Structuration hiérarchique d’assemblage de

nanoparticules magnétiques pour des applications en

tant que capteurs plasmoniques
Mathias Dolci

To cite this version:

Mathias Dolci. Structuration hiérarchique d’assemblage de nanoparticules magnétiques pour des ap-
plications en tant que capteurs plasmoniques. Other. Université de Strasbourg, 2018. English. �NNT :
2018STRAE001�. �tel-01891078v2�

HAL Id: tel-01891078

https://theses.hal.science/tel-01891078v2
Submitted on 12 Oct 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 UNIVERSITÉ DE STRASBOURG

ÉCOLE DOCTORALE de Physique et Chimie-Physique (ED182)

Institut de Physique et Chimie des Matériaux de Strasbourg
(UMR 7504 CNRS – Unistra)

THÈSE
présentée par :

Mathias DOLCI
soutenue le : 15 février 2018

pour obtenir le grade de : Docteur de l’université de Strasbourg

Discipline/ Spécialité : Chimie-Physique des Matériaux

Design of magnetic iron oxide nanoparticle

assemblies supported onto gold thin films
for SPR biosensor applications

THÈSE dirigée par :

M. PICHON Benoît Maitre de conférences, université de Strasbourg
RAPPORTEURS :
Mme GOGLIO Graziella Professeur, université de Bordeaux
Mme SZUNERITS Sabine Professeur, université de Lille

AUTRES MEMBRES DU JURY :

M. CLOAREC Jean-Pierre Professeur, Ecole Centrale de Lyon
M. SCHAAF Pierre Professeur, Université de Strasbourg
 A Nina
pour notre complicité éternelle
 Remerciements

Je profite de ces quelques lignes qui me permettent une liberté personnelle totale pour
énoncer quelque chose que personne n’oublie mais qui est trop souvent tût : une thèse ne se fait pas
seul ! Ainsi j’aimerais remercier toutes les personnes qui ont œuvrés pour ce travail de façon directe
ou indirecte. Sans vous, la suite de ce manuscrit n’aurait pas pu aboutir, j’espère que vous prendrez
conscience, en lisant ces quelques lignes, de l’importance de votre contribution.

Je remercie Stephan Haacke et Pierre Rabu pour m’avoir accueilli respectivement au sein de
l’IPCMS et du DCMI.

Evidemment mon travail de thèse n’aurait pu être mené à bien sans l’aide précieuse de mon
« papa de thèse ». Ces remerciements te sont adressés Benoît. Merci à toi, nous avons parcouru tout
ce chemin ensemble, trois ans c’est long et pourtant j’ai l’impression que tout s’est passé très vite.
Merci pour cette formation, puisqu’au final c’est de cela qu’il s’agit et en effet j’ai appris énormément
avec toi autant d’un point de vue scientifique avec ta rigueur qui te tient à cœur de ne pas avancer
trop vite sans qu’on soit sûr à 101% de ce qu’on fait mais aussi de ton énergie à vouloir faire plein de
nouvelles choses (peut-être même trop de choses), autant d’un point de vue personnel avec tes
paroles de sage-philosophe et tes blagues (un peu moins sages). Dans tous les cas merci, grâce à toi
j’ai passé une excellente thèse et je te remercie de tout cœur. J’espère que l’avenir nous réservera de
belles surprises.

Bien sûr, s’il y a un « papa » à cette thèse, il y a aussi une « maman », une mère protectrice qui
protège ses petits doctorants et n’hésite pas une seconde à s’impliquer s’ils ont le moindre souci.
Sylvie, merci pour ton chaleureux accueil dans l’équipe. Nous avons travaillé « de loin » pour cette
thèse, mais je savais que tu étais là, juste derrière moi et que je pouvais m’appuyer sur toi si besoin. A
toi aussi je te souhaite le meilleur, et à très bientôt je l’espère.

Pour compléter la famille il faut nécessairement que je remercie les membres de l’équipe avec
qui j’ai travaillé. Merci à Damien, le « grand frère » (ou le petit ?), merci pour les discussions de stabilité
colloïdale, de fonctionnalisation de nanoparticules et de Pokémons. Merci également pour le TEM et
le temps que tu y as passé (merci Lina de ne pas avoir mis le feu au microscope). J’aimerais évidemment
remercier mes plus que collègues de chimie 5 avec qui j’ai passé le plus grand de mon temps. Merci à
toi Céline, depuis le début on en a vécu des aventures… grâce à toi je sais me servir de l’infrarouge,
trier les solvants et tirer des flèches sur des cochons. Merci aux autres membres du labo, Geoffrey,
Francis (mon frère), Paula, Kevin, je vous souhaite une bonne continuation pour la suite.

Merci à Cédric, excellent microscopiste et mauvais partenaire de course (ou l’inverse, je ne me

souviens plus), je pense que tu garderas un bon souvenir de moi, en tout cas tu garderas un souvenir…

Merci aux autres membres du départements avec qui j’ai eu des vives et moins vives
discussions, Guillaume, Emilie (tatie), Didier, Silviu, Nathalie, Geneviève, Pierre, Christophe, Anne,
François, Marc et Guido. Merci à vous de m’avoir fait passer un bon séjour au DCMI que je n’oublierais
pas de sitôt.

Merci également à Sophie et Gilles. Le troisième étage était bien plus rigolo avec vous, merci de m’avoir
aidé lorsque j’en avais besoin, merci pour les galettes, petits pains, chasse aux œufs et tout le reste qui
a contribué à la bonne ambiance du troisième.
Je remercie Bobby, Bobby et Bobby, sans vous cette aventure aurait été bien différente. Grâce à vous,
j’ai l’impression d’avoir passé trois ans de franche rigolade et je regrette déjà les discussions au détour
d’un couloir du troisième. Merci Bobby pour m’avoir accompagné entre midi et deux pour rejoindre
notre charmante forêt et revenir au boulot en mode « serpillère », je te souhaite plein de réussite et
j’espère qu’on pourra se revoir incessamment sous peu. Merci Bobby pour toutes les discussions qu’on
ait pu avoir sur des sujets allant de l’eurodance des années 2000 à la politique communiste du bloc
soviétique. Merci également pour ces fameuses soirée crêpes qui ont fortement contribué à
augmenter mon cholestérol et me faire douter d’un éventuellement déclenchement d’un diabète de
type II dans mon organisme. Enfin merci Bob, on en a vécu des choses toi et moi, entre complicité
extrême et rivalité scientifique permettant d’avancer ensemble dans une ambiance détonante. Je ne
pense pas retrouver une ambiance au labo comme celle que nous avons eu, écouter France Gall et
Daniel Balavoine le volume à fond tout en chantant pendant mes manips me manquera j’imagine… Je
ne te dis pas au revoir, mais à bientôt, car nous devons encore faire une collab’ ensemble.

Je tenais à remercier également toutes les personnes extérieures à l’équipe qui m’ont apporté
leur aide pour la réalisation de ce travail de thèse.

Merci à mes collègues « parisien » sans qui je n’aurais pas pu développer toute la partie ‘’SPR’’ de ma
thèse. Merci Greg, pour toute l’aide et les explications qui vont avec sur la résonnance plasmon ainsi
que pour ton enthousiasme et ta bonne humeur. Le jeune padawan est devenu grand maintenant,
j’espère un jour devenir maître ! Merci J-F, pour ton accueil et la patience que tu as eu de m’aider à
faire les mesures pendant la finalisation de ton manuscrit. J’espère que le stress occasionné ne laissera
aucune rancœur à mon égard. Merci Julien pour les très intéressantes discussions et les explications
détaillées que j’ai pu avoir sur la SPR. A tous les trois, merci et bonne continuation, j’espère que nos
chemins scientifiques (et personnels) se recroiseront.

Je tiens à remercier vivement Fouzia et Déborah qui m’ont permis de poursuivre mes études SPR sur
place à Strasbourg. Un bon quart de ma thèse à pu être fait grâce à l’accès au banc SPR de l’ICS, c’est
dire à quel point ce fut important. Merci Fouzia pour cette fructueuse collaboration. Déborah un
immense merci pour m’avoir « sauvé la vie » face à cet appareil SPR, je pense que sans tes
interventions j’aurais abandonné mon dernier chapitre.

