undergo a red-shift with peaks at
610 nm (red) and 800 nm (near IR) respectively (see figure 2.21).
This indicates that there are at least two different fluorescence phenomena,
emitting at different wavelengths with different characteristic decay times.
To investigate the time dependency of the fluorescence emission, a series of
emission spectra were acquired with a constant acquisition time of 40 µs, and a
variable lag time of 0 to 900 µs. The fluorescence intensity for two wavelengths
(700 nm and 800 nm) were plotted against the lag time, yielding fluorescence
decay curves for each phenomenon (see figure 2.22).
49
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
One can thus separate one broadband short lived emission with a peak around
810 nm, and one more defined peak at 700 nm, characterised by a longer fluores-
cence lifetime.
Figure 2.22: Emission intensity decay over time. For λ = 700 nm and
λ = 800 nm.
Bulk ruby (α-alumina) has two well known emission peaks at λ = 693.2 nm
and 694.5 nm [40]. These so-called R emission bands are caused by Cr3+ ions
substituted to Al3+ on octahedral sites.
At high Cr3+ concentration (> 1%), asymmetric peak broadening of these R-
lines is observed, and a broadband emission peak appears at λ = 750 nm caused
by pair and cluster interaction between chromium ions [41]. At room tempera-
ture however, and contrary to our observations, the fluorescence lifetime of these
phenomena are identical. This indicates that while chromium pair and cluster
interactions cannot be ruled out, they cannot explain the short lived fluorescence
phenomenon observed.
Patra et al. [82] found that in γ-alumina, the two discrete R-lines found in
α-alumina tend to convolute of into a single broader emission band ranging from
690 to 720 nm. They attribute this effect to the distortion of octahedral sites in
the γ-lattice.
50
2.5 Chromium doped alumina
Pillonet et al. have characterised the fluorescence of chromium doped δ- and
γ-alumina calcinated from boehmites or via a sol-gel route [84]. They also observe
a short lived (time constant in the tens of µs range) broadband IR radiation with
an excitation peak at 607 nm and an emission peak around 820 nm in addition
to the convoluted R-lines. They attribute this phenomenon to Cr3+ ions located
in weak crystal field sites, in the vicinity of cation vacancies.
The exact time constant related to each event cannot however adequately be
measured with our setup. The Tecan microplate reader is indeed more adapted to
the automatic measurement of multiple samples in solution than to the precise,
time-resolved characterisation of optical phenomena in solids.
2.5.3.3 Effect of chromium concentration and annealing time on flu-
orescence
The effects of chromium concentration and annealing time on the fluorescence
intensity of chromium doped alumina were investigated using a bi-factorial ap-
proach. The annealing temperature chosen was 900 ◦C. The levels of the factors
were chosen as shown on figure 2.23.
Figure 2.23: Bi-factorial plan for the fluorescence response of chromium
doped alumina. Influence of annealing time and chromium concentration.
Annealing time has a negative influence on the fluorescence signal, while high
chromium concentrations is beneficial.
51
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.24: Fluorescence response surface for chromium doped alu-
mina: influence of annealing time and chromium concentration. λem =
800 nm, λex = 615 nm
52
2.5 Chromium doped alumina
2.5.3.4 Fluorescence imaging
The powder could be imaged on a confocal laser scanning microscope, despite
the long time constant of the main emission peak. The signal was however very
weak even with a high gain and an integration time pushed to upper the limit (a
few microseconds/pixel). The effect was to increase the noise and the acquisition
time, given the large number of points needed to create an image. These problems
cannot be easily resolved and reduce the applicability of chromium doped alumina
for CLSM.
The long fluorescence lifetime is however not a disadvantage in the case of
classical widefield fluorescence microscopy, where both the excitation and the
acquisition occur in a continuous way. The signal in this case is strong, with a
good contrast and little noise. Figure 2.26 is a fluorescence microscopy image of
the drop of suspension after drying, reconstructed automatically by the machine
from an array of smaller images. This techniques makes it possible to images such
large objects.
Figure 2.25: Fluorescence microscopy images of chromium doped alu-
mina agglomerate.
53
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.26: Fluorescence microscopy images of chromium doped alu-
mina. Dried suspension drop
54
2.5 Chromium doped alumina
2.5.4 Application to multifunctional particles
The applicability of these fluorophores in multifunctional particles for biomedical
applications was assessed by combining them with SPIONS in a mesoporous silica
matrix.
2.5.4.1 Effect of SPIONS on fluorescence
In order to assess the functional compatibility between SPIONS and chromium
doped alumina nanoparticles (ruby), their optical interactions were measured in
suspension. SPIONS and ruby suspensions were mixed together in various pro-
portions in order to measure the loss of fluorescence caused by the SPIONS. The
SPION concentration was varied between zero and the concentration typically
found in our multifunctional particles, i.e. approximately 5 mg Fe/ml.
Figure 2.27: Effect of SPION concentration on the fluorescence emission
intensity of chromium doped alumina. Even at SPION concentrations equal
to those present in the particles, 70% of the fluorescence signal is still detected.
The relative intensity in figure 2.27 is given as the ratio of intensities between
the suspension with SPION and the one without SPIONS. At 5 mg Fe/ml, the
fluorescence signal decreases by approximately 30 % with respect to a SPION-free
suspension. The optical interactions between the two particles is thus relatively
benign, as most of the signal can still be detected. Furthermore, the decrease is
not linear, as SPIONS seem to absorb only a certain part of the spectrum (λ <
550 nm) but not all of it.
55
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Another strategy was also attempted to combine both types of particles in a
silica matrix, while limiting agglomeration of the boehmite during annealing. The
idea was to directly incorporate the pristine boehmite powder and chromium ions
inside the MS matrix alongside SPIONS, and only then perform the annealing
step. The boehmite is well dispersed in the silica matrix, and remains so during
annealing.
Different quantities of SPIONS in prehydrolysed TMOS were thus mixed to
HT-100 boehmite suspension in chromium nitrate solution. This mixture was
allowed to form a gel by addition of a base, and then annealed at 900 ◦C for one
hour. The SPION concentrations are given here as a percentage of the mass of
SPIONS to the mass of boehmite present in the sample (see figure 2.28).
Figure 2.28: Relative fluorescence intensity for different SPION concen-
trations. A: No SPIONS; B: 0.25 % SPIONS; C: 7.55 % SPIONS
The fluorescence signal is strongly reduced (55 %) with 0.25 % SPIONS, and
decreases to zero with 7.5 % SPIONS. If one compares this to figure 2.27, where
the SPION concentrations reaches maximum 25 % (5 mg Fe/ml) without a major
loss in signal, we can deduce that absorbance is not the key phenomenon at play
56
2.5 Chromium doped alumina
here. It is more likely that SPIONS interfere with the synthesis of chromium
doped alumina.
Chromium and iron oxide have a high affinity and form mixed oxides together
over a wide range of concentrations, as can be seen from the Fe2O3-Cr2O3 phase
diagram (figure 2.29). Iron oxide nanoparticles have even been used to purify
drinking water from chromium contamination [126].
Figure 2.29: Phase diagram between iron oxide and chromium oxide. C
= corundum structure; CC = chromium rich corundum; FC = iron rich corundum.
From [73].
If one synthesises the chromium doped alumina nanoparticles separately, they
can be combined to SPIONS at high concentrations without catastrophic loss of
functionality. When they are synthesised in the presence of SPIONS however,
the chromium ions seem to combine preferentially with the iron oxide and very
low levels of fluorescence of alumina are observed.
This closes the door to transforming well dispersed boehmite nanoparticles
into so many γ-alumina particles directly inside the already formed MS-SPION
beads, which would have the great advantage of conserving its excellent particle
size distribution.
57
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
2.5.4.2 Synthesis of MS-SPION-Ruby multifunctional particles
The encapsulation of chromium doped alumina particles alongside SPIONS inside
mesoporous silica particles was attempted using chromium doped alumina. The
experimental procedure is identical to the one described in section 2.3.1 except
that the PVA-SPIONS suspension in the emulsion were replaced with a 1:1 vol:vol
mixture of PVA-SPION suspension and chromium doped alumina suspension.
No fluorescence could be detected on the particles, and the continuous phase of
the emulsion seemed to be slighlty yellow. This indicates that the ruby particles
did not remain in the dispersed phase but were extracted into the oil. The
SPIONS on the other hand remained, as observed before, inside the particles.
Attempts using PVA-coated ruby particles yielded identical results: the particles
could not be kept inside the emulsion droplets.
2.5.5 Conclusion and outlook
The applicability of chromium doped alumina nanoparticles as near-IR fluorescent
markers is promising, although most aspects investigated here, from character-
isation to synthesis, would need to be completed and optimised. The quantum
yield of the material should be measured. This aspect was not addressed in this
study, but would be a necessary characterisation to have an idea of the intrinsic
value of this fluorophore for imaging applications.
The synthesis of well defined, homogeneously doped alumina nanoparticles
cannot without great difficulties and tedious processes (attrition milling, cen-
trifugation), be achieved from a commercial powder.
A good approach would be a “bottom-up” synthesis via, for example organic
alumina precusors. Zhou et al. have for example synthesised γ-alumina nanocrys-
tals from aluminum acetylacetonate in nonaqueous media [127]. One could imag-
ine to adapt this synthesis and perform it in the presence of chromium species. It
might even be possible to influence the relative intensity of each fluorescence phe-
nomenon, as well as to improve the quantum yield, by controlling the synthesis
conditions.
58
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
2.6 Laser annealing of silica coated ZnS:Mn2+
nanoparticles
2.6.1 Introduction
This section investigates the applicability of silica-coated manganese-doped zinc
sulphide (ZnS:Mn2+@SiO2) as multicolor luminescent material for display ap-
plications. Localized annealing of compacted ZnS:Mn2+@SiO2 powder was at-
tempted by heating the material with different lasers, so as thermally induce flu-
orescence color changes at desired sites on the surface. These thermally affected
zones were characterized by SEM, photoluminescence and cathodoluminescence.
The applicability of the method to the production of multicolour luminescent
images was assessed.
This work was conducted with the assistance of two successive students of the
material science master program, Michael Stuer and Amelie Bazzoni [8, 102].
2.6.2 Theory
2.6.2.1 Silica coated ZnS:Mn2+ nanoparticles
ZnS:Mn quantum dots are fluorescent have an orange emission peak (595 nm)
which is due to the T1(G)-A1(G) Mn2+ d-d transition [9] (see figure 2.30), and
a size dependant excitation peak which is around 300 nm (UV) for our synthesis
route. Pure nanosized ZnS has an emission peak at 420 to 450 nm, but doping
with transition metals such as Mn2+, Cu2+ or Eu3+ creates different emission
peaks in the visible spectrum [87, 118].
ZnS nanocrystals have a cubic zinc blende structure, also called sphalerite,
which consists in primitive cubic sulphur lattice with zinc occupying half the
tetrahedral sites. Wurtzite is a high temperature polymorph of ZnS, thermody-
namically stable above 1020 ◦C, but also found as a metastable phase at room
temperature in reductive conditions. It consists in an hexagonal close-packed
sulphur lattice with zinc occupying half the tetrahedral sites.
The special optical properties of these nanosized semiconductor crystals mainly
come from the confinement in space of their conduction band electrons and va-
59
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.30: Jablonski diagram for ZnS. Possible transitions and correspond-
ing colours in ZnS with or without manganese doping.
lence band holes, to dimensions approaching the exciton Bohr diameter. This
confinement causes an increase in the fluorescence efficiency, a shortening of the
fluorescence lifetime, and an increase in the width of the band gap compared to
the bulk material.
Photoluminescence (PL) properties, as well as other physical properties such
as crystal structure and defect density [32, 33], are also strongly dependant on
the condition of the nanocrystal’s surface. The presence of surface defects or
transition metal impurities near the surface, which act as surface states or“traps”,
tends to have a quenching effect and to therefore decrease the fluorescence yield
[123]. The presence of organic functional groups from surfactants or capping
agents on the surface can passivate such surface structural defects, and cause a
strong increases in fluorescence yield and PL intensity. Coating quantum dots
with another material possessing a wider band gap has likewise been shown in
many system to enhance their quantum yield. In particular, the effects of silica
coatings on the physical and photoluminescence properties of ZnS:Mn quantum
dots have been extensively studied [23, 52], and include a further increase in
fluorescence yield due to passivation of surface electron traps and an enhanced
thermal stability due to the reduced mobility of surface atoms.
60
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
2.6.2.2 Effect of annealing
Benedikt Steitz et al. [100, 101] found that thermal annealing of silica coated
ZnS:Mn nanoparticles in different atmospheres results in changes in the fluores-
cence emission spectra of these quantum dots.
The emission intensity of the ZnS:Mn2+@SiO2 passes through a minimum
when annealed at 600 ◦C due to the sphaerulite to wurtzite phase transformation
of ZnS through a non emissive transition phase. Upon annealing at higher tem-
peratures (>800 ◦C) ZnS:Mn2+@SiO2 yields different high intensity emissions in
different atmospheres. In air, green emissions (530 nm) due to the formation of
a new manganese doped zinc silicate phase called willemite (Zn2SiO4:Mn2+) is
obtained, probably due to the oxidation of sulphur into gaseous sulphur dioxide
gas. In formier gas, an emission peak is found again at 590 nm (orange) due to the
conservation at higher temperatures of ZnS:Mn quantum dots in their wurtzite
phase.
2.6.3 Materials and methods
ZnS:Mn nanoparticles were precipitated from a solution of zinc sulphate upon
addition of sodium sulphide in the presence of L-cysteine as surfactant. Silica
coating was carried out in a second stage by condensation of tetramethoxy or-
thosilicate (TMOS) on the quantum dots in aqueous environment. This suspen-
sion was then freeze dried to produce a fine white powder (see [101] for details).
Oven annealing was carried out in a thermo-gravimetric analysis (TGA) ma-
chine (Mettler). The ZnS:Mn2+@SiO2 powder samples were placed into alumina
70 µl TGA crucible, and heated up at 20 ◦C/min to temperatures between 400 ◦C
and 1200 ◦C, in a flow of air or formier gas. Immediately upon reaching the desired
annealing temperature, the samples were left to cool down to room temperature
in the gas flow.
96 wells microplate: a layer of alumina powder (α-Al2O3, AT-DS) was first
compressed in each well to about half height to act as a support layer. A
layer of ZnS:Mn2+@SiO2 powder was then compressed on top of it (see fig
2.31A). The well plates were placed in a chamber under a Nd:YAG (λ=1064
61
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
A B
C D
Figure 2.31: Different sample preparation methods for laser annealing.
A and B: 96 well microplate; C: quartz glass plates sandwich structure; D: pressed
pellet;
62
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
nm) laser (Quantronix® Condor100) in Q-switched mode. The laser was
programmed to fire inside the wells. This setup allowed to fire the laser with
a wide sweep of parameters (dwell time, power and laser frequency) directly
arranged in an array. It also allowed the characterisation of fluorescence
directly in a microplate reader.
Quarz glass plates: the setup consists in a sandwich structure between two
quartz glass plates separated by a PTFE spacer. The spacer has a window in
which ZnS:Mn2+@SiO2 powder is deposited (figure 2.31C). This method was
designed to provide as a“writing”surface for the laser, and was characterised
by photographs under UV light.
Pressed pellet: 2 to 5 g of ZnS:Mn2+@SiO2 powder were pressed into pellets in
a 2 cm diameter matrix placed in a uniaxial press at a pressure of 20 MPa.
The pellets thus formed are quite fragile and consume a large amount of
powder for a relatively limited writing surface when compared to the quartz
glass sandwich structure.
Laser annealing was attempted with the following lasers:
1. Nd:YAG infared laser (1064 nm, Q-switched mode.) Quantronix® Condor100.
2. Argon continuous laser Innova 300 (514 nm, 5 W).
3. UV pulsed excimer KrFNe-filled laser (248 nm) Lambda Physik LPX-100
laser.
4. Nd:YAG infared laser (1064 nm) SLAB LASAG.
All laser annealing attempts were done in air.
Fluorescence spectra were acquired on a Tecan Saphire2 microplate reader in
top mode with zero lag time and 40 µs integration time.
The photographs of fluorescent samples were taken under a shortwave UV
lamp (Waldmann Lichttechnik GmbH).
X-ray diffraction was realised on a Philips X’pert diffractometer with a Cu
Kα-radiation.
63
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Cathodoluminescence images and spectra were realised on a Cambridge S-360
CL system fitted with a LaB6 filament producing a 1 to 40 keV beam with cur-
rents from a few pA to 1 µA. The emitted light is collected by a semi-elliptical
mirror into a monochromator and detected by an InGaAs photodiode. The sam-
ple investigated was a pressed pellet exposed to a Nd:YAG (λ=1064 nm) laser
(Quantronix® Condor100) in Q-switched mode (19 A, 2.2 kHz for 15 seconds ;
570 J).
2.6.4 Results and discussion
2.6.4.1 Oven annealing
Samples annealed in air at 600 ◦C did not show any emission immediately after an-
nealing, in accordance with previous observations, but started to develop a strong
blue emission within 48 h (see 2.33). This effect was not previously reported and
is due to the gradual oxidation of ZnS:Mn2+ into zinc oxide, as demonstrated
by XRD spectra. This new property is of interest as it allows the production
of a third color, blue, in addition to the original orange emission, and the green
emission of the manganese doped zinc silicate with one starting material.
Figure 2.32: Colours obtained after annealing of ZnS:Mn2+@SiO2 at
different temperatures and in different atmospheres.
64
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
Figure 2.32 shows a summary of the different colors obtained during the dif-
ferent annealing processes. The maximal intensities of the different spectra were
set as 1 so as to allow comparison between them, although the intensity of the
blue emission is more intense than the green one. Human vision is however not
linear and sees green and orange better than blue [30], which is why the apparent
brightness of the different colors seems to be in the same order of magnitude on
figure 2.32.
Figure 2.33: Emission spectra of annealed ZnS:Mn2+@SiO2 samples.
The maximal emission intensity was set as 1 for all spectra. Excitation at λex=
301 nm for pristine material and 1100 ◦C in formier gas, 370 nm for 600 ◦C in air
and 250 nm for 1100 ◦C in air.
These colors could potentially act as primary colors for the development of
RGB (red, or rather orange, green and blue) static multicolor fluorescent images.
2.6.4.2 Laser annealing
Laser induced emission colour change can be achieved using both infrared lasers
(SLAB LASAG and Quantronics Condor100).
65
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
The two other lasers (green and ultraviolet) do not achieve sufficient heating
to create the willemite phase, and only result in a loss of fluorescence at the im-
pact points. This can be caused either by a moderate heating of the material
below 500 ◦C (see figure 2.32), or to bleaching of the quantum dots. Upon in-
tensive irradiation, light induced lattice defects in the quantum dot, sometimes
partially reversible, create extra non radiative recombination pathways and cause
a significant loss of signal [107]. The relatively short wavelengths used in both
cases (514 nm and 248 nm respectively) are susceptible of being absorbed by the
QD’s and to cause photobleaching.
No change in agglomeration or appearance of the powder (as seen under optical
microscope) at the impact site can be detected, indicating that the temperature
probably did not exceed 500 ◦C. No blue emission is detected at the impact site at
any time after the experiment, which means that the zinc sulphide is not oxidised
to zinc oxide, as is observed in the oven-annealed samples heated at 600 ◦C after
24 hours.
Laser induced thermal annealing can however be achieved with both pulsed
IR lasers used. The first attempts were conducted on the Quantronics® laser in
well plates, testing a wide range of lamp currents, pulse durations and irradiation
times in order to find appropriate conditions.
The thermal change occurs only after a certain threshold of illumination time
(up to tens of seconds), before which no effect can be detected, and then progresses
extremely rapidly. In all the wells where an effect can be seen, the diameter of the
thermally affected zone (1 to 2 mm) is substantially larger than the beam diameter
(10 µm). The response is thus strongly non-linear with respect to nominal total
power and irradiation time.
This effect can be explained by the fact that silica is quite transparent to
IR radiations, and that with a fixed pulse duration of only 130 ns, the heat
generated by each pulse is dissipated before the next pulse arrives. In this case, the
temperature increase remains moderate and no thermally induced phase change
is observed at first. The continuous illumination of the same spot for tens of
seconds can however cause optical breakdown of the material. The laser creates
coloured centres in the material which increases the absorbance locally, making
it more liable to create other coloured centres, resulting in a feedback effect that
66
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
leads to a sudden, non linear increase in temperature, and the formation of a
plasma [81]. This is how a thermal annealing effect is finally obtained with this
laser, proving that the concept is valid, but showing the necessity of optimising
the laser parameters in order to decrease the diameter of the thermally affected
zone and shorten the irradiation time.
On the SLAB LASAG infrared laser, contrary to the Quantronix®, the pulse
duration can be chosen. The experiments in this case was performed on quartz
glass sandwich samples, and the powder was exposed to single, long laser pulses.
The pulse duration was varied between 0.75 and 3 ms, which is three to four
orders of magnitudes greater than the duration of the pulse on the other IR laser.
The thermal effect in this case is probably also due to the creation of a cascade
of coloured centres, but the process is better controlled and occurs over a single
pulse. The effect is therefore more local, the thermally affected zone is in the
order of the beam diameter (200 µm), which allows the creation of patterns.
In order to have an effect visible by eye, square shaped arrays of laser impacts
of 2x2 mm2 were created. The impacts are spaced by 170 µm in one direction
and 200 µm in the other.
Figure 2.34: Array of laser impact points. Optical microscopy.
The effect of thermal treatment by laser in air is comparable to that of the tra-
ditional oven treatment and yields a strong emission in the green. The total pulse
energy is an important factor, and seems to yield similar results with different
parameters. Squares 15 and 13 have the same energy but different pulse lengths
67
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Square number Pulse duration [ms] Voltage Energy [J]
1 3 250 2.4
2 3 250 2.4
3 3 350 3.4
4 2 250 1.6
5 1.5 250 0.97
6 1.5 250 0.92
7 1.25 250 0.75
8 1 250 0.55
9 0.75 250 0.36
10 0.75 250 0.36
11 0.5 250 0.16
12 2 235 0.92
13 2 210 0.55
14 3 197 0.53
15 0.75 271 0.55
Table 2.8: List of conditions tested. The numbers correspond to the squares
on figure 2.35.
A B
Figure 2.35: Laser annealing of ZnS:Mn2+@SiO2 powder in different
conditions. A: photograph under white light. B: photograph under UV light.
68
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
and voltages. They however look similar in their colour and intensity. Squares
with the highest powers look more green (> 1 J), those with intermediate powers
(> 0.5 J) look more blue, and those with the lowest powers are hardly affected at
all and remain orange. It is interesting to note that no photobleaching effect is
visible on these squares although they have been exposed to strong IR radiations.
This characterisation method is qualitative and does not give exact emission
spectra, due to the difficulty in measuring fluorescence intensities on an arbitrary
surface. Cathodoluminescence could not be attempted on these samples due to
the quartz glass plate covering the powder, which could not be removed without
modifying the sample too much.
It should be noted that a non emissive zone can be seem at the edge of
the laser treated zone, which appears as a darker rim around the squares. The
powder undergoes partial sintering during the laser treatment and the treated
zones appear more dense and whiter. There is some shrinkage of the powder
linked to sintering and a relative depletion of powder at the border zones occurs,
creating this dark rim effect.
As a proof of concept, the laser was programmed to draw the EPFL logo on
a quartz glass sandwich sample. The result is encouraging although the letters
remain rather ill-defined due to the size of the pixels (laser impacts) available
with respect to the size of the writing surface.
A B
Figure 2.36: EPFL logo drawn by IR laser on a bed of ZnS:Mn2+@SiO2.
A: photograph under white light. B: photograph under UV light.
69
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
2.6.4.3 Cathodoluminescence
The laser’s thermally affected zone was characterised by cathodoluminescence.
Cathodoluminescence is a technique which consists in using an electron beam to
excite the fluorescence of samples, and to collect and analyse these light emissions
using a semi-elliptical mirror coupled to a monochromator. It was particularly
adapted in our case as a space resolved-spectrum of the impact zone could thus
be acquired.
Figure 2.37: Emission spectra of the laser treated and non treated zones
by cathodoluminescence.
The emission peak in the thermally affected zone is situated around 550 nm,
similarly to the emission peak of the sample annealed at 1200 ◦C in air (figure
2.33). The non affected zone has a the usual orange-red emission peak (around
600 nm) of the pristine powder. The results correspond to those obtained with
photoluminescence. Two secondary peaks are present at 450 nm and 650 nm on
both spectra. The former is most likely the emission peak of undoped ZnS, quite
visible in the non-treated spectum, and much less so in the thermally treated one.
The latter could however not be attributed to any event in particular.
70
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
A B
C D
Figure 2.38: Cathodoluminescence of ZnS:Mn2+@SiO2 before and after
laser treatment. A: luminescence in pristine region; B: SE image of laser im-
pacted region; C: luminescence at 450 nm of zone shown in B; D: luminescence at
550 nm of zone shown in B.
71
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
The morphology of the compressed powder becomes smooth and porous during
laser treatment, with spherical pores that could point to some gas formation
(bubbles, probably SO2) in the melted silica (figure 2.38B). The luminescence is
generated homogeneously over the whole surface of the pristine sample (2.38A).
In the laser treated region however, the light seems to be more intense coming
from the pores (2.38C and D). This is probably due to the greater penetration
depth of electron compared to visible light. Only light generated near the surface
can be detected, the rest being absorbed before reaching the surface and the
detector. Pores appear brighter because they constitute pathways through which
light can emerge from within the material.
EDX mapping of the pristine powder surface show inhomogeneous concen-
trations of zinc and sulphur, indicating that the quantum dots are present as
agglomerates inside the silica. In the laser treated zone on the contrary, these
elements are homogeneously distributed and seem to have diffused throughout
the material.
2.6.5 Conclusion and outlook
The applicability of ZnS:Mn2+@SiO2 powder for the production of multicolour
fluorescent images by laser annealing was proven. A two colour picture (the EPFL
logo) was realised thanks to a first optimisation of the laser treatment parameters,
which made it possible to diminish considerably the size of the thermally affected
zone and the irradiation time. It was seen that blue or green hues could be
obtained by adapting the total energy of the laser pulse.
The system remains however quite experimental and a great deal of optimi-
sation and further characterisation is necessary to progress towards a possible
application. Firstly, laser annealing could be attempted in different atmospheres,
such as argon of formier gas in order to achieve different colours, or even non emis-
sive powder. A moderate oven pre-annealing in formier gas at 400 ◦C could for
example extinguish all luminescence in the sample, yielding a dark background.
Subsequent laser annealing in the same atmosphere would rekindle a orange-red
emission at desired sites only. One could then attempt to create real multicolour
72
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
pictures using the three primary colours available (red, green and blue) as laser
annealed colour pixels.
The laser treatment also leaves room for improvement, in diminishing the
diameter of the thermally affected zone for example, to gain in resolution. The
mounting of the samples and the size distribution and properties of the powder
also have to be re-thought. Finally, the excitation by UV light could be replaced
by a more convenient electroluminescence setup, by placing the powder in a strong
electric field.
73
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
Figure 2.39: SE image of compacted ZnS:Mn2+@SiO2 surface and chem-
ical mapping of the surface.
74
2.6 Laser annealing of silica coated ZnS:Mn2+ nanoparticles
Figure 2.40: SE image of laser treated ZnS:Mn2+@SiO2 surface and
chemical mapping of the surface.
75
2. MULTIFUNCTIONAL MESOPOROUS SILICA PARTICLES
76
3
Titanium surface structuring
77
3. TITANIUM SURFACE STRUCTURING
3.1 Introduction
3.1.1 Titanium
Titanium is a transition metal found in various naturally occurring ores such as
rutile or ilemite. Despite its abundance in the earth’s crust, it is one of the most
recently discovered structural metals. Its mechanical resistance is similar to that
of steel while its density is twice as low, making it a particularly essential material
for the aerospace industry. It is also widely employed in the automotive industry,
in the chemical industry as a catalyst, as a structural material in sporting items
as well as for dental and orthopedic applications [7].
Beyond its excellent mechanical properties, its biocompatibility, its resistance
to corrosion as well as its availability make it an important material for many
current and future applications.
Titanium is mostly used either as a pure metal (grades 1 to 4), or as alloys
containing for example aluminium and vanadium (grade 5). The pure metal has a
hexagonal close packed structure (α-phase) at room temperature. Above 882 ◦C,
an allotropic transformation occurs from hexagonal close packed to body centered
cubic structure (α→ β).
The Ti-6Al-4V alloy (Grade 5) has a duplex microstructure with both α and
β phase at room temperature thanks to the stabilising effect of aluminium on
the high temperature phase. These alloys have excellent mechanical properties,
good workability thanks to the ductile β phase, and are widely used in orthopedic
implants.
3.1.2 Electrochemical properties
3.1.2.1 Passivity
Titanium is a passive metal that has a thin oxide layer on its surface, called
passive film, which separates it from its environment. This tenacious oxide is
stable over a broad range of pH and potential conditions as can be seen from the
Pourbaix diagram of titanium (figure 3.1). The typical thickness of such passive
films is in the order of 1 to 3 nm.
78
3.1 Introduction
Figure 3.1: Pourbaix diagram for titanium in aqueous environment.
According to [113].
The dissolution rate of a passivated metal (or passive dissolution) is very low
and constant over a large potential range. This explains its extreme chemical
resistance in all sorts of aggressive environments.
3.1.2.2 Anodic oxide film growth
Anodic oxide growth involves ion transfer reactions at the metal-oxide interface
and at the oxide electrolyte interface. It also involves ion transport inside the
oxide layer from one interface to the other. These phenomena occur under the
influence of a strong electric field (typically 108 V/m) in the film.
The reaction at the metal-oxide interface consists in the creation of titanium
ions from the bulk metal, which take up cationic sites in the oxide, according to
the following equation [54]:
Ti→ Ti4+(ox.) + 4e− (3.1)
There are two possible charge transfer reactions at the oxide-electrolyte inter-
faces: dissolution of the metal ions, or the deposition of O2−(ox.) ions.
Ti4+(ox.) → Ti4+
(aq.) (3.2)
79
3. TITANIUM SURFACE STRUCTURING
O2−(ox.) + 2H+
(aq.) ↔ H2O (3.3)
The transport mechanism inside the oxide film is governed by high field con-
duction. In the absence of an electric field, the transport of ions in the oxide is
governed by diffusion in the form of thermally activated jumps between sites in
the lattice, without preferential directions. In the presence of an electric field,
the potential barrier for the jumps in the direction of the field is reduced and the
one for the reverse direction is increased. A net ionic current flows which is an
exponential function of the electric field.
The thickness, structure and quality of the film thus formed depend mostly
on the applied potential profile (potentiostatic, potentiodynamic, galvanostatic
etc. ) and on the composition of the electrolyte. In the standard conditions used
throughout this study (potentiostatic mode, 0 to 100 V in 0.5 sulphiric acid), the
film thickness t depends linearly on the maximal applied voltage V following a
simple rule t = 2.5·V [6]1. The structure of the oxide layer is mostly amorphous at
anodisation voltages below 20 V, a mix of amorphous oxide and anathase between
20 and 50 V and fully crystallised anathase between 50 and 100 V.
At higher anodisation voltages, film destruction due to dielectric breakdown
starts to appear in the form of crater-like pores, mostly identified as polycrys-
talline anathase [115]. The mechanism for this breakdown is not clear, but the
influence of temperature elevation and internal stresses on the crystallinity of the
film is pointed out. The crytallinity of the layer could increase its conductance,
or cause an embrittlement leading to cracks.
3.2 Structuring of titanium surface
The objective of the current chapter is to create a general technology platform to
achieve nanometer-scale structuring of anodisable metals using particle lithogra-
phy, and to attempt to understand the mechanisms at play in such systems. The
two main directions in the surface structuring of titanium are anodic structura-
tion by anodisation in specific electrolytes, and top-down lithography methods.
1Reference from the thesis of P.-F. Chauvy [16].
80
3.2 Structuring of titanium surface
These methods are often used in combination as they do not necessarily produce
features at the same length scale.
Kim et al. [50] have formed various nanostructures by anodising titanium in
different mixtures of phosphoric acid and hydrofluoric acid. Titania nanotubes
with a mean diameter under 100 nm are grown out of the surface. The structures
obtained are quite homogeneous, but direct morphological control of the features
produced is difficult with such methods.
Another method, named local anodic oxidation (LAO) consists in performing
a local anodisation of the titanium surface using an AFM tip as an electrode, by
applying a voltage between the tip and the surface. In ambient air, the titanium
surface is covered by a thin water film which dissociates under the influence of the
electric field, allowing the adsorption of OH− ions and the growth of an anodic
oxide layer [31]. The molar volume increase due to the anodisation process results
in topographic structures.
This method was briefly considered for this work as a potential complementary
study to the e-beam experiment, as it could have made it possible to locally
measure anodisation currents around surface defects such as masks of adsorbed
particles. This could however not be realised for technical reasons, but remains
an interesting option for further studies
The other strategy comprises various lithography techniques, including classi-
cal lithography [128], laser lithography [16, 17, 42] and electron beam lithography
[48, 49, 55]. These methods all rely on a system of masks, either deposited
or constituted of titanium oxide grown from the surface itself. Electrochemical
treatments, either anodic oxidation or electrolytic dissolution of the material are
then performed selectively and in a space resolved way. These techniques are of-
ten grouped under the term electrochemical micromachining (EMM). The typical
length scale of such structurations is in the micrometer range upwards.
Chauvy et al. [17] have shown that the patterning of anodised titanium can be
directly achieved laser irradiation, without the need for a photo-resist. The oxide
layer can be locally sensitised towards dissolution by a single long UV laser pulse.
Through-mask dissolution is then performed to create cavities in the surface.
Through-mask anodisation of e-beam lithography samples was performed by
Kern et al. using 50 nm diameter circular SU-8 masks and anodising up to 150
81
3. TITANIUM SURFACE STRUCTURING
V without delamination or dielectric breakdown of the mask. The structures
produced are characterised by SEM but no specific attention was paid to the
topography of the sample after through-mask anodisation. The main focus of the
work remains through-mask dissolution anodisation.
Chu et al. [19] have created well organised titania rod-like structures using a
combined anodisation approach. A layer of aluminium is sputtered on titanium
and then anodised. The well characterised hexagonal planar pore structure of
anodised aluminium is then used as a mask through which the anodisation of
the underlying titanium is performed. Titania grows inside the pores, which
leaves an hexagonal array of vertical titania rods (60 nm height, 30 nm diameter)
after the removal of the alumina mask. This method represents an interesting
mixture between anodic structuration (of the alumina mask) and though-mask
electrochemical machining.
No mention could be found in literature of the use of particle lithography
techniques for the structuring of titanium surface by anodisation, either at the
nanometer or at the micron scale.