Je tenais également à remercier Philippe et Anne pour leur aide et la belle collaboration qui a pu
commencer, même si j’ai l’impression que je quitte l’aventure alors qu’elle n’est encore qu’un
bourgeon. J’espère qu’elle donnera de jolies fleurs et vous souhaite à tous les deux une bonne
continuation.

Un grand merci à Spiros et Vaso. Vous avez été très gentils avec moi, à prendre le temps de me montrer
comment fonctionnait l’XPS, de m’aider à interpréter les données, de me passer des échantillons,
même si la file d’attente s’étalait sur des mois.

Merci à toi Simon pour les mesures PM-IRRAS, ta gentillesse et ta bonne humeur étant toujours au
rendez-vous c’était toujours avec grand plaisir (et un peu de nostalgie) que je me rendais à Mulhouse.

Je remercie Xavier et Didier, bien évidemment pour l’aide informatique et l’ATG, mais aussi pour les
« délicieuses » discussions du bureau qui agrémentait mes journées.

J’aimerais faire une mention spéciale pour le bureau 2006 qui a vu passer pas mal de monde
depuis mon arrivée en tant que petit nouveau et à présent je suis l’ancien qui laisse sa place aux
nouveaux. Merci à Julien, Mathilde, Xiaojie, Mathilde, Zo, Elodie, Florian et Kübra pour avoir parsemé
le bureau d’une ambiance « chaleureuse », pleine d’assiduité, de thé et de belles chansons. Matoutou,
on n’en a passé des bons moments ensemble, la sueur des joueurs de foot devenait de l’eau bénite, je
te souhaite une bonne continuation là où le vent te portera (et voudra de toi). Mathilde, je ne pense
pas trouver un autre bureau que nos éructations puissent embaumer comme le bureau 2006, bonne
continuation à toi aussi. Elodie, je te souhaite plein de réussite, toi aussi tu es une « grande »
maintenant, tu n’as plus besoin de t’appuyer sur les gens, laisse les gens s’appuyer sur toi.

Merci aux gens qui m’ont aidé le long de la route, parfois juste de passage, mais qui m’ont aidé
bien plus qu’ils ne le pensent. Merci à maman Yu qui a su diriger d’une main de fer chimie 5 lorsqu’on
était encore que des « bébés », Aurélie ma première partenaire de course qui m’a sauvé la vie plus
d’une fois. Merci Cristina, sans toi je n’aurais jamais su comment jurer ni même connaître le générique
de DBZ en espagnol. Merci aux personnes du DMO qui m’ont aidé malgré mes (fortes) lacunes en
chimie, merci à Catalina, Mathilde, Audrey et Emilie. Malgré le fait que je détestais faire de la chimie,
c’était toujours avec plaisir que je venais vous demander de l’aide, merci d’avoir été gentilles et
patientes avec moi. Merci à Zo, on s’est rencontrés quand j’étais en stage, après tu es revenu, on a
même partagé un bureau. J’aimerais te remercier pour les échanges qu’on a eu, ce fut très agréable
(pas tout le temps, mais la majorité du temps), je te souhaite une bonne continuation. Merci à Andra,
pour moi tu reste la première roumaine à qui j’ai parlé et regarde, aujourd’hui je parle presque
roumain ! Je te souhaite le meilleur, prends soin de toi. Merci à tous les autres doctorants qui ont
parsemé ma thèse de bons moments, merci à Stas, Ivan, Julien, Beata, Jérôme, Ouda, Lydia, Juliane,
Melania, Etienne, Michal, Guillaume, Florian et Ufuk.

Je remercie mes amis de master Ziyad, Kassioge et Kübra. Grâce à vous j’ai pu passer une
merveilleuse expérience humaine et je ne saurai jamais assez vous remercier. Kübra, nous t’attendons
avec impatience, dépêche-toi de devenir docteur ! Je voulais te remercier pour les pauses thé, les
discussions personnelles (t’étais un peu mon psy), et toutes les belles choses que tu as faites pour moi.
Ziyad je ne sais même pas ce que je peux te souhaiter, car je sais que tu vas réussir ce que tu
entreprendras, merci pour m’avoir fait réfléchir, questionner à chaque paroles (check). Kassio merci à
toi également, je te souhaite plein de bonne choses, ta gentillesse est au-delà de tout ce que j’imaginais
avant de te rencontrer, prends soin de toi.

Je remercie enfin les stagiaires que j’ai pu encadrer et qui m’ont aidé dans cette aventure.
Merci à Emelyne, Solène, Eduard, Tom, Romain, Sorina, Kévin, Yu-ting et Pierre qui prend la relève.
Mention spéciale à Mégane (Musch Musch), merci à toi, je te félicite d’avoir réussi à allier rigueur
scientifique et franche rigolade…on s’est bien marré. Je te souhaite plein de réussite. Merci également
à tous les stagiaires qu’ils soient d’ici ou d’ailleurs pour le temps passer à apprendre et jouer au tarot.
Merci à l’équipe de Roumanie : Petronela, Mihail, Antonia, Roxana, Andra, Laura, Adelina et Valentina.
Je remercie pêle-mêle Pedro, Maria, Connor, Maëlle, Clémence, Frédérique, Camille, Salma, Debora,
Camille, Wenja et tous ceux que j’ai oublié. Je remercie également l’équipe brésilienne Antonia,
Camilla, Isis, Pedro et Hermano pour leur accueille chaleureux qui me donne fortement envie de
retourner à Brasilia.

Je n’oublie bien sûr pas les footeux, merci à vous les gars pour les moments de détente
(toujours) dans la bonne ambiance. Merci à Christophe, Rémi, Micka, Gad, Sylvain, Laure, Anna,
Marvin, Michel, Vincent, Tom, Maria, et bien sûr tous ceux que j’ai oublié en cours de route.

Pour finir j’aimerai remercier les personnes totalement extérieures à ce travail de thèse, mais
qui pourtant était là quand j’en avais besoin et qui m’ont apporté le soutien moral qui m’a permis
d’arriver au bout. Merci à mes parents, Nina ma sœur et Valentina ma compagne pour m’avoir soutenu
jusqu’au bout même dans les moments les plus durs.

A vous tous, merci.