82
3.3 Electrochemical polishing of titanium
3.3 Electrochemical polishing of titanium
3.3.1 Introduction
In order to better understand the effect of particles on the topography of anodised
titanium, very flat, smooth titanium surfaces were a prerequisite for this work and
had to be produced in sufficient numbers. The samples must be smooth enough to
allow the detection of the defects caused by the particles. If the original roughness
of the surface is of the same order of magnitude as the pits or bumps caused by the
particles, they cannot not be detected. The samples must therefore be polished
to sub-nanometer roughness. The surface should furthermore be macroscopically
homogeneous and regular to serve as model surface for the ordered deposition of
particles by dip coating and other methods. Too great a variance in the quality
of the samples would render the investigation of the deposition mechanisms very
difficult by causing meniscus jumps during dip coating and inhomogeneity in
wetting ability. The number of samples produced also had to be large enough
to allow the study of deposition processes. Electrochemical polishing was chosen
for this study as it allows the production of very smooth surfaces and presents
several advantages with respect to mechanical polishing:
� No residual stress in the superficial layers
� No impurities or inclusions
� Time gains
� Better results in the micro-roughness range (<1 µm).
The electropolishing process was adapted from a method developed at EPFL
by Olivier Piotrowski [85]. Two directions were developed in parallel: an increase
in the quality of the polishing in order to meet the requirements mentioned above,
and an increase in the number of samples that could be produced. As a first ap-
proach towards increasing the quality of the polishing, a systematic experimental
optimization method was applied, starting from Piotrowski’s experimental setup.
New setups were also imagined to increase productivity and polish several samples
at once.
83
3. TITANIUM SURFACE STRUCTURING
3.3.2 Theory
It is not our purpose here to go in depth into theory, as electrochemical polishing
is a complex subject on which much research has been done. A few basic concepts
ought however to be introduced in order to understand the problems encountered
and how they were solved.
The principle of electrochemical polishing is the dissolution of a metal in an
electrolyte by a combination of a chemical potential and an electrical (anodic)
potential. In the case of titanium, the native inert oxide layer present on its
surface necessitates the use of high anodic potentials and concentrated acidic
electrolytes. The polishing effect is generally explained by the fact that the rate
of dissolution is limited by the speed of material transport at the surface. The
diffusion rate of ionic species involved in the dissolution process is the rate limiting
step in the reaction. If the initial roughness is in the order of magnitude of the
diffusion layer, ions dissolving from peaks have comparatively a shorter diffusion
path to tread than from valleys (see figure 3.15). As a result of their greater
accessibility, peaks dissolve faster than valley and a levelling effect occurs.
Figure 3.2: Levelling effect. Scheme of a sinusoidal surface of period λ and
amplitude ε. The solid line represents Nernst’s diffusion layer (thickness δ). Picture
adapted from [85].
Several conflicting theories exist as to exactly which diffusion phenomenon
limits the process. According to some, it is the diffusion of metallic ions produced
at the anode, for others, it is the diffusion of acceptor compounds, or complexing
ions from the solution to the anode. The results obtained by O. Piotrowski point
84
3.3 Electrochemical polishing of titanium
to the diffusion of tetravalent titanium ions as the limiting phenomenon in our
case.
Another similar phenomenon called brightening occurs at the surface, but at
a scale of the order of a few nanometers. It is also a mass transport limitation,
and is due to the existence at the surface of a thin layer called brightening layer
or Hoar layer. An explanation for this phenomenon is the formation of a contam-
inated oxide layer permeable to metallic ions. The effect of this layer, also called
adsorbate layer or anhydrous layer, is to annihilate the effect of crystallographic
orientation and grain boundaries on the dissolution rate and thus avoid crystallo-
graphic etching [37]. This is an advantage in our case as it allows the production
of very smooth homogeneous surfaces from a polycrystalline material.
Polishing and brightening occur only with a limitation in mass transport,
which is not obtained for all metals and in all situations. Achieving this condition
depends in our case on the experimental setup. Under mass transport limitation,
the current density at the anode is independent from the potential applied. It
goes through a plateau over a wide range of anodic potentials. Above this range,
sparking, heating and gas evolution appear and no polishing occurs. Below this
range, the current is limited mainly by the Ohmic resistance of the electrolyte.
As for all valve metals, the anodic dissolution of titanium takes place after the
dissolution of its native passive titanium dioxide layer. With increasing anodic
polarisation, the formation of pits in this passive layer occurs randomly across
the surface, as determined by local conditions (eg. defects, impurities). Polari-
sation curves (current intensity measured as a function of increasing voltage) by
Piotrowski for titanium in methanol-sulphuric acid electrolyte show that current
intensities measured are very irreproducible and that the surfaces thus treated are
not polished, but inhomogeneously pitted, and that dissolution only take place
in these pits.
On the other hand, inverse polarisation curves, starting at high potentials
and going down to zero, show good reproducibility, and the samples treated are
smooth. When starting at high anodic potential, pitting very quickly occurs all
over the surface so that it can be considered as homogeneously activated. This
activated surface can then dissolve homogeneously and does not form pits. This
85
3. TITANIUM SURFACE STRUCTURING
depassivation treatment avoids the random character of the pitting process and
was applied in our polishing process.
Under polishing conditions, the current density strongly depends on the speed
of the electrolyte flow on the anode. A high electrolyte flow speed on the titanium
surface will increase the value of the current density plateau and therefore the
dissolution rate. The best results are obtained with a homogeneous laminar flow
of electrolyte across the surface. To achieve this, a relative movement between
the sample and the electrolyte is needed, either by stirring the electrolyte or by
moving the sample in the electrolyte. Both types of approach were attempted,
namely rotating anodes, and stirred electrolyte.
3.3.3 Materials and Methods
3.3.3.1 Electrolyte
Figure 3.3: Titanium sample before polishing.
The electrolyte used for the electrochemical polishing of titanium is 3 M sul-
phuric acid (sigma 95-97%) in methanol (puriss, H2O<0.2%). Sulphuric acid is
added dropwise to methanol in a calibrated flask under magnetic stirring. The
magnetic stirrer is then removed and the flask is filled up to the mark with
methanol. Special care is taken to avoid excessive heating of the mixture, as
the mixing and the reaction between methanol and sulphuric acid is strongly
exothermic. The electrolyte is used freshly prepared.
86
3.3 Electrochemical polishing of titanium
3.3.3.2 Sample preparation
The samples are machined at a high level of finish out of a grade 2 titanium
plate of 0.5 mm thickness (cold laminated and annealed according to ASTM B
265 Gr. 2, Bibus Metalle AG) into rectangular strips of dimensions 40 x 7 mm2,
or alternatively into discs of 15 mm diameter. They are cleaned for 5 minutes
in an ultrasonic baths in D.I. water, acetone and isopropanol successively. The
samples were then masked as shown on figure 3.3, leaving an area of 0.7 cm2 on
one end to be polished and an uncovered segment at the other end for electrical
contact (see figure 3.3). Masking was done using painter’s adhesive tape, which
has excellent chemical resistance to the polishing electrolyte.
3.3.3.3 Rotating disc electrode
This setup is composed of a Pine® Modulated Speed Rotator (MSR, see figure
3.4) fitted with a disc rotating electrode, using the 15 mm titanium discs. The
electrolyte is cooled to a temperature of 15 ◦C in a double wall 1 l glass container
fitted to a cryostat circulator. The circulator is filled with a cooling fluid composed
of 40 %vol ethanol in water. The counter electrode is composed of a tubular
stainless steel grid covering the inside of the glass container.
3.3.3.4 Multisample rotating electrode
The multisample rotating electrode was constructed to increase the productivity
of rectangular samples. Rectangular samples were preferred to discs for their
potential application in dip coating. It is composed of a central cylindrical alu-
minium piece, on which up to six titanium strips can be fixed (see figure 3.4B).
This electrode is mounted on a shaft compatible with the Pine ® MSR. The end
of the strips are immersed in the electrolyte and rotated at variable speeds. The
setup is otherwise identical to the rotating disc electrode.
3.3.3.5 Electrochemical cell
The setup is based on a magnetically stirred double walled glass electrochemistry
cell with five necks. A homemade sample holder fixes the rectangular sample
87
3. TITANIUM SURFACE STRUCTURING
A B
Figure 3.4: Rotating electrode setups. A: Pine® Modulated Speed Rotator
with disc electrode; B: Multisample rotating electrode with up to six samples.
88
3.3 Electrochemical polishing of titanium
vertically, the end dipping in the electrolyte. Two platinum counter-electrodes of
1 cm2 each are fixed at the other side of the cell. 250 ml of electrolyte are poured
into the cell and cooled to −15 ◦C under vigorous magnetic stirring. The sample
is oriented to a 30 ◦ angle with respect to the flow to ensure homogeneous laminar
flow across its surface.
3.3.3.6 Multisample electrochemical cell
This setup is built around a stainless steel refrigerated cathode (also connected to
the cryostat circulator) placed in a magnetically stirred bath containing 600 ml of
electrolyte. The bath is hermetically covered with a lid equipped with an O-ring
and maintained with plastic clamps (see figure 3.5). The lid is perforated to fit
four independent sample holders, each capable of maintaining one rectangular
sample in position. The samples are connected electrically by crocodile clamps,
and the cathode is connected by a standard connector through the lid. Polymeric
foam is placed around the bath for thermal insulation, and to minimize frost
formation on the setup.
Figure 3.5: Multisample electrochemistry cell. Complete setup with four
sample holders (A); Refrigerated cathode and lid (B).
89
3. TITANIUM SURFACE STRUCTURING
3.3.3.7 Roughness measurements
Larger roughness scales measurements were done using a UBM laser profilometer
with a 1 µm diameter laser and a vertical resolution of 10 nm. The scanning length
was 2000 mm with a resolution of 2000 points/mm. 5 profiles were measured
for each sample. Small scale roughness measurements were done by atomic force
microscopy (AFM) using a Veeco CPII microscope in non-contact mode on 20x20
µm2 or 25x25 µm2 surfaces with a resolution of 256x256 points.
3.3.3.8 Roughness parameter
For each profile, an ideal baseline can be plotted where all the total area above
the line is equal to the total area below. Different roughness parameters can then
be calculated with respect to this baseline. Ra is the most generally used and is
calculated by taking the arithmetic average of the difference between each profile
point and the baseline over the whole profile length.
Ra =1
n
n∑i=1
|zi|
Where n is the number of points in the profile, zi is the height difference between
the profile and the baseline. Ra was the only roughness parameter used for all
measurements.
Roughness values are strongly dependant on the scale at which one performs
the topographic measurement. For example, mechanical polishing yields very flat
surfaces when measured on the millimetre scale and quite rough surfaces on the
micrometre scale, whereas the opposite is observed for electrochemical polishing.
Our topographic profiles were therefore analysed using a multi-scale roughness
analysis program called “Scale Crawler” developed by P.F. Chauvy at EPFL [18].
This program breaks down topographic profiles into baseline segments of vari-
able lengths (scales) and measures the corresponding average roughness for each
scale. It thus makes it possible to plot the roughness of the sample against the
scale and gives a complete overview of the topography.
90
3.3 Electrochemical polishing of titanium
3.3.4 Experimental optimisation of polishing
3.3.4.1 Experiment plan
The rotating disc electrode was used to perform a first study of the polishing
process. The factors chosen were the rotation speed of the electrode, the duration
of the polishing process and the applied anodic potential. It should be noted here
that all potentials given are measured between the counter electrode (cathode)
and the working electrode (anode), and that no standard reference electrode was
used. A 2x3x3 complete experiment plan was designed; the different factors and
levels can be seen in table 3.1.
Factor Definition Levels
1 2 3
A Rotation speed [rpm] 500 750 -
B Time [min] 5 6 7
C Anodic potential [V] 10 15 20
Table 3.1: Structure of the experiment plan.
The samples were characterised by laser profilometer, and the value of rough-
ness Ra at a scale of 200 µm was taken as the measured response. The 18 ex-
periments were conducted in random order so as to minimise systematic errors.
All samples were first treated at 30 V for 30 seconds, so as to activate the entire
surface. The voltage was then decreased to the nominal values indicated in table
3.2.
3.3.4.2 Results and discussion
The results obtained from this experimental optimisation plan are summarised in
figure 3.6.
The list of significant effects and interactions, as well as the magnitude of
these effects on the roughness parameter is shown in figure 3.7.
As can seen on figure 3.7, factor C, the applied potential, does not have an
influence on the roughness of the sample. This is not unexpected as polishing
91
3. TITANIUM SURFACE STRUCTURING
Figure 3.6: Results: Experimental values and values predicted by the statistical
model (Adjusted values) for all the different treatments.
Figure 3.7: Effect of the factors. Only four factors have an influence on the
roughness.
92
3.3 Electrochemical polishing of titanium
Treatment Exp. n◦ Rotation [rpm] Time [min] Potential [V]
111 1 500 5 10
211 2 750 5 10
121 3 500 6 10
221 4 750 6 10
131 5 500 7 10
231 6 750 7 10
112 7 500 5 15
212 8 750 5 15
122 9 500 6 15
222 10 750 6 15
132 11 500 7 15
232 12 750 7 15
113 13 500 5 20
213 14 750 5 20
123 15 500 6 20
223 16 750 6 20
133 17 500 7 20
233 18 750 7 20
Table 3.2: List of experiments.
occurs under mass transport limitation, and therefore independently of potential.
Both rotation speed and time have a strong influence on the roughness.
The effect of factors spread on three levels is interpreted as the sum of two
orthogonal polynomials that can be determined independently, one linear poly-
nomial multiplied by a factor (B1) and a parabolic polynomial multiplied by a
second factor (B2). In this instance, the effect of B is significant for both these
components. The four factors shown in figure 3.7 account for 90.16 % of the
experimental variability, the remainder being considered as experimental error.
Experiment plans aim to describe a local response surface for a given system.
The physical interpretation of each significant effect is not always possible, nor
is it the purpose of the method. For simple effects, such as the effect of A, the
rotation speed, we can say that faster electrolyte flow on the surface improves
93
3. TITANIUM SURFACE STRUCTURING
the quality of polishing. Conversely, the effect of AB2, the interaction of factor A
with the quadratic component of factor B does not necessarily lead to meaningful
conclusions on the mechanisms at play. The best approach is therefore to plot
the response surface, as is done on figure 3.8, and choose the appropriate factor
levels to minimize Ra.
Figure 3.8: Response surface at 20 V. Roughness parameter Ra as a function
of polishing time and rotation speed.
In this case, one would choose long polishing times and fast electrolyte flow
rates.
94
3.3 Electrochemical polishing of titanium
3.3.5 Other factors affecting polishing.
3.3.5.1 Effect of water.
The surface roughness of the samples produced was very variable, and although
good quality surfaces could sometimes be obtained, the method lacked repro-
ducibility. The whole process was therefore investigated and as a result, the
reproducibility and the roughness of the surfaces produced were significantly im-
proved. The reproducibility of the method can be impaired by two main factors
which had to be well understood and controlled. The first of these factors was
the influence of water in the polishing electrolyte. The electrolyte must be water
free as the presence of water causes re-passivation of the titanium and results in
pitting.
Characteristic polishing defects are caused by very small water concentrations.
The effect of large quantities of water on this electrochemical polishing process
was investigated by Olivier Piotrowski, who found that with more than 5 % water
content, the titanium surface is passivated.
Once formed, the oxide film tends to prevent surface dissolution. He also
found that polishing could be obtained with electrolytes containing up to 5 %
water provided the voltage condition chosen ensured that a saline dissolution film
had a chance to form in the first place. It should however be noted here that a
major difference exists between roughness characterisation methods implemented
by O. Piotrowski and those used here. Piotrowski used laser profilometry as a
main characterisation method, which gives very good results in the micron size
range, but cannot detect topographic features in the nanometric scale.
Given the objectives of this project, and the typical length scale of the struc-
turations created, AFM was used as topographical survey tool throughout this
study. As a result, many surfaces deemed “mirror polished” by laser profilometry
were in fact quite rough on the nanoscale, and typical polishing defects caused by
the presence of very small water contents were not detected. They however rep-
resented a real quality and reproducibility problem for our application. A typical
example of such water-caused polishing defects can be seen on figure 3.9.
The topographic features are only a few nanometers in height, but they are in
the range of our nanostructurations. The exact mechanism for this phenomenon
95
3. TITANIUM SURFACE STRUCTURING
Figure 3.9: AFM-Typical polishing defects caused by small amounts of
water in the electrolyte. The electrolyte was contaminated though condensation
from ambient air.
is not known, but it could be due to a local breakdown of the brightening layer
(or anhydrous layer) due to the presence of water.
3.3.5.2 Sources of water contamination.
Assuming the use of anhydrous starting chemicals, there are three main sources
of water contamination. The first one is the fact that these chemicals (methanol,
puriss. and sulphuric acid 95-97%) are very hygroscopic and that their exposure
to moist air will in time deteriorate the quality of the electrolyte. The second,
and perhaps major source of contamination is the condensation of water from the
atmosphere onto the surface of the electrolyte which is cooled at −15 ◦C. Surfaces
at such temperatures tend to condense water very efficiently, even more so when
the ambient relative humidity is high.
This effect was first suspected when looking at the quality of polished samples
produced over a period of one year. It was noticed that samples produced during
spring and summer presented more polishing defects than those produced during
winter. The ambient relative humidity during summer can be as high as 80
%, whereas it usually drops to 20 % during the winter. To avoid this seasonal
96
3.3 Electrochemical polishing of titanium
variance, the different polishing setups were placed in a glove box under a flow of
dry compressed air. The relative humidity could thus be lowered to values well
below 20 %, which was the lower measuring limit of the hygrometers used. The
preparation of the electrolyte and the handling of water sensitive chemicals were
also done in the glove box.
The third source of water contamination comes from the electrolyte itself.
Methanol and sulphuric acid react together according to [85]:
MeOH + H2SO4 → MeOH+2 + HSO−4 → MeOSO3H + H2O (3.4)
The first reaction produces persulphate ions and protonated methanol, and is
necessary for the dissolution reaction of titanium. The second reaction produces
methyl sulfonate and water and is therefore detrimental to the efficiency of the
electrolyte. It is furthermore dangerous for the health as methyl sulfonate is
poisonous. The second reaction is slow at room temperature, but can proceed
faster in an excess of sulphuric acid, or in case the electrolyte heats up too much.
This can happen during the preparation of the electrolyte, which is a strongly
exothermic process. To avoid this problem, the electrolyte must be prepared fresh
and prevented from heating up above 60 ◦C.
3.3.5.3 Non-anodic current limitations.
The second problem encountered can be summarised as non-anodic current lim-
itations. The current in the electrochemical cell is limited during polishing by
mass transport at the anode, which is mainly determined by the speed of the
electrolyte flux on the anode surface, as well as by the surface area of the anode.
Indirectly therefore, increasing the flux (faster stirring), or polishing large sur-
faces increase the demand in electrical current. As long as the electrical source
can provide it, the polishing process goes on. As soon as some other factor be-
comes limiting, the roughness increases dramatically, particularly on the parts of
the sample where the flux is highest.
The non-anodic current limiting factors are the cathodic reaction and the
maximal power of the electrical source. In the latter case, not much can be done
apart from decreasing the size of the sample. In the former case however, the
97
3. TITANIUM SURFACE STRUCTURING
surface area of the cathode can be increased. This explains why when polishing
in the exact same conditions two samples of slightly different sizes, the results
could be extremely different. The cathodic reaction can also be hindered by
surface contaminants (mainly titanium), which is easily remedied by inversing the
polarity of the polishing cell for a few seconds to dissolves them off the platinum
cathode.
3.3.6 Assessment of the different polishing setups.
3.3.6.1 Rotating electrode setups
The rotating disc electrode was a useful setup for the optimization of the pol-
ishing parameters. It had however certain drawbacks, such as the fact that the
electrolyte bath was not confined hermetically, letting substantial water conden-
sation take place and by that limiting the quality of the samples. Furthermore,
the sample shape was not adapted to the envisaged dip coating process. The
multisample rotating electrode did not have this problem, as it was built for rect-
angular samples, but the results obtained were disappointing. First of all it also
suffers from water condensation, and secondly, an inadequacy in the design cause
turbulent flow to appear already at low rotation speeds. Furthermore, the de-
mand in current for six samples was greater than what could be provided by the
power source. This option was therefore not pursued any further.
3.3.6.2 Fixed electrode setups
The conclusions from the experiment plan, in combination with the considerations
detailed in section 3.3.5.1 were implemented to the single sample electrochemical
cell setup with success. The parameters chosen in this case were a duration time
of 10 minutes and the maximal stirring speed attainable with the magnetic stirrer,
approximately 900 rpm. It is with this setup that the best results were obtained.
Figure 3.10 shows the plot of the roughness parameter Ra against the scale before
and after optimization as measured by AFM.
The roughnesses obtained before optimisation are similar to Piotrowski’s re-
sults at the same scale. The best samples produced in controlled atmosphere with
the single sample electrochemical cell do not exceed Ra values of 1.75 nm at 20
98
3.3 Electrochemical polishing of titanium
Figure 3.10: Multiscale roughness measurement from AFM profiles for
samples polished before and after optimisation.
µm segment length. An example of such a surface can be seen on figure 3.11. A
slight crystallographic etching is visible, with level differences between grains of
the order of 1 nm.
The multisample electrochemical cell worked well in principle and was used
to produce samples for the particle deposition studies. The results obtained were
similar to those for the single sample cell. For practical reasons however, the
single sample cell was preferred when the accent had to be put on quality rather
than quantity. It was furthermore never placed in controlled atmosphere, which
would certainly have improved its performance.
99
3. TITANIUM SURFACE STRUCTURING
Figure 3.11: AFM-Polished titanium surface.
100
3.4 Particle-substrate contact
3.4 Particle-substrate contact
3.4.1 Introduction
To understand and control the masking effect of particles on the titanium sur-
face, the dimension and nature of the contact between the polymeric beads and
the titanium surface was measured, modeled and eventually modified. In that
perspective, the effect of heat treatments and acetone vapour on the particle-
substrate contact radius was investigated.
3.4.2 Materials and methods
Commercially available polystyrene (PS) particle suspensions in water were used
for all experiment. They were purchased (estapor ®, France) in two different
nominal diameters, 200 and 500 nm (ref. K020 and K050 respectively) as 10%
w/w suspension in water.
Electropolished titanium substrates were treated for 5 minutes in air plasma
to enhance their wettability. 10 µl drops of 0.01 %vol suspension of 500 nm or
200 nm diameter PS particles in D.I. water were deposited on the surface. The
suspension was allowed to dry in a climatic chamber at 80% RH and 20 ◦C.
The heated samples were placed in a stove at 150 ◦C for different times. They
were then taken out and either dipped in D.I. water at room temperature, or left
to cool in air.
Samples treated with acetone vapour were placed in a sealed chamber con-
taining a petri dish filled with acetone. They are taken out after certain times
and dried in a flux of compressed air.
The particle-substrate contact diameter was measured on a high resolution
scanning electron microscope with an in lens secondary electron detector, at a
tilt angle of 85◦ so as to be able to image the contact site. The measurements
were done using imageJ image analysis program. Ten measurements were made
of each type of particle.
101
3. TITANIUM SURFACE STRUCTURING
3.4.3 Results and discussion
3.4.3.1 Pristine particles
The results of the measurements can be seen in table 3.4. Figure 3.12 shows
views of the particles at 85◦ tilt angle. The contact diameter is visible and can
be measured with around 5 % error on ten measurements.
Figure 3.12: Contact diameter measurement. 500 nm diameter particles.
A classical model to describe adhesive contact between deformable surfaces is
the one developped by Johnson, Kendal and Roberts (JKR) in 1971 [43]. It is a
simple extension of the Hertz theory of elastic contact that takes into account
deformations and stresses caused by adhesion phenomena at the contact site
[56, 67] and is valid under the following assumptions [95]:
� Deformations are assumed to be purely linear elastic
� Materials are elastically isotropic
� Young’s modulus and Poisson’s ratio do not change under pressure
102
3.4 Particle-substrate contact
� The contact radius is small compared to the radius of the particle.
The contact radius a0 between an adhesive deformable sphere of radius R and
a stiff surface can be described by equation 3.5.
a0 = 3√
9πγR2(1− η)/E (3.5)
Where γ is the interfacial energy between the polystyrene and the titanium,
E and η are Young’s modulus and Poisson’s ratio for polystyrene.
Young’s Modulus (at RT) 2.2 GPa [88]
Young’s Modulus (at 150 ◦C) 1 MPa [98]
η Poisson’s ratio 0.34 [98]
H Hardness 27 MPa [88]
Glass temperature 110 ◦C [98]
Melting Point 270 ◦C [98]
Surface energy 0.034 J/m2 [97]
Table 3.3: Properties of polystyrene. From literature.
The contact areas predicted by this model for polystyrene particles on tita-
nium can be seen in table 3.4.
R [nm] a0 JRK [nm] a0 measured [nm]
100 23.2 38
250 42.8 70
Table 3.4: Contact radii predicted by the JKR model vs. measured
contact radii.
The contact radii predicted by the JKR model are in a reasonable order of
magnitude, although significantly lower than those measured. This indicates that
the assumptions made above are not all true and that the JKR model cannot be
applied directly. It is likely, as reported by Rimai et al. [89], that some plastic
deformation occurs at the contact site.
103
3. TITANIUM SURFACE STRUCTURING
The Maugis-Pollock model [68] is an extension of the JKR model taking into
account plastic deformations at the contact zone. It predicts a square root depen-
dency between the contact radius and the particle radius according to equation
3.6.
Ha20 = 2RwA (3.6)
where wA is the work of adhesion, H is the hardness of the polystyrene. wA
is related to the surface energies of the material and their interfacial energy by
wA = γT i + γPS − γT i−PS (3.7)
Rimai et al. [88] have studied the adhesion of polystyrene spheres on silicium
wafers in quite a lot of detail. They have found the Maugis-Pollock model to
be suitable and were able to deduce a reasonable value for the work of adhesion.
Although the substrate is different, the surface energy of silicium (0.75 J/m2)
is in the same order of magnitude that of anathase (0.74 J/m2 [59]), the low
temperature polymorph of titanium dioxide. It is thus arguable that the works
of adhesion would be comparable. The radii measured in our case were therefore
directly plotted alongside theirs to assess the applicability of their model to our
situation. The particles they used were larger than ours and ranged from 0.81
µm to 5.9 µm, as can be seen in figure 3.13.
The results follow the trend predicted by the Maugis-Pollock model, our points
being very close to the linear regression curve established by Rimai et al. For our
situation, wA was therefore taken as equal to the one measured by them, namely
0.72 J/m2.
3.4.3.2 Effect of heat
The main effect of heating above Tg is to decrease the Young’s modulus of the
particles by several orders of magnitude.
The minimal temperature to obtain this effect is around 150 ◦C, which is
higher than the literature value of 110 ◦C for Tg of polystyrene (table 3.3). This
could be explained by the small dimensions of the particles, which can cause shift
in the glass temperature of the particles with respect to the bulk material due
104
3.4 Particle-substrate contact
Figure 3.13: Linear plot of the contact radius a0 against the square root
of the particle radius. The results of Rimai et al. (diamonds) and our results
(triangles) follow the Maugis-Pollock model.
Figure 3.14: Effect of heat on the contact diameter. 500 nm diameter
particles heated at 150 ◦C for 20 seconds.
105
3. TITANIUM SURFACE STRUCTURING
to entropic effects [69] as we approach a particles sizes similar to the radius of
gyration of the polymer chains. It could also be due to the level of cross-linking
of the particles’ surface. The heating time did not have a noticable effect on the
morphology of the deformed particles as for all times between 10 second and 5
minutes, the contact radius did not change significantly.
The JKR model cannot be applied to the case of the heated particles as
the contact radius is no longer small compared to the particle radius. Another
model developed by Lau et al. [56] describes the spreading of soft latex particles
on a substrate. The latex particles they use are styrene butadiene copolymer
with different degrees of cross-linking. These polymers have a glass transition
temperatures below room temperature and are therefore used in the rubbery
state.
At temperatures above Tg, Young’s modulus for polystyrene decreases by two
to three orders of magnitude. As in the previous models, the final shape of the
deformed particle at equilibrium is determined by the minimization between the
mechanical energy required to deform the particle, and the work of adhesion.
Due to the difficulty of a formal computation of the elastic energy for such high
deformations, this model computes the elastic energy over large deformations by
resorting of a scaling picture (see figure 3.15), assuming a spherical cap geometry
of the deformed particle and volume conservation.
Figure 3.15: Geometry of a deformed elastic sphere. a0 is the contact
radius, h the height of the spherical cap, R the radius of the sphere, uz(ρ) is the
deformation.
106
3.4 Particle-substrate contact
The surface Energy Us is given by:
Us = A · wa = πa20wa (3.8)
Where A is the contact area, a0 is the contact radius and wa is the work of
adhesion.
Assuming a constant total volume, we can calculate the radius R′ of the
deformed truncated sphere.
R′ =1
3
(4R3
h2+ h3
)(3.9)
The contact radius a can then be calculated to be:
a =√R′2 − (R′ − h)2 (3.10)
Using the value for the work of adhesion wa found by Rimai et al. [88], see
section 3.4.3.1 we can calculate the surface energy for all deformations h/2R.
The elastic energy as a function of h is thus calculated according to equation
3.11
Uel =8√3KR3Φ(h/2R) (3.11)
where K is a stiffness parameter defined as K = E/(1 − η2), and Φ(x) is a
scaling function given as:
Φ(x) =√x−1 − x2
[(1− x)2 − 4
9(1− x)(x−1 − x2) +
4
45(x−1 − x2)
](3.12)
Minimizing the total energy Utot = Uel + Us, and knowing the exact material
constants of the polymer, we can predict the equilibrium contact radius (see figure
3.16).
The main unknown parameter in this case is the exact value of Young’s mod-
ulus for the particles at 150 ◦C, this temperature being in the zone of transition
around Tg. Using the literature value of Young’s modulus for bulk polystyrene
above Tg (see table 3.3) does not give credible results.
107
3. TITANIUM SURFACE STRUCTURING
Young’s modulus was therefore varied in order to fit the model to the experi-
mental data. The conclusion is that if this model is correct, the 500 and 200 nm
particles do not have the same mechanical properties. The 200 nm particles seem
to be stiffer than the 500 nm particles.
Figure 3.16: Elastic, surface and total energies for the spreading of 200
nm and 500 nm PS particles at 150 °C. The energy minima are at h/2R =
0.57 and 0.65 respectively.
R [nm] h measured [nm] Calculated Young’s modulus [MPa]
100 116± 5 7.8
250 353± 14 3.3
Table 3.5: Calculated Young’s modulus according to the Lau et al.
model.
This phenomenon of size dependent stiffness for latex microspheres has been
reported in literature, particularly for cross-linked particles above Tg [36]. During
synthesis, the surface of the particles is cross-linked in order to obtain the desired
mechanical and surface properties. As a result, a core-shell effect is observed,
where the superficial layer around the particles is stiffer than the inside. The
absolute thickness of this layer is constant for all particle sizes, and this shell
effect is thus more dominant on the mechanical properties of smaller particles.
This could explain why the smaller particles are stiffer.
108
3.4 Particle-substrate contact
Figure 3.17: 500 and 200 nm deformed particles (150 °C for 5 minutes).
The two particles seem to have different stiffnesses.
The calculated moduli (table 3.5) are in the correct order of magnitude, the
literature value for the bulk being around 1 MPa (see table 3.3).
3.4.3.3 Effect of acetone
Another approach to modify the particle-substrate contact area was to make use of
the capillary condensation effect which occurs at the particle-substrate interface.
The interface region is a very narrow gap which tends to condense vapours from
the ambient atmosphere, as schematized on figure 3.18.
Figure 3.18: Capillary condensation around a spherical particle in pres-
ence of a vapour. Picture taken from [44].
Samples were placed in a saturated acetone atmosphere for 10 minutes before
they were taken out. The effect of this treatment was observed by SEM and can
be seen on figure 3.19.
109
3. TITANIUM SURFACE STRUCTURING
Figure 3.19: Effect of acetone capillary condensation. Acetone acts as a
solvent for PS and softens the particles.
This effect is interesting to modify the particle-surface contact. Heating was
however preferred for all structuring experiments as it is easier to control.
110
3.5 Particle deposition
3.5 Particle deposition
3.5.1 Introduction
Self assembled monolayers (SAM) are 2 dimensional periodic lattices formed of
monodisperse spherical colloidal particles. Self assembly of colloidal particle can
be achieved by making use of attractive capillary forces due to solvent evaporation
(see figure 3.20).
Figure 3.20: Schematic illustration of the phenomenon of particle diffu-
sion toward the meniscus encountered in processes such as dip coating
and drop-drying. Figure adapted from [44].
Evaporation causes a solvent flow towards the meniscus which carries the par-
ticles along with it. They accumulate at the meniscus without agglomerating and
begin to arrange themselves in a close-packed conformation. They are then forced
111
3. TITANIUM SURFACE STRUCTURING
into adhesive contact with each other and with the surface by the interstitial cap-
illary forces encountered during drying and form a hexagonal close packed array
of one or more layers. To achieve colloidal self assembly of particles on a surface,
a good colloidal stability of the particles against agglomeration is essential. The
particles must have strong repulsive interactions, due to electrostatic charges at
the surface for example, even at relatively short ranges (5 to 10 nm). They must
moreover have a strong repulsive interaction with the surface. In our case, the
latex particles used are negatively charged in water at neutral pH, with a zeta
potential of -50 to -70 mV. The titania surface is also negatively charged, which
provides adequate conditions of particle monolayer deposition.
The particle size distribution is also an important factor for the quality of the
SAM’s produced. Inhomogeneity in the size of the particles has a detrimental
effect on the ordering of the colloidal crystal produced. The particles used in our
case are monodisperse, with less than 5% dispersion in particle diameter and can
thus form high quality SAM’s.
One of the original objectives of this study was to optimise the particle depo-
sition process in order to be able to coat large surfaces (several tens of cm2), such
as the surface of implants, with dense self assembled monolayer from colloidal par-
ticles. The experimental difficulties encountered with the polishing process, as
well as those linked to the actual structuring of surfaces by anodisation, reduced
the urgency of optimising this aspect of the method.
Deposition methods were however found that allow the production of dense
homogeneous SAM’s over hundreds of microns, up to a few mm2, which was quite
sufficient for the requirements of this research.
3.5.2 Materials and methods
Polystyrene particle suspensions (estapor ®, France) in water were used for all
experiment.
Dip coating was realised on a dip-coater at various speeds using rectangular
polished titanium samples. The samples were pre-treated 5 minutes in air plasma,
dipped at various speeds in different particle suspensions (concentrations between
0.1 and 10 % w/w) and then dried in ambient air.
112
3.5 Particle deposition
Drop drying was performed on rectangular polished samples after 5 minutes of
air plasma treatment. A 10 µl drop of particles suspension at various concentra-
tion was pipetted at the center of the polished area. The drop was then allowed
to dry in a climatic chamber at 20 ◦C and either 10 or 80 % relative humidity.
The samples were characterised by scanning electron microscopy.