I. CHAPITRE I - State of Art ................................................................................................................. 5
A. Fe3O4 Nanoparticles ..................................................................................................................... 6
1. Iron oxide properties in bulk ................................................................................................... 6
a) Crystalline structure ............................................................................................................ 6
b) Magnetic properties ............................................................................................................ 7
c) Electronic properties ........................................................................................................... 8
d) Optical properties ................................................................................................................ 8
2. Iron oxide nanoparticles synthesis ........................................................................................ 10
a) Main synthesis method ..................................................................................................... 10
b) Thermal decomposition synthesis ..................................................................................... 10
c) Composition as function of the size of the nanoparticle .................................................. 12
B. Nanoparticles assembly............................................................................................................. 13
1. Nanoparticle assembly methods ........................................................................................... 13
a) Assembly through substrate functionalization ................................................................. 14
(1) Self-assembled monolayer of organic molecules ...................................................... 14
(2) Mixed self-assembled monolayer ............................................................................. 15
b) Nanoparticle assembly through chelating interactions .................................................... 16
c) Nanoparticle assembly through covalent binding............................................................. 17
d) Nanoparticle assembly through multiple hydrogen binds ................................................ 18
2. Magnetic and optical properties of nanoparticles ................................................................ 20
a) Magnetic properties of nanoparticles ............................................................................... 20
(1) Magnetic properties of a single nanoparticle ........................................................... 20
(2) Magnetization dynamics ........................................................................................... 22
(3) Surface effect............................................................................................................. 23
b) Magnetic properties of nanoparticle assembly................................................................. 24
(1) Dipolar interactions ................................................................................................... 24
(2) Influence of the inter-particle distance ..................................................................... 24
C. Bio-sensors ................................................................................................................................ 26
1. Generalities of biosensors ..................................................................................................... 26
2. Surface plasmons .................................................................................................................. 26
a) Historical context............................................................................................................... 26
b) Theory................................................................................................................................ 27
c) Excitation condition of surface plasmons with Kretschmann configuration .................... 28
 d) Influence of parameters on plasmons excitation .............................................................. 30
(1) Choice of the material ............................................................................................... 30
(2) Metal permittivity...................................................................................................... 31
(3) Temperature and metal thickness............................................................................. 32
(4) Refractive index at the surface .................................................................................. 33
(5) Penetration depth ..................................................................................................... 33
3. Plasmonic biosensors ............................................................................................................ 35
a) Generalities on plasmonics biosensors ............................................................................. 35
b) Localized surface plasmon resonance sensors (LSPR) ....................................................... 36
(1) Colorimetric assays .................................................................................................... 38
(2) Surface nanostructured architecture ........................................................................ 38
c) Propagative surface plasmon resonance (SPR) ................................................................. 39
4. Plasmonics biosensors optimization ..................................................................................... 40
a) Intrinsic properties ............................................................................................................ 40
b) Surface functionalization ................................................................................................... 43
5. Iron oxide nanoparticles based biosensors ........................................................................... 44
D. Bibliography............................................................................................................................... 47
II. CHAPITRE II - Structuration of Iron Oxide Nanoparticle Assemblies ............................................. 55
A. Nanoparticle synthesis .............................................................................................................. 56
1. Experimental details .............................................................................................................. 56
a) Nanoparticles synthesis conditions ................................................................................... 56
b) Purification of the nanoparticles ....................................................................................... 58
2. Structural characterizations .................................................................................................. 61
a) Size and morphology of the nanoparticles ........................................................................ 61
b) Colloidal stability of nanoparticle suspension ................................................................... 62
c) Structural characterization of the nanoparticles .............................................................. 63
3. Magnetic characterizations ................................................................................................... 64
a) Magnetic measurements as function of the applied field ................................................ 64
b) Magnetic measurements as function of the temperature ................................................ 66
4. Conclusion ............................................................................................................................. 68
B. Nanoparticle and substrate functionalization ........................................................................... 69
1. Nanoparticle functionalization .............................................................................................. 69
a) Experimental section ......................................................................................................... 69
b) Characterization after functionalization ........................................................................... 70
 2. Substrate functionalization ................................................................................................... 72
a) Synthesis of 11-mercapto-undecyn................................................................................... 73
b) Self-assembled monolayer formation ............................................................................... 74
c) Characterization of self-assembled monolayers ............................................................... 75
3. Conclusion ............................................................................................................................. 76
C. Nanoparticle assembly prepared by “click” chemistry ............................................................. 78
1. Assembly of 10 nm nanoparticles ......................................................................................... 78
a) Assembly protocol ............................................................................................................. 78
b) Scanning electron microscopy characterization................................................................ 79
2. Study of the kinetics of the assembly.................................................................................... 80
3. Variation of assembly parameters ........................................................................................ 82
a) Nanoparticle sizes.............................................................................................................. 82
b) Concentration of the nanoparticle suspensions ............................................................... 85
c) Stability of the assembling solution in time ...................................................................... 87
4. Conclusions............................................................................................................................ 89
D. Characterization of the nanoparticle assemblies ...................................................................... 90
1. Composition and structural characterizations ...................................................................... 90
a) Atomic force microscopy ................................................................................................... 90
b) Water contact angle .......................................................................................................... 90
c) Polarized Modulated Infrared Reflection Absorption Spectroscopy ................................ 91
d) X-Ray Photoelectron Spectroscopy ................................................................................... 92
2. Magnetics characterization ................................................................................................... 93
a) Cycles of magnetization as function of external field ....................................................... 93
b) Magnetic measurements as function of the temperature ................................................ 97
3. Conclusion ............................................................................................................................. 99
E. General conclusion .................................................................................................................. 100
F. Bibliography............................................................................................................................. 101
III. CHAPITRE III –Nanoparticle self-assembly by multiple hydrogen binding between nucleosides
103
A. Nanoparticle assembly through nucleosides .......................................................................... 104
1. Substrate post-functionalization ......................................................................................... 104
a) Experimental procedure .................................................................................................. 105
b) Characterization .............................................................................................................. 106
(1) Water contact angle ................................................................................................ 106
 (2) X-Ray photoelectron spectroscopy ......................................................................... 107
(3) Phase Modulated Infrared reflection absorption spectroscopy ............................. 107
2. Nanoparticle post-functionalization.................................................................................... 108
a) Control of the quantity of azide groups at the surface ................................................... 108
(1) FT-IR spectroscopy measurements ......................................................................... 109
(2) DLS measurements .................................................................................................. 110
b) Post-functionalization with thymine groups ................................................................... 111
c) Characterization of the NP@Thym.................................................................................. 111
(1) TEM analysis ............................................................................................................ 111
(2) FT-IR measurements................................................................................................ 112
(3) DLS measurements .................................................................................................. 113
3. Molecular recognition through nucleosides ....................................................................... 115
a) Experimental protocol ..................................................................................................... 115
b) Influence of experimental parameters............................................................................ 116
(1) Assembly time ......................................................................................................... 116
(2) Influence of the quantity of functional groups at the nanoparticle and SAM surfaces
117
(a) On the SAM surface ............................................................................................. 117
(b) On the nanoparticle surface ................................................................................ 118
(3) Solvent influence ..................................................................................................... 119
c) Influence of the temperature .......................................................................................... 122
d) Influence of the rinsing step ............................................................................................ 122
B. Conclusion ............................................................................................................................... 123
C. Bibliography............................................................................................................................. 124
CHAPITRE IV - Biomolecules Recognition mediated by iron oxide nanoparticle assemblies supported
onto gold thin films ............................................................................................................................. 125
D. Biotin-Streptavidine ................................................................................................................ 128
1. Design of the bio-platform .................................................................................................. 128
a) Grafting of an alkyne derivative of biotin onto nanoparticle assembly .......................... 128
b) Characterizations ............................................................................................................. 129
(1) Microscopy characterizations.................................................................................. 129
(2) Water Contact Angle ............................................................................................... 130
(3) X-Ray Photoelectron Spectroscopy ......................................................................... 131
(4) Phase Modulation Infrared Reflection Absorption Spectroscopy measurements.. 132
 c) Monitoring of the bio-platform build-up by surface plasmon resonance measurements
134
d) Performing SPR measurements to determine nanoparticle density .............................. 136
2. Detection of streptavidin .................................................................................................... 139
a) Influence of nanoparticle assembly on the detection..................................................... 139
(1) Detection of streptavidin by performing angular interrogation ............................. 140
(2) Detection of streptavidin by performing spectral interrogation ............................ 141
b) Understanding the role of nanoparticle assembly on the detection .............................. 142
(a) Decay length ........................................................................................................ 143
(b) Sensitivity factor .................................................................................................. 143
(c) Available surface area ......................................................................................... 146
3. Detection platform optimization......................................................................................... 150
a) Influence of the spatial distribution of biotin groups at the nanoparticle surface ......... 150
(1) Preliminary study: influence of the accessibility of the biotin groups on a gold
surface151
(a) SEM analysis ........................................................................................................ 152
(b) PM-IRRAS measurements .................................................................................... 152
(c) SPR measurements .............................................................................................. 154
(2) Variation of the size of the nanoparticles ............................................................... 155
4. Kinetics measurements ....................................................................................................... 159
(1) Kinetic measurement and determination of the limit of detection of the Streptavidin
160
(2) Determination of the association constant by the coefficient rate method .......... 162
(3) Determination of the association constant by the Langmuir binding isotherm ..... 164
5. Biofouling............................................................................................................................. 165
6. Conclusion ........................................................................................................................... 167
E. Iminosugar – Glycosidase ........................................................................................................ 168
1. Context and motivation....................................................................................................... 168
2. Substrate post functionalization by iminosugar.................................................................. 168
a) Experimental protocol ..................................................................................................... 168
b) Characterizations ............................................................................................................. 169
(1) Scanning electron microscopy ................................................................................. 169
(2) Water contact angle ................................................................................................ 170
(3) XPS ........................................................................................................................... 170
(4) PM-IRRAS ................................................................................................................. 171
 (5) Construction of the film followed by Surface Plasmon Resonance Measurements 172
3. SPR measurements for the glycosidase .............................................................................. 173
a) α-Mannosidase solution preparation .............................................................................. 173
b) SPR measurements .......................................................................................................... 174
c) Kinetics measurements ................................................................................................... 176
(1) Sensorgrams ............................................................................................................ 176
(2) Determination of the rate constant affinity of the iminosugar and α-mannosidase
couple 177
d) Competitive association .................................................................................................. 179
e) Regeneration of the surface ............................................................................................ 181
4. Biofouling............................................................................................................................. 183
a) Control experiment: α-mannosidase on NPs@N3 surface .............................................. 183
b) Control experiment: BSA on iminosugar surface ............................................................ 184
5. Conclusion ........................................................................................................................... 185
F. General conclusion .................................................................................................................. 186
H. Bibliography............................................................................................................................. 187
Introduction