3.5.3 Dip coating
No dense SAM’s could be achieved by dip coating. The deposits produced were
generally quite inhomogeneous, with substancial de-wetting of the sample and
many meniscus jumps, and great local variations in particle coverage densities.
Figure 3.21 shows some of the different structures observed for a range of dip
conditions. For these reasons, and because of the quality and reproducibility of
drop drying, dip coating was abandoned as a deposition method.
In certain conditions, however, some order due to long-range electrostatic
repulsion between the particle was observed, as can be seen on figure 3.21B. This
ordering effect was used for the first structuring attempts (section 3.7.3.1).
Blattler et al. [12] have successfully deposited PS particle monolayers on glass
slides or silicium wafers by using very low retraction speed (from 1 to 10 µm/s)
in a strictly controlled atmosphere with 50 ± 5 % RH and 22± 1 ◦C. These very
low speeds are not attainable with our setup which furthermore cannot be placed
in a controlled atmosphere.
3.5.4 Drop drying
Drop drying is easier to control and more reproducible than dip coating. The rel-
ative humidity level and the temperature determine the evaporation rate. Figure
3.22 shows the results obtained for the drying of 10 µl drops of 0.1 % suspension
of 200 (K020) and 500 (K050) nm particles at 10 and 80 % relative humidity.
Samples dried at 80 % RH give better long range order than those dried at
10 %. Slower drying leaves more time for the migration and arrangement of the
particles, which can thus from larger ordered domains. Figures 3.22 A and B
show dense SAM’s spreading over several microns.
113
3. TITANIUM SURFACE STRUCTURING
A B
C D
Figure 3.21: Different deposits obtained by dip coating of 200 nm par-
ticle. A: 10% 10 mm/min.; B: 0.25 %, 1000 mm/min; C: 1%, 600 mm/min; D:
1%, 300 mm/min.
114
3.5 Particle deposition
A B
C D
Figure 3.22: Self assembled monolayers by drop drying. A: K050, 80%
RH; B: K050, 80 % RH; C and D: K050, 10% RH.
115
3. TITANIUM SURFACE STRUCTURING
3.6 Air plasma treatments
3.6.1 Air plasma for particle deposition
An important use of air plasma in this study is to increase the surface energy
of the polished titanium substrates. All deposition processes being based on
aqueous particle suspensions, the relative hydrophobicity of the titanium after
polishing is a problem as unwetting of the surface, as well as meniscus jumps are
detrimental to the desired homogeneity and ordering of the coating. After a five
minute treatment in air plasma, the surface is wet perfectly by water, without
the need for additional surfactants. This effect lasts for approximately one hour,
depending on the ambient humidity conditions, and so deposition processes were
performed within a few minutes after the end of the plasma treatment.
3.6.2 Air plasma for SAM preparation
The preparation of dense particle monolayers with high particle-substrate con-
tact radii cannot be realised directly as adjacent particles submitted to a heat
treatment above Tg tend to coalesce, as can be seen on figure 3.14. Such a heat
treatment, whilst increasing the particle-surface contact radius, causes the parti-
cles in a self assembled monolayer to coalesce into a film, which is not a desired
effect of the treatment.
Oxygen plasma oxidises the polystyrene particles and causes an erosion of the
material surface and a more or less uniform decrease in radius [12]. Applied to
dense particle monolayers, a short treatment causes the particle-particle contacts,
which are necessary to the ordering of the layer, to be eroded away, leaving isolated
particles in an ordered hexagonal array (see figure 3.23).
The thermal treatment (150 ◦C for 30 seconds in a stove, followed by tempering
in water at RT) can then be applied, increasing the particle-substrate contact radii
without significant coalescence of the particles.
The appropriate time for such plasma treatments depends on the original size
of the particles. 200 nm diameter particles subjected to an identical 2 minutes
treatment do not retain their order but are partially destroyed.
116
3.6 Air plasma treatments
A B
Figure 3.23: Dense monolayer of 500 nm particles before (A) and after
(B) 2 minutes oxygen plasma treatment. The particles retain their order but
do not touch each other.
Figure 3.24: Ordered array of hemispherical PS particles. Obtained from
a particle monolayer by plasma and thermal treatments.
117
3. TITANIUM SURFACE STRUCTURING
Figure 3.25: 200 nm diameter particles after 2 minutes plasma treat-
ment.
3.6.3 Air plasma for cleaning
After anodisation, the particles are removed using a combination of ultrasound
bath in acetone and oxygen plasma. A study of the effect of plasma treatment
was realised to assess its efficiency as a cleaning method. A sample was anodised
in the presence of deformed PS particles. The sample was treated with oxygen
plasma for one minute and then imaged by SEM under 85◦ tilt. The same oper-
ation was repeated 4 times, each time imaging the exact same locations on the
sample. The particles could thus be almost completely removed by plasma clean-
ing, although some small polymer deposits could remain even after prolonged
treatment. Blattler et al. report large star-shaped PS residues (> 200nm) that
cannot be removed by plasma cleaning. Such large residues were not seen in our
study, which could be due to the use of particles from a different supplier, with
possibly a smaller degree of cross-linking.
118
3.6 Air plasma treatments
A B
C D
Figure 3.26: Deformed particle after different plasma treatment times.
A: 1 min.; B: 2 min.; C: 3 min.; D: 4 min.
119
3. TITANIUM SURFACE STRUCTURING
A B
C D
Figure 3.27: Deformed particle array after different plasma treatment
times. A: 1 min.; B: 2 min.; C: 3 min.; D: 4 min.
120
3.7 Surface structuring with polystyrene particles
3.7 Surface structuring with polystyrene parti-
cles
3.7.1 Introduction
A first step was the characterisation of the effect of isolated particles on the
topography of the surface using atomic force microscopy. Determining whether
there is a bump or a cavity in the oxide layer underneath the particle was to
give us indications on the mechanism at work. A second step was to investigate
the effect of a self assembled monolayer of particles on the topography. Finally,
structuring attempts were made with heat deformed particles, both as individual
particles and SAM’s.
To detect structures caused by particles, it is important that the anodisation
process itself does not create too many topographic feature. Most studies were
therefore realised at an anodisation voltage of 40 V because it is an appropriate
voltage for the growth of a smooth, amorphous oxide layer with few anodisation
defects and a layer thickness of 100 nm which is enough to start observing effects
on the topography.
The influence of the particle-surface contact radius, the effect of deposition
in SAM’s as well as the effect of anodisation conditions (anodisation voltage and
voltage-time profiles) were investigated.
3.7.2 Materials and methods
PS microspheres of 200 nm or 500 nm diameter are used (estapor ®K020 and
K050) for all experiments. Several dilutions between 10 and 0.05 % vol were
prepared in DI water and used either in dip-coating or drop drying as described
previously.
Sparse coverages were obtained by dip-coating in 0.25% vol particle suspension
at 1000 mm/min. and drying in ambient air.
Self-assembled monolayers were obtained by depositing a 10 µl drop of 0.2 %
vol of particles on a surface previously treated in air plasma for 5 minutes, followed
by slow drying in a climatic chamber at 20 ◦C and 80 % RH.
121
3. TITANIUM SURFACE STRUCTURING
Anodisation was performed using a computer driven Agilent ®6030 A source
at different voltages in 500 ml of 0.5 M sulphuric acid against a cathode made
of a 30 x 10 cm2 stainless steel grid lining the walls of a 1 l glass beaker. The
samples were dipped approximately 1 cm into the electrolyte.
The particles were removed from the substrates after anodisation either by 20
minutes ultrasound treatment in D.I. water followed by 5 minutes in air plasma
in the case of the pristine particles, or by a 15 minute treatment in air plasma in
the case of the deformed particles. The samples were observed in a SEM (Phillips
XLF30) without further preparation after dip coating, after anodisation and after
cleaning.
AFM was performed using a Veeco Instruments® CPII setup with MPP-
11123-10 tips in non-contact mode.
3.7.3 Results and discussion
3.7.3.1 Influence of the contact radius
A difficulty in investigating the effect of a single particle on the topography of
the anodised surface is to be sure that any feature detected by AFM is indeed
caused by the presence of the particle during anodisation. To do so, the best way
would be to take a SEM picture of a certain area of the sample after dip coating,
perform the anodisation, remove the particles and then make an AFM picture of
the exact same spot.
Figure 3.28: Structuring with particles. The reality look more like this than
like figure 1.1.
This is however practically very hard to do, as moving the tip of the AFM
to a predefined location is not possible in our setup. The best option then was
to have a particle coating that was homogeneous on the majority of the surface,
and presented recognisable features (interparticle spacing, general appearance),
122
3.7 Surface structuring with polystyrene particles
so that the features then detected by AFM, if they showed the same type of
properties, could be without doubt attributed to the influence of the particles.
Figure 3.29: Sample coated with 200 nm particles. Dip-coating at 1000
mm/min in 0.25 %vol suspension.
The samples chosen to perform the topographical study present an homoge-
neous distribution of particles that is quite recognisable, as can be seen from
figure 3.29.
Due to strong repulsion forces between them, the particles are spread out to
a large majority as single particles, with a few pairs and triplets. The particles
retain this type of order after anodisation, indicating that they remain still during
the process.
The topography of the particle covered sample (figure 3.30) presents many
protrusions between 10 and 15 nm in height and around 80 nm in radius, spread
evenly across the surface. The right picture gives an idea of what the sample
would have looked like before the particles were removed, under the assumption
that the detected features were indeed caused by them. If one compares it to figure
3.29, one sees that the general appearance of the two pictures is very similar. One
can see that pairs and triplets are at the correct distance from each other (200
nm), indicating beyond doubt that the detected topographical features are the
effect of the particles. Figure 3.31 shows a plot of the profile across two of these
defects, and allows us to measure their dimensions.
123
3. TITANIUM SURFACE STRUCTURING
Figure 3.30: AFM image of anodised samples. All pictures are 5x5 µm2.
The picture on the right gives an idea of what the sample looked like before the
particles were removed (200 nm circles represent the particles).
Figure 3.31: Topography profile across protrusions. The peaks measure 10
to 15 nm in height and around 80 nm radius.
124
3.7 Surface structuring with polystyrene particles
These protrusions are very small and difficult to detect even with atomic force
microscopy. They were also difficult to reproduce. No structuration was observed
in the case of self-assembled monolayers of both 200 nm and 500 nm particles.
Heat deformed particles on the other hand have a significant effect on the
topography.
Figure 3.32: Effect of heat deformed particles on the topography.
Figure 3.33: Surface structuring with deformed particles. AFM, 500 nm
particles, anodised at 40 V.
As seen on figure 3.14, 500 nm diameter particles have contact radius after
heat treatment of the order of 700 nm. This large contact radius causes a much
more visible effect on the topography than in the case of the pristine particles.
As can be seen on figure 3.34, these structures are typically composed of
a protruding rim on the outside, immediately followed by a depression, or pit,
125
3. TITANIUM SURFACE STRUCTURING
Figure 3.34: Profile across the imprint of a deformed particle. The rim is
well visible on either side of the profile.
approximately 6 to 7 nm in depth and a radius corresponding to the particle-
surface contact radius.
At 40 V anodisation voltage, the oxide layer is approximately 100 nm thick,
and the height difference between anodised and non-anodised titanium is around
50 nm. This means that although the particles have an effect on the growth of
the oxide film, they do not hinder it completely: anodisation occurs underneath
the particles. This is perhaps more surprising than for the case of the pristine
particles, where the particle-substrate contact area is a hundred times smaller,
and the contact radius is in the size range of the oxide layer thickness. In the
case of the deformed particles however, with contact diameters almost reaching
micron sizes, it poses the question of the growth mechanism of the oxide layer
underneath the mask.
In the z direction, the electric conductivity of the oxide layer is the limiting
factor for the propagation of the oxidation process. As seen before, and for our
anodisation conditions, it virtually comes to a standstill after reaching a thickness
of 100 nm. One can thus argue that in the presence of a mask, the same limitation
should apply to the the horizontal propagation of the oxidation process as well,
assuming it occurs using the same transport mechanisms (i.e. ionic conduction in
high fields). Anodisation should thus not be able to progress further horizontally
than about 100 nm from the edge inwards. It however evidently does, which
points to a different transport mechanism than high field ionic conduction.
126
3.7 Surface structuring with polystyrene particles
3.7.3.2 Effect of particle deposition
Figure 3.35: Surface structuring with self-assembled monolayers of de-
formed particles. SEM, 500 nm particles, anodised at 40 V.
The effect is enhanced when the particles form SAM’s. The average height
of the structurations in SAM’s anodised at 40 V is around 12 nm, that is to say
twice the height of the structurations for isolated particles (figure 3.39). The
phenomenon causing the protrusion of the rims thus appears to be additive.
Some SAM’s obtained through optimised heat and plasma treatment do not
coalesce, and can form regular honeycomb structures over surfaces up to sev-
eral hundreds of µm2 (figures 3.36 and 3.37A). Straight wall-like segments form
between particles separated by narrow gaps (< 200 nm), as are found in such
non-coalesced SAM’s, which outline a well defined hexagonal array. These struc-
tures, which could be named narrow gap structures, are 30 to 50 nm in height.
This is an enhancement of the rim effect of approximately five to six times that of
single isolated particles, and by that a promising feature for surface structuring.
In other cases, the particles are still in contact in some points, and limited
coalescence occurs, yielding somewhat different structures (figure 3.37B).
127
3. TITANIUM SURFACE STRUCTURING
Figure 3.36: Regular honeycomb structurations caused by SAM’s of
deformed particles. Anodisation at 40V.
A B
Figure 3.37: Details of structurations by SAM’s. A: non-coalesced (regular)
monolayer; B: partially coalesced monolayer.
128
3.7 Surface structuring with polystyrene particles
3.7.3.3 Influence of the anodisation voltage
Different anodisation voltages were applied between 10 and 100 V on 500 nm
diameter heat deformed particles.
A B
C D
Figure 3.38: Imprint of deformed particles after anodisation at different
voltages. 20 V (A), 40V (B), 80V (C) and 100 V (D).
The effect of anodisation voltage on the height of the structuration produced
was characterised by two measurements realised on AFM profiles, as can be seen
on figure 3.39. The first (Rim) was the height of the top of the rim with respect
to the rest of the surface, while the second (Max) was the total amplitude of the
structuration, between the highest and the lowest point (see figure 3.34).
129
3. TITANIUM SURFACE STRUCTURING
Figure 3.39: Height of topographic features as a function of anodisation
voltage. Rim: height difference between the rim and the rest of the surface; Max:
total amplitude of the structuration, between the highest and the lowest point.
These height differences evolve similarly with the anodisation voltage. The
height of the rim represents for approximately 30 % of the total height of the
structuration, irrespective of the anodisation voltage. The effect of particles on
topography does not increase linearly with layer thickness, but is enhanced at
higher anodisation voltages.
Another notable influence of the anodisation voltage is the morphology of the
oxide layer formed under the masks. Figure 3.38 shows details of the oxide layers
underneath particles at different voltages, and while a more or less homogeneous
surface is found at 20 and 40 V, radial dendritic-like structures can be seen on
samples at 80 and 100V.
The oxide layer on these samples is visibly less dense underneath the mask.
A network of canals are formed between the edge and the centre of the imprints
(see figure 3.40). This canal network constitutes a possible alternative pathway
to high field ionic conduction for the transport of matter underneath the mask. A
model to explain this phenomenon could be the competition between the oxidation
reaction and the dissolution of titanium ions during anodisation.
Under normal anodisation conditions with unlimited spacial access to the
130
3.7 Surface structuring with polystyrene particles
Figure 3.40: Structure of the oxide layer underneath the particles. An-
odisation at 100 V.
electrolyte, only approximately 20% of the titanium ions arriving at the oxide-
electrolyte interface go into solution instead being oxidised [26]. It is possible
that this ratio is different underneath the mask, as the access to the electrolyte
is limited and the sites in the vicinity of the edges cannot provide sufficient ad-
sorbed oxygen to oxidate all the titanium ions. As a result, a large proportion of
the titanium starts to dissolve, creating cavities and canals in the underlying ma-
terial. These could then constitute pathway for the infiltration of the electrolyte
underneath the mask, allowing the propagation of the oxidation process further
than would have been possible with a classic high field conduction model. Such a
combination of dissolution and oxidation (growth) could explain how the anodic
oxidation of titanium can progress underneath masks.
Another interesting feature on figure 3.40A are the well visible narrow gap
structures. The growth of the titanium oxide dendrites (for lack of a better
word) seem to originate preferentially from these structures. The masking effect
is probably increased by the proximity of the other particles which could enhance
the dissolution-growth effect.
3.7.3.4 Influence of the anodisation profile
Anodisation was performed in the majority of cases in potentiostatic mode, with a
voltage step function (A, see figure 3.41). To study the effect of the voltage profile
131
3. TITANIUM SURFACE STRUCTURING
on the structuration, some samples were anodised in potentiodynamic mode. Two
voltage increase rates were applied, 20 V/s (B) and 400 V/s (C).
Figure 3.41: Scheme of the different anodisation treatments. A: Poten-
tiostatic mode; B: Potentiodynamic mode at 400 V/s (not to scale); C: Potentio-
dynamic mode at 20 V/s.
Samples anodised in potentiodynamic mode at either sweep rates did not
present any structurations at the voltage applied (40V), and no imprints of the
particles could be found. The rapid increase in voltage in the case of the po-
tentiostatic anodisation is limited only by the power of the electrical source, and
corresponds to a high initial growth rate of the oxide. This seems to be a deter-
mining factor in the formation of the topographic features.
3.7.4 Conclusion
The effect of adsorbed particles on the topography of the anodic layer was studied
in detail. It was found that anodisation was not prevented by the presence of the
particles, whether deformed of not, and that the anodisation voltage, as well as the
anodisation voltage profile had a strong influence on the topography. Topography
effects are also enhanced when the particles are separated by narrow gaps, which
results in larger structures in particle monolayers.
132
3.8 Model experiment using e-beam lithography
3.8 Model experiment using e-beam lithogra-
phy
3.8.1 Aim of the experiment
The following observations were during preliminary studies on the effect of polystyrene
particles on the topography of anodised titanium:
1. Particles do not prevent anodisation of titanium at the contact zone.
2. They nonetheless have an effect on topography.
3. They tend to leave protrusions rather than pits.
4. This effect is small and difficult to detect.
5. The particle-substrate contact area is small.
These observations prompted the posing of the following questions:
1. How large must a particle or mask be to prevent anodisation?
2. Are there specific conditions at the edges that locally cause the occurrence
of protrusions or pits?
3. Where are these topographic features situated with respect to the edge?
4. What is the effect of the mask’s size on the topography?
5. What is the structure of the oxide layer grown underneath the mask?
To answer these questions, a model experiment was designed. The idea con-
sists in creating an array of circular masks of different diameters on the titanium
surface, starting from the diameter of the particles and increasingly larger (100nm,
200 nm, 400 nm, 800nm, 2 µm, 4 µm and so on up to 100 µm). The titanium
is then anodised, the masks are removed by cleaning, and the topography of the
ensuing oxide layer is measured by atomic force microscopy. The process flow
diagram, along with a scheme of the pattern created can be seen on figure 3.42.
133
3. TITANIUM SURFACE STRUCTURING
Figure 3.42: Process flow and pattern for e-beam lithography. 1: Spin
coating of e-beam resist; 2: Exposure; 3: Development; 4: Anodisation; 5: AFM
characterisation.
Having such broad range of mask sizes on a single sample allows the covering
of several length scales at once, while the array design makes it easy to locate
with precision where each mask was situated, even after it was removed.
The smallest features (100 and 200 nm diameter masks) were too small to
be produced by conventional lithography technique. Electron beam lithography
was however perfectly adapted to our requirements, as it allows the fabrication
of both the smallest and the largest masks with good writing precision.
The principle of electron beam lithography is to expose a resin-coated surface
to an electron beam according to specific patterns. The beam trigger a chemical
reaction in the resin (e-beam resist) and causes it either to polymerise (positive
resist) or to degrade (negative resist). The resin is then developed, and the
unwanted areas are removed, leaving the desired pattern. For positive resists,
as was used in our case, only the exposed areas remain after the development
step, whereas the opposite is true for negative resists. The main advantage of e-
beam lithography is its capacity to go beyond the diffraction limit of light, which
typically limits the applicability of traditional lithography to the fabrication of
134
3.8 Model experiment using e-beam lithography
structures in the sub-micron range.
The use of this technique in our case was however not straightforward, and
several experimental difficulties had to be solved. Existing e-beam systems are
made to operate on standard silicium wafers, and the shape and dimensions of our
samples were not necessarily adapted to the processes used in microtechnique.
Furthermore, most available e-beam resists were silica based and had to be
removed with hydrofluoric acid. This was not an option in this case as there was a
good chance that the HF would interact with the amorphous titanium oxide layer.
A new polymeric resin was therefore tried, which despite wide use in lithography
had not been thoroughly tested for e-beam.
3.8.2 Materials and methods
Polished rectangular titanium samples were taken to a clean room under yellow
light, cleaned in an ultrasonic bath for five minutes in acetone and isopropanol
and then dried under a clean nitrogen flux. They were then taped to a silicium
wafer with Capton adhesive tape, placing the polished area at the center of the
wafer. The wafer was then fixed on a spin coater and the samples were spin coated
at 2000 rpm for 1 minute with 3 ml of nLOF resist in PGMEA at a dilution of
1/3. The samples were removed from the wafer and heated on a hot plate at
110 ◦C for 1 minute 45 seconds to bake the resin.
They were then placed in the electron beam lithography setup, and exposed
to the pattern chosen. For technical reasons the design was separated into two
parts. The arrays of smaller masks were written with a fine resolution. The larger
masks were then written with a coarser resolution so as reduce the writing time.
The pattern was written twelve times at increasing electron dosages in order to
find optimal conditions, starting a 200 µC/cm2, and increasing by a factor 1.2 at
each iteration up to 1486 µC/cm2 .
The sample were then post-baked at 110 ◦C on a hot plate for 1 minute, and
dipped in AZ 726 developer containing 2.38 % tetramethyl ammonium hydroxide
for 30 seconds under gentle swirling. They were then rinsed for 30 seconds in D.I.
water. The patterns were visible by eye at that stage.
135
3. TITANIUM SURFACE STRUCTURING
Anodisation was performed in potentiostatic mode at 40 or 55 V in 1 M or
0.5 M H2SO4 for 2 seconds. The resin was removed by 20 min ultrasound bath
in acetone followed by 5 minutes in air plasma.
The topographies were characterised by SEM and AFM as described before.
Focused ion beam (FIB) cross sections were performed using a NVision 40
CrossBeam (Zeiss) with the help of Dr. Marco Cantoni (CIME, EPFL).
3.8.3 Results and discussion
3.8.3.1 E-beam lithography
Figure 3.43 shows a general top view of the twelve repetitions of the design with
increasing dosages the top right to the bottom left corner, row by row. The
process yielded well defined structures for the weakest exposition dosage (200
µC/cm2). Larger dosages cause charging and decreases the resolution of the
patterns drastically.
Figure 3.43: Overview of the e-beam patterns for different electron
dosages. Leak currents and charging cause blurring of the features at higher
dosages.
At 200 µC/cm2 the fine features are well fabricated, although due to resolution
problems, the circularity of the smallest masks is not perfect as can be seen on
figure 3.44. This is a limitation of the technique and cannot easily be remedied.
The thickness of the masks was measured by SEM by imaging at a wide tilt
angle of 85 ◦. This imaging technique is very sensitive to topographical features,
136
3.8 Model experiment using e-beam lithography
Figure 3.44: Arrays of masks by e-beam lithography.
the grazing incident electron beam giving excellent contrast, which combined to
the large field depth of scanning electron microscopy makes it an indispensable
characterisation tool. The thickness of the masks is approximately 200 nm and
is homogeneous throughout the sample (figure 3.45).
Figure 3.45: Tilt view of the 4 µm, 800 nm, 400 nm and 100 nm masks.
137
3. TITANIUM SURFACE STRUCTURING
3.8.3.2 Effect of the masks on topography
The samples were imaged by SEM at each stage, before and after anodisation
and after removal of the masks.
Figure 3.46: 2 µm mask at different stages of the process.
Figure 3.46 shows the surface (2 µm mask) at these different stages. A certain
amount of delamination (mask peel-off) takes place during the process. It oc-
curs probably after the main topographical features (outermost rims) have been
formed, as they are well visible underneath the peeled off mask.
As can be seen on figure 3.49, the small masks are too small to prevent anodi-
sation, and the oxide layer is formed underneath. For the larger masks however (4
µm diameter), a relatively deep hole (45 nm deep) is visible in the centre, corre-
sponding to the level difference between anodised and non-anodised titanium. For
this voltage and in these conditions, anodisation stops abruptly approximately 1
µm away from the edge of the masks. Figure 3.51 shows the height profiles across
the different masks. The step between the anodised and the non anodised metal
is visible on the 4000 nm profiles anodised at 40 V.
138
3.8 Model experiment using e-beam lithography
Figure 3.47: SEM image of the e-beam sample after anodisation and
removal of the masks.
Figure 3.48: Tilt SEM image of the e-beam sample after anodisation
and removal of the masks.
139
3. TITANIUM SURFACE STRUCTURING
Figure 3.49: AFM topography of the e-beam sample after anodisation
and cleaning. Anodisation at 40 V in 0.5M sulphuric acid.
Figure 3.50: AFM topography of the e-beam sample after anodisation
and cleaning. Details of the 100 and 200 nm masks.
140
3.8 Model experiment using e-beam lithography
Figure 3.51: Height profiles across the different masks. Anodisation at 40
V in 0.5 M sulphuric acid.
There are rims around all the masks of about 10 nm in height, except for the
smallest 100 nm masks which did not leave any detectable topographical features.
Concentric height minima and maxima are visible on the edges of the masks.
The effect of mask diameter on the amplitude of the topographic structura-
tions at the edges is shown in figure 3.52. The measurement displayed here is
identical to Max on figure 3.39, namely the height difference between the first
maximum (top of the rim) and the minimum point of the oxide layer (see figure
3.51).
There is no significant effect of mask diameter at 40 V anodisation voltage.
At 55 V, the effect of mask diameter is more pronounced but does not seem to
evolve linearly.
3.8.3.3 Effect of anodisation parameters
There are notable differences in topography between the sample anodised at 55
V and the ones anodised at 40 V. As can be seen on figure 3.53, the oxide layer at
55 V could form on the whole sample surface, including underneath the masks.
The titanium was completely anodised, even under the largest 100 µm masks.
The reason for this difference probably lies in the fact that the electric field
141
3. TITANIUM SURFACE STRUCTURING
Figure 3.52: Effect of mask diameter on the edge structure.
A B
Figure 3.53: Effect of the anodisation parameters. Topography left by the
400 nm (A) and 800 nm (B) masks for different anodisation conditions.
142
3.8 Model experiment using e-beam lithography
trough the masks was strong enough to cause sparking across it, damaging it and
making it permeable to the electrolyte. The damage is visible on figure 3.55.
Figure 3.54: Profiles across the different masks for different anodisation
conditions.
The sparking results in a very rough surface, presenting needle-like structures,
some of which are visible sticking out through the mask on figure 3.55. The
topography at the rim is similar for all samples and all mask sizes, although the
height difference between the first maximum and the first minimum is larger for
the 55 V samples than for the 40V samples: 20 nm instead of 10 on average.
This can be explained by the higher thickness of the oxide layer formed at 55 V
(approximately 140 nm).
The electrolyte concentration does no seem to have a very notable effect on
the topography, except that the transition between anodised and non-anodised
titanium seems to be smoother. Another noteworthy difference is that neither
the 55V-0.5M sample nor the 40V-1M samples had visible topographic effects
underneath the 200 nm masks, whereas the 40V-0.5M samples clearly did (figure
3.50). This could either be an effect of the anodisation conditions themselves,
143
3. TITANIUM SURFACE STRUCTURING
Figure 3.55: Damage through a 100 µm mask after anodisation at 55 V.
or to some batch effect, these two samples having been prepared and anodised
at a later date. The production of such samples is very time consuming and
requires trained staff in a clean room; statically significant numbers could not be
produced, and so it is difficult to assess the reproducibility of the process. The
samples do however look identical in the SEM.
The 800 nm 55V-0.5M sample (figure 3.53) has radial structures that could be
an indication of the way the oxide layer grew underneath the mask. These struc-
ture are similar to the ones observed on the substrates structured with particles
at 80 V and 100 V (see figure 3.40).
3.8.3.4 Cross-sections by focused ion beam
The focused ion beam technique was very helpful in investigating the structure
and morphology of the oxide layer underneath the masks. The principle of the
technique is to use a focused beam of gallium ions to mill away materials in a
very controlled and precise way. It is coupled to a scanning electron microscope,
which makes it possible to realise cross sections, TEM samples and other nano-
and micrometer scale machining tasks.
In our case, the technique was employed to image cross sections of the oxide
layer, especially underneath the masks. A conducive carbon layer is first deposited
on the site to avoid charging of the sample, starting from a gaseous precursor
injected into the microscope chamber and which decomposes into amorphous
carbon under the action of the ion beam. A trench is then milled away so as to
give access to the desired cross section (figure 3.56).
The information sought in doing this experiment was chiefly concerning the
situation of the metal-oxide interface, of which nothing could be learned from
144
3.8 Model experiment using e-beam lithography
A B
C D
Figure 3.56: Different steps of the FIB technique. A: original surface; B:
carbon layer deposition; C: milling; D: detail of the cross section.
145
3. TITANIUM SURFACE STRUCTURING
atomic force microscopy or the other techniques used so far. The second point
of interest was to image the oxide layer itself and try to see if any morphological
differences could be found between regular areas and under-mask zones.
Figure 3.57 shows the different cross sections on the e-beam sample anodised at
40 V in 0.5 M sulphuric acid. A shows the 4 µm mask’s imprint. The oxide layer is
visible on either sides of the imprint and gradually becomes thinner in the center
where it disappears. The original titanium surface is situated approximately in
the middle of the oxide layer, indicating that it grows in both directions equally.
The layer is about 100 nm thick, which agrees with the 2.5 nm/V rule applied to
potentiostatic anodisation of titanium between 0 and 100 V [6].
The rim protrusions are visible on the cross section and are marked by ar-
row heads. The metal oxide interface presents fluctuations much like the oxide
electrolyte interface.
When looking closely at the oxide layer underneath the masks, one can see
that its colour is slighly darker than for the rest of the layer, and that small
pores are present. These images were taken using the backscattered electron de-
tector, and a darker colour in that case indicates a lower density of the material.
The eroded titanium underneath the masks presents so-called waterfall or cur-
tain effect, visible as a series of vertical lines. This effect is seen when the ion
beam encounters a material with density variations, which cause it to be slightly
deflected, resulting in these lines.
3.8.3.5 Conclusion
The observations made on the e-beam samples support the ones made on the par-
ticle structurations in section 3.7: the growth mechanism underneath the masks
imply the formation of pores in the layer. In the case of low anodisation voltages,
i.e. slower film growth rates, the pores are not easily visible on the layer in top
views, but appear in cross sections. They are smaller and less oriented, and seem
to be dispersed randomly in the layer. For higher anodisation voltages (> 50 V),
the growth rate is higher, the pore are larger, further apart and more oriented.
Not only do the pores orient radially, they also tend to align in concentric
circles, as can be seen from the topography profiles on figure 3.51. Although
146
3.8 Model experiment using e-beam lithography
the exact correlation between concentric rings and pores is not clear and would
require a slice by slice FIB tomography of the entire imprint, it is reasonable
to assume that they are indeed correlated. The e-beam experiment allowed a
thorough characterisation of the effects of the masks’ diameter, and provided a
complete picture of the typical edge defects caused by masks during anodisation.
147
3. TITANIUM SURFACE STRUCTURING
A
B
C
Figure 3.57: FIB cross-section across the different e-beam imprint. A:
4 µm imprint (the top images are AFM 3D reconstructions of the edge of the
imprint.); B: 2 µm and 800 nm imprints; C: 200 nm imprint. Arrow heads mark
the edges of the masks, arrows mark the edges of the oxide layer.
148
3.9 Finite element modeling
3.9 Finite element modeling
3.9.1 Introduction
Finite element modeling was used to gain insight into the mechanisms behind the
observed topographic structuring effect. A commercial program called “COM-
SOL ® Multiphysics” was used. This program makes it possible to create 2D or
3D models and to apply several types of physical calculations at once. These dif-
ferent “physics”, as they are called, can be coupled together in order to take into
account their interactions. A classical example of this is the electrical conduction
through a resistance. The electric current density is influenced by the conduc-
tivity of the material, which is a temperature dependent parameter. The electric
current produces heat (Joule effect), increasing the temperature and modifying
the conductivity. The two parameters are thus intimately coupled and can be
calculated by two different sets of partial differential equations using COMSOL
multiphysics.
3.9.2 Single physic model
First, a simple model was created to study the way the electric current at the
metal oxide interface behaves around an insulating mask. The “DC-conductive
media” physic was chosen and a 2D model representing the e-beam lithography
system was drawn. The system was schematized as shown on figure 3.58, and
each domain of the system was given a conductivity value taken from literature,
corresponding to titanium, PMMA and the electrolyte, 0.5 M sulphuric acid (table
3.6). The passivation layer was omitted for this first model.
The two vertical external boundaries were set on electrical insulation (normal
component of the current density jn = 0), the top external boundary was set as
grounded, and the bottom boundary was set to a constant electrical potential of
40 V.
The solution can be seen on figure 3.59, with the arrows showing the mag-
nitude and direction of the current, and the surface showing the total current
density. The current density at the edge of the masks much higher than else-
149
3. TITANIUM SURFACE STRUCTURING
Domain Material Conductivity [S/m]
Substrate Titanium 7.4 x 105
Passivation layer Titanium dioxide 1 x 10 −6
Mask PMMA 1 x 10−14
Electrolyte 0.5 M H2SO4 4.3 x 104
Table 3.6: Electrical conductivities.
Figure 3.58: Scheme of the model system. 1: titanium substrate, 2: PMMA
mask, 3: Electrolyte.
150
3.9 Finite element modeling
Figure 3.59: Current density around a mask. Arrows show the direction of
the current, colour scale shows the current density
where in the layer. The growth rate of the oxide layer, which is influenced by the
local current density, would therefore also be higher at the edge.
3.9.3 Two physics model
To simplify the model and shorten calculation times, only the oxide layer itself
was included in this second model. The conductivity of the electrolyte and of
the titanium substrate are indeed so much higher than that of the oxide layer,
that these interfaces can be considered as conductive boundaries (tangential com-
ponent of the current jt = 0). The PMMA mask was replaced by a completely
insulating external boundary (jn = 0) instead of a continuity condition at an
internal boundary with an extremely insulating material.
The native layer has normally a thickness ranging between 1nm and 5 nm [94]
but due to meshing difficulties, such a small layer caused the calculations to stop.