1
 Iron oxide nanoparticle assemblies represent a high interest in the development of nano-
devices for application in spintronics such as high data storage, magneto-resistive devices and sensors.
Iron oxide is a very well known, low-cost and abundant material which presents advantageous
magnetic and optical properties. The main interest in nanoparticles comes from their properties
resulting from their high surface/volume ratio in comparison to the bulk state.

Therefore, the design and synthesis of iron oxide nanoparticles became a hot-topic owing to
their intrinsic properties which directly dependent of their shape, size and composition. During the last
decades, synthesis techniques considerably evolved and the current challenges are today to control
the formation of more complex structures such as new morphologies or core/shell composition. In this
context, the thermal decomposition technique opened wide perspectives towards the fine design of
nanoparticles by adapting carefully the operating conditions.

Although nanoparticles can be easily studied as powders, the preparation of highly ordered
assemblies onto surfaces is a promising way to investigate their individual versus collective properties.
Therefore, a main challenge is to develop new and efficient strategies to finely control the spatial
arrangement of such nano-building blocks. Therefore, in the frame of the bottom-up approach, highly
stable colloidal suspensions are required.

The control of the assembly of nanoparticles can be easily performed through specific
interactions between functional groups localized at both nanoparticle and substrate surfaces. In this
context, the “click” chemistry approach is very efficient and easy to process. The “click” chemistry
allows controlling the assembly through robust and covalent binding. Another interesting way is
multiple hydrogen bonds which are also highly specific to control the assembly of nanoparticles.

Then, the control of the spatial arrangement allows studying the physical properties of the
nanoparticles such as the optical or magnetic properties. Moreover, the properties of the substrate
which supports the nanoparticles can also be controlled. Indeed, plasmonic materials such as gold thin
films present surface plasmon resonance which can be tuned as function of the structuration of
nanoparticle assembly. These plasmonic properties allow performing biosensing to detect different
species in solution. Therefore, the nanoparticle assemblies supported onto plasmonic substrates are
very appealing for biosensing.

This manuscript will be divided in 4 sections:

· The first chapter presents the state of art related to the synthesis and assembly of iron oxide
nanoparticles and their structural, magnetic and optical properties. Two assembling methods
will be presented: the covalent linkage of the nanoparticles to the surface by the CuAAC “click”
chemistry and the reversible boundary through multiple hydrogen bonds between
nanoparticles and substrates. Moreover, if focuses on the way surface plasmon resonance
present a particular interest for biosensing applications.

· The second chapter is dedicated to the elaboration of iron oxide nanoparticles by thermal
decomposition and their assembly onto a gold substrate thanks to the CuAAC “click” chemistry
reaction. The structural properties of the nanoparticles and their assemblies were
studied. The collective magnetic properties were studied as function of the size and the spatial
arrangement of the nanoparticles.

2
· The third chapter focuses on the assembly of nanoparticles via multiple hydrogen bonds. We
took advantage on the “click” reaction to graft complementary nucleosides groups on the
surface of the nanoparticles and the substrates. The mechanism of the assembly is then
studied as function of different assembling parameters such as solvent, temperature and
quantity of functional groups.

· Finally, the fourth chapter is dedicated to the design of a new and versatile approach to
produce SPR biosensors. We took advantage of the nanoparticle assemblies presented in
chapter 2 as an original detection platform which can be functionalized by specific
bioreceptors. We show how iron oxide nanoparticles can be of interest to enhance the
sensitivity of a gold thin film. We present the different parameters in order to design the most
efficient detection platform. We studied the kinetics of two different biomolecular recognition
processes which are among the most popular in biomedicine: the detection of proteins and
enzymes.

3
4
I. CHAPITRE I - State of Art

5
 A. Fe3O4 Nanoparticles

1. Iron oxide properties in bulk

a) Crystalline structure

Magnetite (Fe3O4) exhibits a spinel structure AB2X4 where A and B represent the cations and X
the anions.[1] It crystallizes in a cubic system (Fd3m space group) with a lattice parameter of 0.8394 nm.
The oxygen atoms form a face centered cubic (FCC) arrangement with octahedral (Oh) and tetrahedral
(Td) sites where the cations can be inserted (figure I-1). A unit cell is composed of 32 oxygen atoms
which define 64 Td sites which one eighth are occupied (A site) and 32 Oh sites which half are occupied
(B site). Magnetite is a mixed oxide constituted of Fe2+ and Fe3+ in an reverse spinel
(Fe3+)Td(Fe3+Fe2+)Oh(O2-)4 where the tetrahedral sites are occupied by Fe3+ and octahedral sites occupied
by Fe3+ and Fe2+.

Oxygen
2+ 3+
(Fe , Fe ) octahedral site (B site)
3+
Fe tetrahedral site (A site)

Site A

Site B

Figure I-1. Schematic representation of the magnetite structure

Maghemite (γ-Fe2O3) is the oxidize form of the magnetite and in this case the iron cations are
all trivalent Fe3+ which induce vacancies in the structure noted □. The resulting structure of maghemite
is (Fe3+)Td[(Fe3+)5/3□1/3]Oh(O2-)4. The structures of magnetite and maghemite are very similar and it
requires different characterization techniques to differentiate them. X-Ray diffraction (XRD)
measurements combined to pattern refinement allow determining the lattice parameter which is
slightly different between magnetite (8.396 ± 0.001 Å, JCPDS file n°00-019-0629) and maghemite
(8.338 ± 0.001 Å, JCPDS file n°00-013-0458).

6
 b) Magnetic properties

Magnetic properties in a material are defined by the magnetic moments carried by each atom.
Therefore, the magnetization of a material for a given volume is equal to the sum of all magnetic
moments which constitute the material. Applying a magnetic field to the material allows aligning these
magnetic moments along the field. Therefore, the susceptibility of the material corresponds to the
ratio of its magnetization M versus the applied field H:

=
!
"
Equation I.1

Two classes of material can be distinguished as function of their behavior when an external
magnetic field is applied.
The first class corresponds to weak interactions between the atoms. Therefore, the material
does not present a spontaneous magnetization. This non-cooperative magnetism is composed of:
· Diamagnetism, where the atoms do not present magnetic moments without external field
and in presence of magnetic field will create weak dipoles with moments in opposition of
the field. The susceptibility is therefore negative and low.
· Paramagnetism, where the magnetic moments are randomly oriented without magnetic
field. The presence of an external magnetic field will align the moments in the direction of
the field. The susceptibility of these materials is therefore positive but low.

The second class presents exchange interactions between their magnetic moments. In this
case of cooperative magnetism, a spontaneous order appears without external magnetic field:

· Ferromagnetism (FM), where the magnetic moments are parallel between them and
present a long range order without magnetic field, e.g. permanent magnetization.
· Anti-ferromagnetism (AFM), where the magnetic moments have an anti-parallel
configuration and compensate, e.g. no macroscopic magnetization.
· Ferrimagnetism (FiM), where magnetic moments have an anti-parallel configuration but
do not totally compensate, e.g. lower permanent magnetization than FM.