As a result, the native layer was drawn proportionally thicker than it actually is,
but still less than a tenth of the size of the mask. Ultimately, it is like simulating
151
3. TITANIUM SURFACE STRUCTURING
Boundary type DC-conductive media Moving Mesh
Vertical (1, 6) Insulation Free displacement
Top (3, 5) Ground Equation 3.14
Bottom (2) Potential (40 V) Equation 3.13
Mask (4) Insulation Free displacement
Table 3.7: Boundary conditions. Edge numbers correspond to figure 3.60
the anodisation at 40 V of sample that would have been pre-anodised at 5 V.
According to Faraday’s law, the amount of product in an electrochemical
reaction is proportional to the amount of charge that has passed through the
system. This is actually a simplification of the situation in the case of the anodic
oxidation of titanium, as a significant part of the current lost in a series of other
reactions. The dissolution reaction of titanium in the electrolyte accounts for
example for approximately 20 % of the total current [16]. For simulation purposes
however, the growth rate of the oxide layer can be considered proportional to the
local current density.
In order to simulate the growth of the oxide layer during anodisation, a second
physic called “moving mesh” was introduced and coupled to the “Direct current
conductive media”. A moving mesh system makes it possible modify the geometry
of a system by moving single points, boundaries or entire domains as a function
of local physical variables. Anodic oxidation was thus simulated by moving the
boundaries of the oxide layer as a function of the local current density. The
bottom boundary (see figure 3.60) is given the displacement condition:
Figure 3.60: Oxide layer with mask. Boundary conditions can be seen in table
3.7.
vn = −F ·MTi · jnρTi · n
(3.13)
152
3.9 Finite element modeling
Where vn is the normal velocity, MTi is the molar mass of titanium, F is
Faraday’s constant, ρTi is the density of titanium and n is the valency number of
titanium ions.
Accordingly, the top boundary is given the condition:
vn =F ·MTiO2 · jnρTiO2 · n
− F ·MTi · jnρTi · n
(3.14)
The mask boundary is left to move freely, as are the two vertical boundaries.
This means that for each Coulomb going through the oxide layer, the top bound-
ary moves up by a certain distance, and the bottom boundary moves down by
another distance . It is worth noting that titanium dioxide has a molar volume
of 19.97 cm3/mol and titanium one of 10.62 cm3/mol. This means that the ve-
locities of the top and bottom boundaries are in fact almost equal and opposite,
as each nanometre of titanium oxidized gives roughly 2 nm of oxide.
3.9.4 Results and discussion
Figure 3.61: Oxide layer growth around a mask. The edge of the mask is
higher than the rest of the layer.
A second situation was simulated where two masks (of different sizes in this
case) are separated by a small gap, as would be the case between adjacent parti-
cles. The idea is to see if the edge effect is additive, and if the topography at a
narrow gap between two masks protrudes more than at the edge of a single mask.
153
3. TITANIUM SURFACE STRUCTURING
Figure 3.62: Detail of the edges at different anodisation times.
As can be seen on figure 3.63, the effect is indeed at least partially additive, as
the protrusion situated between the masks is clearly higher than the rest of the
surface.
Figure 3.63: Oxide growth between two adjacent masks.
3.9.5 Conclusion
This model gives an explanation to the presence of rims around the masks: the
increased current density at the rim causes a faster growth of the oxide at that
location. It describes reasonably well the situation at the very early stages of
anodisation and gives insight into the formation of the topographic features en-
countered in our study. An enhancement of the effect at narrow gaps is observed,
as was the case between adjacent particles in SAM.
The model does not however explain the growth of the oxide layer underneath
the mask. It is does not take into account mask peel-off, titanium dissolution,
sparking, heating or gas evolution. Furthermore, the DC current conductive me-
dia physics relies on an Ohmic conductivity behaviour which does not adequately
154
3.9 Finite element modeling
describe high field ionic conduction. In the real case, the growth rate of the layer
decreases exponentially according to [54, p. 251]:
dL
dt= A i = A exp
(B∆Φ
L
)(3.15)
Where A and B are material constants L is the layer thickness and ∆Φ is the
potential difference across the layer. With an Ohmic behaviour, the growth rate
of the layer follows dLdt
= const.∆ΦL
.
155
3. TITANIUM SURFACE STRUCTURING
156
4
Conclusion and outlook
The original objectives and concepts underlying this project were partially ful-
filled, and partially found to be unrealistic. Surface structuring by anodisation in
the presence of particles was shown to be a valid concept which can produce or-
dered topographic features in the nanometre range. Mesoporous silica multifunc-
tional particles were found to be inadequate for this application. A gel-emulsion
synthesis method was however scaled up and optimised, allowing the production
of up to 60 mg of multifunctional particles per synthesis. This synthesis could be
further improved by using a continuous emulsification process such as for example
a high pressure homogeniser instead of a turbine.
The multifunctional particles developed were thoroughly characterised, and
although they were not applied to surface structuring, they were investigated
for other potential applications such as drug delivery, bio-imaging and display
applications. The interactions of the particles with the model drug paclitaxel
was studied, leading to a better understanding of the drug loading, transport
and release phenomena, and possibly opening the way to their application as
magnetic drug delivery vehicles. The biological effect of the drug loaded particles
should be investigated in future both in vitro and in vivo, along with the potential
enhancement of magnetic resonance imaging due to their magnetic properties.
The production of near-IR fluorescent nanoparticles without heavy metals
was attempted using the chromium doped alumina system. Although interesting
results were obtained, and near IR fluorescence was observed, this development
157
4. CONCLUSION AND OUTLOOK
was dropped after a point as the subject was though too far thematically from
the main focus of this work.
The interesting effects of thermal annealing of silica coated ZnS:Mn nanopar-
ticles on their fluorescence was exploited to successfully produce multicolour flu-
orescent images by laser thermal annealing.
The polishing process was mastered sufficiently to provide excellent starting
surfaces for surface structuring. The effect of particles on the topography was
thoroughly characterised, and the e-beam model experiment as well as the nu-
merical models provided some insight on the system.
The next step would be to optimise particles deposition and deformation, so
as to allow the creation of structured surfaces of the order of mm2, for example
by dip-coating or inkjet printing.
This perhaps opens the way for particle lithography in electrochemistry, not
only for anodisation, but also for through mask titanium dissolution. A few
attempts were conducted during this project, without success, but it could prob-
ably be achieved. The advantages of a well controlled bottom up process such
as colloidal self assembly could be of interest for some applications compared to
complex top-down lithography techniques.
Finally, this system could be used to investigate the role of different struc-
turations at different scales, on the adhesion and proliferation of cells on anodised
titanium, for example in the case of titanium bone implants.
The creation of structured topographies using particles, and the understand-
ing of the phenomena involved, have been general thematic guidelines of this
project. The exploratory nature of the subject however left me the freedom to
be creative in the ideas and methods used to move towards this goal. I enjoyed
this freedom very much, although it was necessarily accompanied by a relative
thematic isolation in my work.
158
Bibliography
[1] Materials Studio. Accelrys Sofware Inc., San Diego USA. 32
[2] I. Abarkan, T. Doussineau, and M. Smaihi. Polyhedron, 25(8):1763–1770,
May 2006. 26, 28
[3] C.A. Aerts, E. Verraedt, R. Mellaerts, A. Depla, P. Augustijns, J. Van
Humbeeck, G. Van den Mooter, and J.A. Martens. The Journal of Physical
Chemistry C, 111(36):13404–13409, 2007. 8
[4] M.S. Alexander, M. M. Kiser, T. Culley, J.R. Kern, J. W. Dolan, J. D.
McChesney, J. Zygmunt, and S. J. Bannister. Journal of Chromatography
B, 785(2):253–261, March 2003. 31
[5] N. Andersson, B. Kronberg, R. Corkery, and P. Alberius. Langmuir, 23(3):
1459–1464, 2007. 8
[6] L. Arsov, M. Froelicher, M. Froment, and H.-L. Goff. J. Chim. Phys., (3):
275–279, 1975. 80, 146
[7] M. Balazic, J. Kopac, M.J. Jackson, and W. Ahmed. International Journal
of Nano and Biomaterials, 1(1):3 – 34, 2007. 78
[8] A. Bazzoni. Effet d’un recuit laser sur la luminescence de quantum dots
de Zns:Mn. Semester project, Ecole Polytechnique Federale de Lausanne,
2008. 59
[9] P.A. Beermann, B.R. McGarvey, B.O. Skadtchenko, S. Muralidharan, and
R.W. Sung. Journal of Nanoparticle Research, 8(2):243, April 2006. 59
159
BIBLIOGRAPHY
[10] B. Bhattacharyya, J. Munda, and M. Malapati. International Journal of
Machine Tools and Manufacture, 44(15):1577–1589, December 2004. 2
[11] J. Bibette, F. Leal Calderon, and P. Poulin. Reports on Progress in Physics,
62(6):969–1033, 1999. 12
[12] T.M. Blattler, A. Binkert, M. Zimmermann, M. Textor, J. Voros, and
E. Reimhult. Nanotechnology, 19(7):075301, 2008. 113, 116
[13] C. Buzea, I. Pacheco, and K. Robbie. Biointerphases, 2(4):MR17–MR71,
December 2007. 44
[14] F. Caboi, G. Capuzzi, P. Baglioni, and M. Monduzzi. Journal of Physical
Chemistry B, 101(49):10205–10212, December 1997. 14
[15] M. Chastellain, A. Petri, and H. Hofmann. Journal of Colloid and Interface
Science, 278(2):353–360, October 2004. 16
[16] P. -F. Chauvy. Electrochemical Micromachining of Titanium using Oxide
Film Laser Lithography. PhD, Ecole Polytechnique Federale de Lausanne,
2002. 80, 81, 152
[17] P. -F. Chauvy, P. Hoffmann, and D. Landolt. Applied Surface Science,
208-209:165–170, March 2003. 81
[18] P.-F. Chauvy. Scalecrawler. micropat SA, Lausanne (Switzerland). 90
[19] S. Z. Chu, S. Inoue, K. Wada, S. Hishita, and K. Kurashima. Journal of
The Electrochemical Society, 152(3):B116–B124, March 2005. 82
[20] B. Coasne, J.P. Pikunic, R.J.-M Pellenq, and K.E. Gubbins. Materials
Research Society Symposia Proceedings, 790:53–58, 2003. 36
[21] P. Costa and J.M. Sousa Lobo. European Journal of Pharmaceutical Sci-
ences, 13(2):123–133, May 2001. 38
[22] G. D. Cremer, B.F. Sels, J. Hotta, M.B.J. Roeffaers, E. Bartholomeeusen,
E. Coutino-Gonzalez, V. Valtchev, D. Vos, T. Vosch, and J. Hofkens. Ad-
vanced Materials, 22(9):957–960, 2010. 3
160
BIBLIOGRAPHY
[23] A.B. Cruz, Q. Shen, and T. Toyoda. Materials Science and Engineering:
C, 25(5-8):761–765, December 2005. 60
[24] E. L. Cussler. Diffusion. Cambridge University Press, 1997. 36
[25] G. J. S. Dawes, L. E. Fratila-Apachitei, B. S. Necula, I. Apachitei, G. J.
Witkamp, and J. Duszczyk. 21(1):215–221, 2010. PMID: 19669866 PMCID:
2805798. 2
[26] J.-L. Delplancke and R. Winand. Electrochimica Acta, 33(11):1551–1559,
1988. 131
[27] K. Dittmar. Nanostructured Titania Coatings for Drug-Eluting Medical
Implants. PhD thesis, Ecole Polytechnique Federale de Lausanne, 2009.
30, 42
[28] A. Dyal, K. Loos, M. Noto, S.W. Chang, C. Spagnoli, K.V.P.M. Shafi,
A. Ulman, M. Cowman, and R.A. Gross. Journal of the American Chemical
Society, 125(7):1684–1685, February 2003. 9
[29] D. F. Evans and H Wennerstrom. The colloidal domain. Wiley-VCH Pub-
lishers, 2nd ed. edition, 1999. 12
[30] J.T. Fulton. Biological Vision: A 21st Century Tutorial.
www.neuronresearch.net, Corona Del Mar, CA. USA, 2004. 65
[31] C. Galli, M. Collaud Coen, R. Hauert, V. L. Katanaev, P. Groning, and
L. Schlapbach. Colloids and Surfaces B: Biointerfaces, 26(3):255–267, Oc-
tober 2002. 81
[32] B. Gilbert, F. Huang, Z. Lin, C. Goodell, H. Zhang, and J.F. Banfield.
Nano Letters, 6(4):605–610, April 2006. 60
[33] N. Goswami and P. Sen. Solid State Communications, 132(11):791–794,
December 2004. 60
[34] A. Han, G. Mondin, N.G. Hegelbach, N.F. de Rooij, and U. Staufer. Journal
of Colloid and Interface Science, 293(1):151–157, 2006. 42
161
BIBLIOGRAPHY
[35] H. Hata, S. Saeki, T. Kimura, Y. Sugahara, and K. Kuroda. Chemistry of
Materials, 11(4):1110–1119, April 1999. 31, 36, 38
[36] J.Y. He, Z.L. Zhang, M. Midttun, G. Fonnum, G.I. Modahl, H. Kristiansen,
and K. Redford. Polymer, 49(18):3993–3999, August 2008. 108
[37] T.P. Hoar, D.C. Mears, and G.P. Rothwell. Corrosion Science, 5(4):279–
289, 1965. 85
[38] S. Hu, T. Liu, H. Huang, D. Liu, and S. Chen. Langmuir, 24(1):239–244,
2008. 8
[39] L. Huynh, J. Grant, J.-C. Leroux, P. Delmas, and C. Allen. Pharmaceutical
Research, 25(1):147–157, 2008. 34
[40] G. F. Imbusch. Physical Review, 153(2):326, 1967. 50
[41] G. F. Imbusch. Journal of Luminescence, 53(1-6):465–467, July 1992. 50
[42] Christian Jaeggi, Philippe Kern, Johann Michler, Thomas Zehnder, and
Hans Siegenthaler. Surface and Coatings Technology, 200(5-6):1913–1919,
November 2005. 2, 81
[43] K. L. Johnson, K. Kendall, and A. D. Roberts. Proceedings of the Royal
Society of London. Series A, Mathematical and Physical Sciences (1934-
1990), 324(1558):301–313, 1971. 102
[44] F. Juillerat. Self-assembled nanostructures prepared by colloidal chemistry.
PhD, Ecole Polytechnique Federale de Lausanne, 2006. 109, 111
[45] K. Kambara, N. Shimura, and M. Ogawa. Nippon Seramikkusu Kyokai
Gakujutsu Ronbunshi/Journal of the Ceramic Society of Japan, 115(1341):
315–318, 2007. 26
[46] B.K. Kang, S.K. Chon, S.H. Kim, S.Y. Jeong, M.S. Kim, S.H. Cho, H.S.
Lee, and G. Khang. International Journal of Pharmaceutics, 286(1-2):147–
156, November 2004. 38
162
BIBLIOGRAPHY
[47] J. Karger and D.M. Ruthven. Diffusion in Zeolites and Other Microporous
Solids. John Wiley & Sons, 1st edition, April 1992. 37
[48] P. Kern, Y. Muller, J. Patscheider, and J. Michler. The Journal of Physical
Chemistry B, 110(47):23660–23668, November 2006. 81
[49] P. Kern, J. Veh, and J. Michler. Journal of Micromechanics and Micro-
engineering, 17(6):1168–1177, 2007. 81
[50] Sung Eun Kim, Jae Hoon Lim, Sang Cheon Lee, Sang-Cheol Nam, Hee-
Gyoo Kang, and Jinsub Choi. Electrochimica Acta, 53(14):4846–4851, May
2008. 2, 81
[51] F. Krauss. Synthesis and Characterization of Highly Loaded Superparamag-
netic Beads. Master, Ecole Polytechnique Federale de Lausanne, 2005. 11,
17
[52] H. Kubo, T. Isobe, H. Takahashi, and S. Itoh. Applied Surface Science, 244
(1-4):465–468, May 2005. 60
[53] C.-Y. Lai, B.G. Trewyn, D.M. Jeftinija, K. Jeftinija, S. Xu, S. Jeftinija, and
V.S.-Y. Lin. Journal of the American Chemical Society, 125(15):4451–4459,
April 2003. 8
[54] D. Landolt. Corrosion and surface chemistry of metals. CRC Press, 2007.
79, 155
[55] D. Landolt, P.-F. Chauvy, and O. Zinger. Electrochimica Acta, 48(20-22):
3185–3201, September 2003. 2, 81
[56] A. W. C. Lau, M. Portigliatti, E. Raphael, and L. Leger. EPL (Europhysics
Letters), 60(5):717–723, 2002. 102, 106
[57] F. Leal-Calderon, B. Gerhardi, A. Espert, F. Brossard, V. Alard, J. F.
Tranchant, T. Stora, and J. Bibette. Langmuir, 12(4):872–874, 1996. 13
[58] Y.-G. Lee, C. Oh, S.K. Yoo, S.M. Koo, and S.G. Oh. Microporous and
Mesoporous Materials, 86(1-3):134–144, November 2005. 8
163
BIBLIOGRAPHY
[59] A. Levchenko, G. Li, J. Boerio-Goates, B.F. Woodfield, and A. Navrotsky.
Chemistry of Materials, 18(26):6324–6332, December 2006. 104
[60] Zheng liang Zhi, Yasutaka Morita, Shouhei Yamamura, and Eiichi Tamiya.