This magnetic order exists below a temperature which is called the Curie temperature (TC) for
ferromagnetic and ferrimagnetic materials or Néel temperature (TN) for anti-ferromagnetic materials.
Above this temperature, the spontaneous order disappears and the material becomes paramagnetic.

Magnetite is a ferrimagnetic material. The Fe3+ ions have a moment of 5 μB and the Fe2+ have
a moment of 4 μB. Magnetic order in reverse spinel is driven by the exchange interactions between A
and B sites. Magnetic moments of Fe3+ ions in Oh and Td sites are aligned but their antiparallel
configuration result in their compensation. Therefore, the net magnetization of 4 μB only accounts
from the Fe2+ in Oh sites.

Octahedral sites: Fe3+ (5 μB) ⇑⇑⇑⇑⇑⇑⇑⇑ Fe2+ (4 μB) ↑↑↑↑↑↑↑↑

Tetrahedral sites: Fe3+ (-5 μB) ⇓⇓⇓⇓⇓⇓⇓⇓

7
 According to cell multiplicity, the magnetic moment per unit cell is 32 μB.[2] In bulk state, the
saturation magnetization for magnetite is 92 emu/g. In the case of maghemite, one Fe 3+ is located in
octahedral sites and 5/3 in tetrahedral sites which lead to a theoretical magnetization of 10/3 μB.[3]
This corresponds to a lower saturation magnetization of 74 emu/g.

c) Electronic properties

Magnetite is considered as a half-metal and has a high conductivity at 300 K for an oxide (200-
300 Ω .cm-1) which is attributed to electron hopping between Fe2+ and Fe3+. However, magnetite has
-1

many conductive regimes as function of the temperature which does not follow any law. At room
temperature, the magnetite has a band gap around 0.1 eV.[4, 5] The maghemite, instead has the
behavior of a semi-conductor with a 2.03 eV band gap.[6]

d) Optical properties

The optical properties of iron oxide are directly dependent on its refractive index. The
refractive index n of a material corresponds to the ratio of light celerity in vacuum c compared to the
phase velocity of the light v in the material:

&=(
'
Equation I.2

Maxwell’s equations predict that celerity c of an electromagnetic wave (such as light) which
propagates through vacuum is related to the dielectric permittivity ε0 and the magnetic
permeability μ0 as:

)=
*
+,- .-
Equation I.3

Therefore in a material with a dielectric permittivity ε and magnetic permeability μ, the phase
velocity of light is:

/=
*
√,.
Equation I.4

Given the equation I.2, the refractive index of a material is defined as:

& = √12 32 Equation I.5

with 12 = the relative permittivity and 32 =

, .
,- .-
the relative permeability of the material. The
magnetic permeability is directly linked with the magnetic susceptibility of the material by:

32 = 1 + Equation I.6

In the case of diamagnetic or paramagnetic materials, the susceptibility is very low (< 10-5)
which reduces the permeability to 1. For magnetite, at room temperature, the susceptibility still
remains very (χ ≈ 10-3). [7-10] The approximation of μr = 1 can be done and the refractive index will only
be dependent of the relative permittivity of the material:

8
 & = √12 Equation I.7

Figure I-2. Evolution of the refractive index as function of the incident wavelength in magnetite. From Query et al.[11]

The refractive index is energy dependent and the relation of n and the wavelength of the
illuminating light have been given by Query in 1985 (Figure I-2).[11] In our study, we will consider the
domain of visible light, in the range 400-800 nm. The refractive index of iron oxide oscillates between
2.46 and 2.28.

9
 2. Iron oxide nanoparticles synthesis

The development of new chemical synthesis strategies recently allowed unprecedented control on
size, shape, composition and structure of nanoparticles.[12-15] Furthermore, highly stable suspensions
of nanoparticles can be prepared which is important for anyone who wants to consider isolated
nanoparticles. Therefore, unprecedented control of physical properties of nanoparticles, especially
new and unexpected physical properties can be achieved.

a) Main synthesis method

Many techniques have been developed to synthesize iron oxide nanoparticles which present
different advantages and drawbacks.[16-19] The main purpose of these methods is to improve the
control of the size and the shape of the nanoparticles such as micro-emulsion synthesis[20],
hydrothermal synthesis[21] and polyol synthesis[22].

The co-precipitation synthesis remains the most used method since it is the easiest to perform;
the solvent used is usually water and conduct to high yields. Nanoparticles are formed with the
addition of a base in acid aqueous solution of iron salts Fe2+ and Fe3+. High quantities of nanoparticles
in water with size range from 2 nm to 50 nm can be synthesized. However, the narrow size distribution
is difficult to control and the nanoparticles tend to aggregate which is not suitable for further control
of the arrangement of the nanoparticles.

b) Thermal decomposition synthesis

The thermal decomposition allows preparing nanoparticles with a high control on the size and
morphology and a good stability in suspension. Therefore, this synthesis method has been chosen for
our work.

The thermal decomposition has been initially developed for the synthesis of quantum-dots and
semi-conductor nanocrystals with a high control on the size and the morphology.[23, 24] Later, this
method has been used to form metal oxide nanoparticles. It consists in the decomposition of a metal
precursor in a high boiling temperature organic solvent in presence of stabilizing agent. This method
promotes the formation of well crystallized nanoparticles functionalized with organic molecules thus
providing high colloidal stability in organic solvent.[18, 25]

The synthesis process is driven by the nucleation and growth mechanisms. LaMer theory
describes that the nanoparticles formation consists in three phases (figure I-3).[26] Phase 1 corresponds
to pre-nucleation step, where the metallic precursor decomposes in monomers which are the base
units of the iron oxide nanocrystals. After the increase of the monomer concentration above
supersaturation level (CS), a rapid self-nucleation happens after reaching a maximal concentration
(Cmax) (phase 2). When the monomer concentration goes down below a critical concentration (Cmin),
the self-nucleation stops and the monomers contributes to nanoparticle growth (phase 3).

10
 Figure I-3. Precursor concentration as function of the reaction time. From LaMer V. K et al.[26]

This method presents the advantage to separate the self-nucleation and growth steps. The fine
control of experimental parameters, especially the temperature rate, results in a narrow size
distribution of the nanoparticles. Moreover, the size and the morphology can be tuned by varying the
synthesis parameters such as the solvent, the temperature ramp, the reaction time, the precursor
concentration and the nature of ligand (figure I-4).[27, 28] During the synthesis the nanoparticles are
covered with a stabilizer, the oleic acid which acknowledges the good stability of nanoparticle
suspension in a variety of organic solvents and allows easily performing ligand exchange for further
processes.

Figure I-4. TEM images of iron oxide nanoparticles synthesized with different solvents. The sizes of the nanoparticles are
ranging from 2.5 nm to 14 nm. From Demortière et al.[28]

11
 The investigation on the commercial precursor low purity which altered significantly
reproducibility led our team to develop an in-house precursor.[29] This synthesis method allowed
developing the nanoparticle design to form complex architectures such as nanoparticles with different
compositions or Fe3O4@CoO core/shell structures.[30]

c) Composition as function of the size of the nanoparticle

As seen previously, the surface of magnetite is partially oxidized in maghemite. Therefore,

nanoparticles display an oxidized layer at their surface which consists in a thickness of few
nanometers.[27] Therefore, the proportion of maghemite is more important for small nanoparticles
with a larger surface/volume ratio.[31] It becomes obvious that γ-Fe2O3 is the dominant phase in small
iron oxide nanoparticles, whereas the proportion of the Fe3O4 gradually increases with the particle size
(figure I-5.a).

a) b)

Wavenumber (cm-1)

Figure I-5. a) Magnetite proportion as function of the nanoparticle size in (γ-Fe2O4)1-x@(Fe3O4) and b) Infrared spectra of
magnetite (black curve) and maghemite (red curve) in the Fe-O region. Extracted from a) data of Park et al.[31] and b)
Baaziz et al.[27]

Moreover, infrared spectroscopy showed that magnetite and maghemite present different
signatures. The magnetite spectrum shows a unique band at 570-580 cm-1 whereas the maghemite
exhibits many bands localized in the area 400-800 cm-1 (figure I-5.b).[32]

The characterizations by infrared spectroscopy and X-ray diffraction allow identifying the
presence of the maghemite phase in the nanoparticles. However, it is not sufficient to determine
precisely the stoichiometry of maghemite in the structure. The Mossbauer spectroscopy is an accurate
technique to determine the oxidation state of iron atoms.[33]

The synthesis of iron oxide nanoparticles in the Fe3O4 phase implies necessarily a layer of γ-
Fe2O3. This partial oxidation is not predominant in the case of large nanoparticles (> 10 nm). However,
this composition may impact the intrinsic properties of the nanoparticles.

12
 B. Nanoparticles assembly

The physical properties of nanoparticles are also dependent of their close environment.
Therefore, interactions with neighbor are of particular interest. The control of the spatial arrangement
of nanoparticles (control of the inter-particle distance and the dimensionality of assembly) represents
great challenges. Many assembly techniques can be used to graft nanoparticles onto surfaces. The
structuration of these assemblies and the linkage with the surface will be dependent of the method
used.

1. Nanoparticle assembly methods

The assembly of nanoparticles onto a surface can be controlled through specific interactions
between functional groups. Three approaches can be considered (figure I-6):

Figure I-6. Schematic representation of the assembly strategies: a) functionalization of the nanoparticles, b)
functionalization of the surface and c) functionalization of both nanoparticles and substrate. From Bellido et al. [34]

· The first approach consists in the assembly of functionalized nanoparticles where the
functional groups directly interact with the substrate surface. Many techniques have been
used to assemble nanoparticles onto substrates such as drop casting[34], dipping[34], spin
coating, dip coating, Langmuir-Blodgett[2, 35-37] and layer-by layer.[38] These techniques
allow easily assembling nanoparticles in mono or multilayers, but suffer from a lack of the
structuration control.
· The second approach consists in the substrate functionalization with molecules carrying
functional groups which interact with the nanoparticle surface.
· The third approach is a combination of both methods by a functionalization of nanoparticle
and substrate surface. The considerable advantage of this method is the specificity which
allows a good control of the assembly structuration.

This last method of assembly will be the most interesting in our case since the high specificity
will allow the control of the structuration of the nanoparticle assembly. The main advantages in the

13
surface functionalization of substrate for nanoparticle grafting are the easiness to process by simple
dipping of substrate in nanoparticle suspensions and the high versatility.

Therefore, the molecular patterning onto surface will allow controlling the spatial distribution
of the grafted nanoparticles. Thus, it requires the functionalization of substrates by molecules carrying
specific functional groups.

a) Assembly through substrate functionalization

(1) Self-assembled monolayer of organic molecules

Self-assembled monolayers (SAM) of organic molecules are an easy and versatile method to
functionalize and to modulate the surface properties of a substrate.[39-45] Adsorbed molecules onto a
substrate spontaneously self-organize thanks to the Van der Waals and hydrophobic interactions
between the molecules (figure I-7).[45, 46]

Figure I-7. Schematic representation of the self-assembled monolayer formation. From Sugimura, Kyoto.

The molecules are composed of an anchoring group, a spacer and a terminal functional groups:

· The anchor group depends of the substrate surface. In the case of SiO2 surface,
trialkoxysilanes are mainly used.[47-49] For metal oxides, phosphonic acids are preferred.[49-
51]
Thiol groups are the most used in the case of gold surface[52-54] among others such as
thioacetate, di-sulfurs or di-azonium salts. In our case, gold substrates will be used with
thiols molecules because of the strong interaction with gold[52, 53, 55] which is described as
an intermediate between a covalent and non-covalent binding, about 50 kcal/mol.[39]
· The spacer constitutes a physical barrier between the anchoring groups and the terminal
head groups and ensures a thickness of the SAM. The size of the spacer is crucial since the
anchoring groups is rapidly grafted on the substrate surface and the molecules process in
a slow rearrangement in all-trans configuration to maximize the interactions between
chains.[46, 56] Alkyl chains are mostly used to maximize the packing between chains, but
other types of chains with phenylene or ethylene glycol can be used.[57-59]

14
 · The functional head group represents the essential part of the SAM since it defines the
interface and the surface properties. The choice of the functional group will depend of the
target interactions.

In our study, the substrate is a gold thin film supported on a silica wafer since this metal is inert
and its functionalization is well described. Moreover, this metal presents plasmonic properties from
which we will take advantage.

(2) Mixed self-assembled monolayer

The surface functionalization can be modulated by mixing different molecules (with different
terminal functional groups and/or different chain lengths) in the SAM. Here, the objective is to increase
the reactivity by spacing the active with inactive head groups. Thus, the SAM structuration will depend
of the phase segregation phenomenon leading to enriched domains with one of the thiols and thus
influences the spatial repartition of the nanoparticles.[60]

Few methods can be performed in order to generate mixed SAM (figure I-8). A first one consists
in the immersion in a solution of different thiols. A second consists in the preparation of a SAM with a
single thiol and then to perform a partial chemical reaction in order to generate two different terminal
groups.[61] A third one which consists to prepare a SAM with an asymmetric disulfur which carried two
different terminal groups.[62]

Figure I-8. Mixed SAM formation strategies: a) mixed thiols adsorption, b) anhydride method adapted from Yan et al.[61]
and c) asymmetric disulfur adsorption

The characterization of mixed SAMs is not easy especially for determining the spatial
distribution of the different molecules. Several techniques are used such as the infrared spectroscopy
with grating incidence, photo-electron spectroscopy, ellipsometry and water contact angle. Some
structural characterizations are also used such as atomic force microscopy or scanning tunneling
microscopy.[60] Moreover, simulation showed that the spatial repartition of molecules in mixed SAMs
[63]
can lead to segregated structures.

15
 b) Nanoparticle assembly through chelating interactions

Indirect characterization can give also information on the SAM structure by grafting
ferrocenes[52] or nanoparticles.[64] In our team, iron oxide nanoparticles coated with oleic acid have
been assembled onto mixed SAMs of mercaptododecane (MDD) and mercapto-undecanoic acid (MUA)
which present alkyl chains with the same length (figure I-9).

Figure I-9. Mixed SAM preparation by adsorption of mercaptododecane (MDD) and mercapto-undecanoic acid (MUA)
following by iron oxide nanoparticle assembly.

The methyl terminal groups -CH3 is non-chelating for the nanoparticles in contrast to the
carboxylic function (-COOH) which is chelating for the iron oxide surface of nanoparticles. SAMs were
prepared with different ratios of MUA/MDD and were immersed in an oleic acid coated nanoparticle
suspension (figure I-10).

Figure I-10. Nanoparticle assembly onto SAM with ratio of MUA/MDD of a,d) 20%, b,c) 50% and c,f) 80%. a,b,c) scanning
electronic microscopy images and d,e,f) binary images. Scale bar: 100 nm

16
 The spatial distribution is different as function of the mixed SAM composition. For low quantity
of active groups the presence of nanoparticle domains confirms the phase segregation of the thiols
within the SAM. For a larger amount of COOH, the phase segregation disappears and the nanoparticles
arrangement becomes more uniform. The weak interaction between the nanoparticles and the surface
may lead to their rearrangement through the magnetic dipolar interactions.

c) Nanoparticle assembly through covalent binding

The “click” chemistry, introduced by Sharpless in 2001[65] is an ensemble of reaction between

functional groups which satisfied several requirements among which are high reproducibility,
robustness and easiness to process. “Click” chemistry is mainly associated to cycloaddition reactions
involving heteroatoms, such as hetero-Diels-Alder[66] and 1,3-dipolar cycloadditions.[67]
One of the most studied click reaction is the Huisgen cycloaddition also named Copper-
Catalyzed Azide-Alkyne Cycloaddition (CuAAC) (figure I-11).