Chemical Communications, (19):2448–2450, 2005. 3
[61] K.W. Limbach and J. Wei. AIChE Journal, 36(2):242–248, 1990. 37
[62] Y.S Lin, S.H. Wu, Y. Hung, Y.H. Chou, C. Chang, M.L. Lin, C.P. Tsai,
and C.Y. Mou. Chemistry of Materials, 18(22):5170–5172, October 2006. 8
[63] Y. Liu. Colloids and Surfaces A: Physicochemical and Engineering Aspects,
274(1-3):34–36, February 2006. 34
[64] J. Lu, M.P. Rao, N.C. MacDonald, D. Khang, and T.J. Webster. Acta
Biomaterialia, 4(1):192–201, 2008. 2
[65] J. Lowe, H. Li, K. H. Downing, and E. Nogales. Journal of Molecular
Biology, 313(5):1045–1057, November 2001. 32
[66] W.D. Machin and R.J. Murdey. Langmuir, 12(26):6501–6505, 1996. 28
[67] D. Maugis. Langmuir, 11(2):679–682, February 1995. 102
[68] D. Maugis and H.M. Pollock. Acta Metallurgica, 32(9):1323–1334, Septem-
ber 1984. 104
[69] Y. Mi, G. Xue, and X. Wang. Polymer, 43(25):6701–6705, 2002. 106
[70] O. Mondain-Monval, F. Leal-Calderon, and J. Bibette. Journal de Physique
II, 6(9):1313–1329, 1996. 20
[71] Mu, Gu, and Xu. Applied Physics A: Materials Science & Processing, 80
(7):1425–1429, April 2005. 17
[72] K. Mukherjee, S. P. Moulik, and D. C. Mukherjee. Langmuir, 9(7):1727–
1730, July 1993. 14
[73] Y. Murakami, A. Sawata, Y. Tsuru, and K. Akiyama. Journal of Materials
Science, 38(12):2723–2725, June 2003. 57
164
BIBLIOGRAPHY
[74] T. Nakamura, Y. Yamada, and K. Yano. Journal of Materials Chemistry,
16(25):2417–2419, 2006. 9
[75] T. Neuberger, B. Schopf, H. Hofmann, M. Hofmann, and B. von Rechen-
berg. Journal of Magnetism and Magnetic Materials, 293(1):483–496, May
2005. 9
[76] R.I. Nooney, T. Dhanasekaran, Y. Chen, R. Josephs, and A.E. Ostafin.
Advanced Materials, 14(7):529–532, 2002. 8, 9
[77] E. Palin, H. Liu, and T.J. Webster. Nanotechnology, 16(9):1828–1835, 2005.
2
[78] Q. A. Pankhurst, J. Connolly, S. K. Jones, and J. Dobson. Journal of
Physics D: Applied Physics, 36(13):R167–R181, 2003. 9
[79] G. D Parfitt and Kenneth S. W Sing. page 23. Academic Press, London
a.o, 1976. 28
[80] Y. Park, S. Chung, and K. Row. Korean Journal of Chemical Engineering,
16(3):388–391, May 1999. 34, 35
[81] R. Paschotta. Encyclopedia of Laser Physics and Technology. Wiley-VCH,
Berlin, December 2008. 67
[82] A. Patra, R.E. Tallman, and B.A. Weinstein. Optical Materials, 27(8):
1396–1401, 2005. 50
[83] J. Philibert. Atom Movements: Diffusion and Mass Transport in Solids.
Monographies de physique. Editions de Physique, Les Ulis, France, 1991.
39
[84] A. Pillonnet, C. Garapon, C. Champeaux, C. Bovier, H. Jaffrezic, and
J. Mugnier. Journal of Luminescence, 87-89:1087–1089, May 2000. 51
[85] O. Piotrowski. Polissage electrochimique du titane. PhD thesis, Ecole Poly-
technique Federale de Lausanne, 1999. 83, 84, 97
165
BIBLIOGRAPHY
[86] Sang-Hyun [1] Pyo, Jin-Suk [2] Cho, Ho-Joon [2] Choi, and Byung-Hee [3]
Han. Drying Technology, 25:1759–1767, October 2007. 38
[87] S. C. Qu, W. H. Zhou, F. Q. Liu, N. F. Chen, Z. G. Wang, H. Y. Pan, and
D. P. Yu. Applied Physics Letters, 80(19):3605–3607, May 2002. 59
[88] D. S. Rimai, L. P. DeMejo, and R. C. Bowen. Journal of Applied Physics,
68(12):6234–6240, December 1990. 103, 104, 107
[89] D.S. Rimai, L.P. Demejo, and R.C. Bowen. Journal of Adhesion Science
and Technology, 8:1333–1355, 1994. 103
[90] E. Ruiz-Hernandez, A. Lopez-Noriega, D. Arcos, I. Izquierdo-Barba,
O. Terasaki, and M. Vallet-Regi. Chemistry of Materials, 19(14):3455–3463,
July 2007. 8, 26
[91] S. Sakka and R. M. Almeida. page 202. Springer, 2005. 28
[92] W. Sangthong, M. Probst, and J. Limtrakul. Chemical Engineering Com-
munications, 195(11):1486, 2008. 36
[93] S. Santra, R. Tapec, N. Theodoropoulou, J. Dobson, A. Hebard, and
W. Tan. Langmuir, 17(10):2900–2906, May 2001. 9
[94] J. W. Schultze and M. M. Lohrengel. Electrochimica Acta, 45(15-16):2499–
2513, May 2000. 151
[95] U. D. Schwarz. Journal of Colloid and Interface Science, 261(1):99–106,
May 2003. 102
[96] D. Shao, A. Xia, J. Hu, C. Wang, and W. Yu. Colloids and Surfaces A:
Physicochemical and Engineering Aspects, 322(1-3):61–65, June 2008. 8, 9
[97] R.N. Shimizu and N.R. Demarquette. Journal of Applied Polymer Science,
76(12):1831–1845, 2000. 103
[98] G.M. Spinks, H.R. Brown, and Z. Liu. Polymer Testing, 25(7):868–872,
October 2006. 103
166
BIBLIOGRAPHY
[99] M. Staiger, P. Bowen, J. Ketterer, and J. Bohonek. Journal of dispersion
science and technology, 23(5):619–630, 2002. 45, 47, 48
[100] B. Steitz. Coating and functionalization of nanoparticles for biomedical
applications. PhD thesis, Ecole Polytechnique Federale de Lausanne, 2008.
10, 61
[101] B. Steitz, Y. Axmann, H. Hofmann, and A. Petri-Fink. Journal of Lumi-
nescence, 128(1):92–98, 2008. 61
[102] M. Stuer. Luminescence change of quantum dots by laser annealing.
Semester project, EPFL, Lausanne, 2007. 59
[103] M. Suzuki and J. M. Smith. The Chemical Engineering Journal, 3:256–264,
1972. 38
[104] B. Tamamushi and N. Watanabe. Colloid & Polymer Science, (258):174–
178, 1979. 14
[105] B. Tan and S.E. Rankin. Journal of Non-Crystalline Solids, 352(52-54):
5453–5462, December 2006. 15, 28
[106] K. K. Unger, D. Kumar, M. Grun, G. Buchel, S. Ludtke, Th. Adam,
K. Schumacher, and S. Renker. Journal of Chromatography A, 892(1-2):
47–55, September 2000. 8
[107] W. van Sark, P. Frederix, A. Bol, H.C. Gerritsen, and A. Meijerink.
ChemPhysChem, 3(10):871–879, 2002. 66
[108] S.R. Veith, A. Hughes, and S.E. Pratsinis. Journal of Controlled Release,
99(2):315–327, September 2004. 37
[109] R. L. Vinall, B. Gasser, and R. G. Richards. Injury, 26(Supplement 1):
21–27, 1995. 2
[110] L. Z. Wang, P. C. Ho, H. S. Lee, H. K. Vaddi, Y. W. Chan, and Chan Sui
Yung. Journal of Pharmaceutical and Biomedical Analysis, 31(2):283–289,
February 2003. 31
167
BIBLIOGRAPHY
[111] Y. Wang and F. Caruso. Chemistry of Materials, 17(5):953–961, March
2005. 8
[112] K. Wefers and M. Chanakya. Oxides and hydroxides of aluminium. Num-
ber 19 in Alcoa technical papers. Alcoa Research Laboratories, 1987. 45,
46
[113] R.E. White, B.E. Conway, and C. G. Vayenas. Modern aspects of electro-
chemistry. Springer, 2004. 79
[114] Z.Z. Xu, C.C. Wang, W.L. Yang, and S.K. Fu. Journal of Materials Science,
40(17):4667–4669, 2005. 9
[115] J. Yahalom and J. Zahavi. Electrochimica Acta, 16(5):603–607, May 1971.
80
[116] H. Yang, G. Vovk, N. Coombs, I. Sokolov, and G.A. Ozin. Journal of
Materials Chemistry, 8(3):743–750, 1998. 28
[117] L. Yang, Y. Wang, G. Luo, and Y. Dai. Microporous and Mesoporous
Materials, 94(1-3):269–276, September 2006. 15
[118] P. Yang, M. Lu, D. Xu, D. Yuan, and G. Zhou. Applied Physics A: Materials
Science & Processing, 73(4):455–458, October 2001. 59
[119] P. Yang, Z. Quan, Z. Hou, C. Li, X. Kang, Z. Cheng, and J. Lin. Bioma-
terials, 30(27):4786–4795, September 2009. 8
[120] K. Yano and Y. Fukushima. Journal of Materials Chemistry, 14(10):1579–
1584, 2004. 26, 28
[121] C. Yao and T.J. Webster. Journal of Nanoscience and Nanotechnology, 6:
2682–2692, September . 2
[122] S.L. Yuan, Z.T. Cai, G.Y. Xu, and Y.S. Jiang. Chemical Physics Letters,
365(3-4):347–353, October 2002. 14
[123] X. Zhang, H. Song, L. Yu, T. Wang, R. Xinguang, X. Kong, Y. Xie, and
X. Wang. Journal of Luminescence, 118(2):251–256, June 2006. 60
168
BIBLIOGRAPHY
[124] D. Zhao, J. Sun, Q. Li, and G.D. Stucky. Chemistry of Materials, 12(2):
275–279, February 2000. 9
[125] W. Zhao, J. Gu, L. Zhang, H. Chen, and J. Shi. Journal of the American
Chemical Society, 127(25):8916–8917, June 2005. 8
[126] L.-S. Zhong, J.-S. Hu, H.-P. Liang, A.-M. Cao, W.-G. Song, and L.-J. Wan.
Advanced Materials, 18(18):2426–2431, 2006. 57
[127] S. Zhou, M. Antonietti, and M. Niederberger. Small, 3(5):763–767, 2007.
58
[128] O. Zinger, P.-F. Chauvy, and D. Landolt. Journal of The Electrochemical
Society, 150(11):B495–B503, November 2003. 2, 81
169
Education
2010 PhD thesis, Ecole Polytechnique Fédérale de Lausanne.
“Nanostructured functional particles and surfaces.”
Engineering diploma in material science, Eidgenössische Technische
Hochschule Zürich (ETHZ).
2005
1998 International Baccalaureate and Bilingual Diploma (En-Fr),
International School of Geneva. Graduated with honours.
H obbies and interests
Singing, sailing, badm inton, swim m ing, skiing. Active m em ber of a charity organisation.
A rthur Ganz, m ay 2010.
A rthur G anzChem in de Fontenay 13
1007 Lausanne
078 737 47 19
arthur.ganz@ ep$.ch
French (m other tongue)
English (C2), Germ an (C1)
29 years old, m arried, one child
Swiss and French nationalities
PhD in material science
W ork experience
O ct. 2006
to present
Professor’s assistant, Laboratoire de technologie des poudres, Ecole
Polytechnique Fédérale de Lausanne.
- W riting of m y PhD thesis;
- D evelopm ent of innovative technological solutions for the creation of titanium
im plant surfaces;
- Independent m anagem ent of m ultiple projects;
- Supervision of the work of 3 students;
- Teaching of laboratory courses;
- O ral and poster com m unications at national and international conferences;
I developed a complete technology platform allowing the nanoscale topographic
structuring of surfaces. This was an entirely new �eld of research for the laboratory.
2005-2006 Research engineer, Centre des M atériaux, Ecole des M ines de Paris.
- European project, “Nanodiam onds for biom edical applications”;
- Interactions with European and French research groups;
- In charge of coordination and assessm ent of post-graduate training courses;
Beyond my research activities, I acted as the students’ sole representative in the
committee in charge of selecting external course providers.
2004-2005 Diploma thesis, Institut national pour la santé et la recherche m édicale (INSERM )
unit 706, Paris.
- Characterisation of intercellular adhesion forces using deform able substrates;
- Interdisciplinary research project between bio-physics and biology;
- Training in biological techniques;
I conceived a method for coating functional proteins on substrates in a space resolved
way. The results obtained were published in “Biology of the Cell” (cited 28 times).
A pril-July 2003 Internship (4 m onths), RO LEX SA , Geneva.
- D evelopm ent project at the watch laboratory;
- Tribological study for the optim isation of anodised alum inium ;
I conducted a extensive statistical experiment plan which lead to a better control
of wear phenomena on anodised aluminium.
M arch-July 1999
and O ct.-M arch
2002-2003
A gent, SO S EVA SA N SA , Geneva.
- Travellers’ insurance;
- O rganisation of repatriations and sanitary evacuations;
- Assistance to travellers;
- O % cer in charge at the em ergency call centre;
- Responsible for constituting litigation &les for the legal departm ent;
I handled emergency situations (illnesses, accidents), and provided guidance and
assistance to travellers in di!culty abroad.
Internship (6 weeks), Francia$ex SA , Rennes, France.
- Production of house closures;
- W ork at the production line;
- O ptim isation of inventory m anagem ent;
- Contact with people from di*erent backgrounds;
I implemented a more e!cient inventory management system for accessory parts
using the Kanban method.
M arch-A pril 2000
Independent and interdisciplinary, I aim to apply my expertise in material science and
expand my project management skills in a dynamic research and development team.”
“
Computer and technical skills
M S O % ce, O rigin, Latex, Adobe Illustrator, Im ageJ, CO M SO L m ultiphysics.
Atom ic force m icroscopy, scanning/transm ission electron m icroscopy, optical m icroscopy, particle size
m easurem ent, cell culture.
Education
2010 PhD thesis, Ecole Polytechnique Fédérale de Lausanne.
“N anostructured functional particles and surfaces.”
Engineering diplom a in m aterial science, Eidgenössische Technische
H ochschule Zürich (ETH Z).
2005
1998 International Baccalaureate and Bilingual D iplom a (En-Fr),
International School of G eneva. G raduated w ith honours.
H obbies and interests
Singing, sailing, badm inton, sw im m ing, skiing. Active m em ber of a charity organisation.
A rthur G anz, m ay 2010.
Arthur G anzChem in de Fontenay 13
1007 Lausanne
078 737 47 19
arthur.ganz@ ep$.ch
French (m other tongue)
English (C2), G erm an (C1)
29 years old, m arried, one child
Sw iss and French nationalities
PhD in m aterial science
W ork experience
O ct. 2006
to present
Professor’s assistant, Laboratoire de technologie des poudres, Ecole
Polytechnique Fédérale de Lausanne.
- W riting of m y PhD thesis;
- D evelopm ent of innovative technological solutions for the creation of titanium
im plant surfaces;
- Independent m anagem ent of m ultiple projects;
- Supervision of the w ork of 3 students;
- Teaching of laboratory courses;
- O ral and poster com m unications at national and international conferences;
I developed a complete technology platform allowing the nanoscale topographic
structuring of surfaces. This was an entirely new !eld of research for the laboratory.
2005-2006 Research engineer, Centre des M atériaux, Ecole des M ines de Paris.
- European project, “N anodiam onds for biom edical applications”;
- Interactions w ith European and French research groups;
- In charge of coordination and assessm ent of post-graduate training courses;
Beyond my research activities, I acted as the students’ sole representative in the
committee in charge of selecting external course providers.
2004-2005 D iplom a thesis, Institut national pour la santé et la recherche m édicale (IN SERM )
unit 706, Paris.
- Characterisation of intercellular adhesion forces using deform able substrates;
- Interdisciplinary research project betw een bio-physics and biology;
- Training in biological techniques;
I conceived a method for coating functional proteins on substrates in a space resolved
way. The results obtained were published in “Biology of the Cell” (cited 28 times).
A pril-July 2003 Internship (4 m onths), RO LEX SA , G eneva.
- D evelopm ent project at the w atch laboratory;
- Tribological study for the optim isation of anodised alum inium ;
I conducted a extensive statistical experiment plan which lead to a better control
of wear phenomena on anodised aluminium.
M arch-July 1999
and O ct.-M arch
2002-2003
Agent, SO S EVA SA N SA , G eneva.
- Travellers’ insurance;
- O rganisation of repatriations and sanitary evacuations;
- A ssistance to travellers;
- O % cer in charge at the em ergency call centre;
- Responsible for constituting litigation &les for the legal departm ent;
I handled emergency situations (illnesses, accidents), and provided guidance and
assistance to travellers in di" culty abroad.
Internship (6 w eeks), Francia$ex SA , Rennes, France.
- Production of house closures;
- W ork at the production line;
- O ptim isation of inventory m anagem ent;
- Contact w ith people from di*erent backgrounds;
I implemented a more e" cient inventory management system for accessory parts
using the Kanban method.
M arch-A pril 2000
Independent and interdisciplinary, I aim to apply my expertise in material science and
expand my project management skills in a dynamic research and development team.”
“
Com puter and technical skills
M S O % ce, O rigin, Latex, Adobe Illustrator, Im ageJ, CO M SO L m ultiphysics.
Atom ic force m icroscopy, scanning/transm ission electron m icroscopy, optical m icroscopy, particle size
m easurem ent, cell culture.
top related