Figure I-11. Schematic representation of the reaction between an alkyne and an azide in presence of a copper (I) catalyst
(CuAAC "click" chemistry)

This reaction creates a triazole bridge after reaction of alkyne (CΞC) and azide (N3) groups. In
2004, the click reaction was used for the immobilization of functional materials on SAMs (figure I-
12).[68-70]

17
 Figure I-12. CuAAC « click » reaction onto SAMs with a) electrochemically active molecules and b) protein-repellent
molecules. Adapted from a) Collman et al.[68] and b) Li et al.[70]

This concept has found a great interest in surface chemistry and more recently in the assembly
of nanoparticles.[71-75]
One of the great advantages of this assembly method is the covalent bond between the
substrate and the nanoparticles. It allows the strong and irreversible grafting of the nanoparticles
without possible rearrangement.
Another advantage of this method is to be versatile by functionalizing a variety of substrates
and nanoparticles with complementary azide and alkyne groups. Kinge et al. reported first an alkyne-
terminated FePt nanoparticle assembly onto an azido-terminated SAM.[75] However, previous work in
our group showed that the best configuration consist in alkyne terminated gold substrate and azide-
terminated nanoparticles.[76]

d) Nanoparticle assembly through multiple hydrogen binds

Nanoparticles can also be assembled onto surfaces through non covalent binding in order to
favor the rearrangement of the nanoparticles at the surface. This method can be done by two different
manners (figure I-13):

· The use of interactions where the association constant is high (> 105 M-1) such as host-
guest recognition. Many systems have been performed through this recognition,[77, 78] and
the most used example is the cyclodextrin which constitutes a supramolecular trap for
hydrophobic molecules such as adamantane or ferrocene groups.[79-82]
· The use of multiple weak interactions within one molecule to take advantage of the
multiplicity effects.[83-85] Here, the stability is ensured by the high number of bonds
between molecules. Examples of multiple hydrogen bonding can be found in literature

18
 such as the interaction between Hamilton receptor and barbituric acid[78, 86, 87] and base
pairing between complementary nucleosides.[85, 88, 89]

a) b)

Figure I-13. a) Nanoparticle binding mediated by Hamilton-type receptors and b) cyclopetides assembled through host-
guest recognition. Adapted from a) Zirbs et al.[87] and b) Nijhuis et al.[82]

Moreover, many assembly parameters can influence the recognition between the two
elements. The nature of the solvent, the temperature, the number of specific groups can impact the
binding energy. Therefore, it can be used as a tool to control the structured nanoparticles assemblies
(figure I-14).[90, 91]

Figure I-14. Schematic representation of proposed molecular recognition between a silicon surface modified with
adenine and polymer functionalized with thymine. From Viswanathan et al.[92]

Viswanathan et al. showed that the recognition process is promoted when chloroform
(aprotic) is used and no recognition happen in DMSO (protic).[90, 91] The change of solvent allows the
total reversibility of the self-assembly process.

The recognition process at the surface takes advantage of the high number of groups at the
surface of a SAM to enhance the formation of hydrogen bonds. To obtain stable binding of
nanoparticles onto a surface, the strength of interactions between the nanoparticles and the surface
need to be controlled. Therefore, the association constant between molecular receptors located
between the nanoparticles and the surface should be sufficiently strong and is often increased by
multivalent effects.

19
 In literature, the grafting of nanoparticles through hydrogen bonds is often performed in
solution or onto polymeric surfaces. The team of Reinhoundt showed the attachment of silica
nanoparticles through the host-guest recognition by using cyclodextrin.[78, 79, 81] Gold nanoparticles
have been grafted onto substrate through the Hamilton receptor and barbituric acid.[87, 93] Moreover,
gold nanoparticles were grafted through DNA strands which is also a recognition process involving
hydrogen binding.[94-99] However, the assembly of iron oxide onto substrate through such weak
interactions has been poorly reported.[100]

The assembly of nanoparticles through multiple hydrogen bonding requires a specific choice
of solvent and a particularly attention at the density of groups on the considered surface. However,
the high specificity coupled with low energy interactions allow flexible and dynamic coupling between
nanoparticles and substrate and may favor rearrangement by playing with solvent or temperature.[101]

2. Magnetic and optical properties of nanoparticles

a) Magnetic properties of nanoparticles

(1) Magnetic properties of a single nanoparticle

To understand the magnetic properties of magnetite nanoparticles, we will consider first the
bulk state and study the effect of the reduction of the size. In a magnetic material, the total magnetic
energy can be described as the sum of different contributions:

6 = 678 + 69 + 6: + 6; Equation I.8

The exchange energy (Eex) corresponds to the interaction between the spins of metal ions in
the crystal structure. This very strong and short range interaction explains the collective behavior of
magnetic moments in ferro- and ferrimagnetic materials. The anisotropy energy (Ea) conducts to align
the magnetic moments in a specific direction of the structure. The dipolar energy (Ed) is a long range
contribution which comes from the interaction between magnetic moments. The Zeeman energy (EZ)
is an interaction between the magnetic moments and an extern magnetic field.
To minimize the total energy, the magnetite presents a structure in domain separated by walls
called Bloch wall (figure I-15).

Figure I-15. Schematic representation of the energy diminution to create Bloch walls in a magnetic material

20
 All the magnetic moments are parallel inside the same domain. When an external magnetic
field is applied, the moments tend to align in the direction of the field and the Bloch walls are
displacing. Therefore, the magnetization of the material can be measured as function of the applied
field. A hysteresis cycle is obtained and is characterized by its saturation magnetization MS, remanent
magnetization MR and coercive field HC (figure I-16).

Figure I-16. Hysteresis cycle of the magnetization M as function of the applied magnetic field H

Nevertheless, creating a Bloch wall requires energy. Below a critical size rC, the best
configuration becomes a single domain with a magnetization following a specific direction. In this case,
the nanoparticle is composed of a unique magnetic domain with a stable magnetization at room
temperature: the nanoparticle consists in a single blocked domain. Therefore, we use the macro-spin
approximation which assimilates each nanoparticle to a single dipole representing the total
magnetization of the nanoparticle.

In this case, the anisotropy energy Ea tends to align the magnetic moment of a single
nanoparticle in a specific direction. This energy is the sum of different contributions (detailed in
appendix A). Here, the magnetocrystalline energy will be considered predominant and additional
component such as surface, shape and volume contribution can be neglected to explain the
magnetization dynamic.

The magnetocrystalline energy EMC favors the alignment of the magnetic moments in a specific
direction of the crystal lattice, the magnetization easy axis. For a single nanoparticle, the
magnetocrystalline energy at first order is:

69 = <> @A&²(C) Equation I.9

with K the anisotropy constant which depends of chemical composition of the material, V the
volume of the nanoparticle and θ the angle between the magnetization and the easy axis (figure I-17).
Therefore, the magnetization of a nanoparticle presents two minima of energy (θ = 0 and θ = π)
separated by an energy barrier EB = KV (θ = π/2).

21
 Figure I-17. Schematic representation of the free energy of a nanoparticle as function of the magnetization direction.
From Bendanta et al.[102]

For Fe3O4 the magnetization easy axis is parallel to the [111] direction, whereas the γ-Fe2O3
the magnetization easy axis is parallel to the [110] direction. Without external magnetic field, the
nanoparticle magnetization will be oriented parallel or anti-parallel to the magnetization easy axis.

(2) Magnetization dynamics

If the diameter of the nanoparticle still decreases below a certain size, the product KV becomes
lower than the thermal energy. In this case, the magnetization of the nanoparticles rotates from a
direction of easy axis to another without extern magnetic field. The Neel-Brown law shows that for
non-interacting nanoparticles, the switch of direction happens for a characteristic relaxation time τ:

HI
E = EF G JK L Equation I.10

with τ0 ≈ 10-9-10-11, kb the Boltzmann constant, K the anisotropy constant and V the volume of
the nanoparticle. This equation shows that the relaxation time increases when the temperature
decreases and the volume of the nanoparticle increases. For a low temperature or a nanoparticle with
a size large enough, τ becomes longer than the measurement time τm (depending on the equipment)
and the magnetization is blocked.

On the contrary, if the temperature is high enough or the nanoparticle size sufficiently small, the
relaxation time will be shorter than the measurement time τm; it is the superparamagnetic regime. In
this case the magnetic moments of the nanoparticles will oscillate during the measurement and the
resultant magnetization will appear to be zero. This behavior is characterized by the disappearance of
the coercive field and remanent magnetization.

The temperature when the regime is changing (τ = τm) is called the blocking temperature Tb:

MN =
OP
T
QK RS( U )
Equation I.11
T-

22
 Usually, the magnetic measurements performed by SQUID (Superconducting Quantum
Interference Devices) have τm = 100 s and τ0 = 10-9 s, the equation I.11 becomes:

<> = 25XY MY Equation I.12

Therefore, the blocking temperature is depending of the size of the nanoparticle, the
anisotropy and the measuring time. A sample of nanoparticle is never monodisperse and presents a
size distribution which conducts to a distribution of blocking temperatures. Moreover, this equation
does not take in account the interactions between nanoparticles. We will see later that the dipolar
interaction in nanoparticle assembly have an effect on the blocking temperature.

(3) Surface effect

When the size of a nanoparticle decreases, the surface/volume ratio increases strongly, thus
surface effects does. The diminution of neighboring atoms at the surface leads to a break of the
symmetry and creates disorder of the spins at the surface of the nanoparticle. The direct consequence
of this size reduction is that the disordered layer at the surface display a spin glass magnetic behavior
where the magnetization of surface atom is different of the one in the volume of the nanoparticle. This
effect is called spin canting and is responsible of the reduction of saturation magnetization in
nanoparticles. The spin canting is strongly dependent of the size of the nanoparticle but also of the
synthesis method and the ligand bound to the surface atoms.[103, 104] Safronov and al showed in 2013
the dependence of magnetization as function of the thickness of the canting spin layer:[8]

Z[ = ZY\]Q (1 − )
_` b
a
Equation I.13

where Mbulk is the saturation magnetization of the bulk (92 emu/g for the magnetite), D is the
diameter of the nanoparticle and Δ the thickness of the spin canting layer. By considering a constant
canted layer of 2 nm, we can easily calculate the saturation magnetization as function of the
nanoparticle size.

Figure I-18. Saturation magnetization as function of the size by taking in account a canted layer of 1 nm at the
nanoparticle surface. Adapted from Safronov et al.[22]

23
 The magnetization of the nanoparticle is strongly dependent of their size especially for small
nanoparticles. Although, the magnetization tends to stabilize for nanoparticles larger than 20 nm, the
saturation magnetization never reaches the magnetization of the bulk due to such surface disorder.

b) Magnetic properties of nanoparticle assembly

We have shown that the magnetic properties are dependent of the size of the nanoparticles;
however they are also dependent of the interactions between the nanoparticles. Nanoparticles can
interact with their neighbors through dipolar interactions, exchange interactions or super-exchange at
the interface. In our case, the nanoparticles are coated with an organic layer and the dominant
interactions are the dipolar interactions.

(1) Dipolar interactions

The dipolar interaction between two nanoparticles is long range and anisotropic. It depends of
the magnetization of each nanoparticle and of the inter-particle distance. In the case of the macro-
spin approximation,[2] each nanoparticle is assimilated to a unique dipole carrying the total
magnetization. In a random distribution of nanoparticles (powder state), the dipolar energy between
two nanoparticles is:[2, 37, 105]

6: = de9
-. .c
f Equation I.14

with μ, magnetization carried by a nanoparticle and a the distance between two nanoparticle
centers. In the case of 2D triangular array of nanoparticles:

6: = 2.8
.c
9f
Equation I.15

Nevertheless, in the case of a monolayer of nanoparticles with a random orientation, the

dipolar energy cannot be simplified and has to be studied experimentally.

(2) Influence of the inter-particle distance

The influence of the inter-particle distance on the dipolar interactions can be investigated by
studying the variation of the blocking temperature.

Frankamp et al.[106] and Fleutot et al.[105] studied the influence of the blocking temperature by
functionalizing iron oxide nanoparticles with ligand with different sizes. The blocking temperature
increased with the decrease of the inter-particle distance which resulted in stronger dipolar
interactions (figure I-19.a).

24
 Moreover, Poddar et al.[107] studied the blocking temperature as function of the dimensionality
of nanoparticle assembly (figure I-19.b) prepared by the Langmuir-Blodgett technique and by
dispersing nanoparticles into a polymer matrix.

a) b)

Figure I-19. Evolution of TB as function of inter-particle distance a and b) Imaginary component of the susceptibility as
function of the temperature of non-interacting nanoparticle (NIP), monolayer (2D) and multilayers (Q3D) nanoparticles.
From a) Fleutot et al.[105] and b) Poddar et al.[107]

TB markedly increased from dispersed nanoparticles to a monolayer. Moreover, the monolayer

also has a higher blocking temperature than the multilayer. The higher anisotropy in a 2D assembly
due to all the magnetic moments aligned in the plane of the substrate requires a high energy to reverse
the magnetization. However, the multilayer presents a component out of the plane ant therefore a
weaker energy barrier to pass. Similar studies in the team showed similar collective properties of
nanoparticle assemblies.[2, 76]

The influence of the dipolar interactions on the hysteresis cycle is not trivial and the
observations reported in literature are contradictory. Most of the study (theoretical and experimental)
showed that the coercive field HC is decreasing when the dipolar interactions is increasing[2, 106, 108, 109]
and the ratio of the remanent magnetization on the saturation magnetization MR/MS increase.[107, 108,
110, 111]
However, these values are also dependent of the anisotropy energy and therefore are
dependent of the size of the nanoparticles.[37, 112]

25
 C. Bio-sensors

1. Generalities of biosensors

The aim of a biosensor is to detect species such as ions, molecules or proteins. Nowadays the
detection of different molecules became really important for food safety, environmental, medical and
security purpose.[113, 114] Moreover knowing the concentration especially for diagnosis applications is
critical. The use of fluidic system where the analyte can be in contact with the recognition element
allow the creation of portative and easy to use devices which give raise to the current interest of
biosensing. Biosensors are composed of a biological recognition element integrated with a transducing
element. Therefore, an analyte will give a biological response which is converted in an optic or electric
signal thanks to the transducer. Such monitoring coupled with the versatility of the recognition
methods allow detecting variety of analytes which tells about the growing interest for biosensing.

Biosensors can be classified as function of the element to detect or of the type of signal which
is measured. Therefore, we can have among other, amperometric, voltametric or optic detection.[113]
In each case, a crucial factor for the detection is the recognition of the receptor with the target
element. We will focus on biosensors with optic transducers which use materials carrying plasmonic
properties. The main advantages of such plasmonics sensors are the use of label-free detection
protocols which allow detecting biomolecules in their natural forms, in-situ and in real time. The
sensitivity and the limit of detection allow small quantities of analyte to detect. These sensors usually
use the propagative or localized surface plasmon resonance of plasmonic materials such as noble
metals. After presenting historical and theoretical part to generate and excite surface plasmon, we will
focus on its interest to perform biosensing and how to improve the efficiency of such devices.

2. Surface plasmons

Surface plasmon occurs at the surface of metallic materials and refers to the collective
oscillation of the conduction electrons. These oscillations can be coupled with an incident light in order
to give birth to a resonance phenomenon called surface plasmon resonance (SPR). The SPR is used to
probe the change of the environment at the metal surface since it depends of the change in refractive
index.

a) Historical context

The first and most famous observation of optical phenomenon due to surface plasmon dates
back to the IVe century with the Lycurgus cup (figure I-20). The color of the cup is different depending
on the illumination’s orientation (green with the reflected light - when the illumination comes from
outside and red with the transmitted light - when the illumination comes from inside the cup). This
typical behavior is due to the presence of gold and silver nanoparticles which exhibit plasmonic
properties. We have to wait the beginning of the XXe century for the identification of the surface
plasmon by Wood who highlighted anomalies in the diffraction spectrum of reflected light through
diffracting grating.[115] In 1941, Fano gave an explanation to this phenomenon and showed that these

26
anomalies are due to the excitation of electromagnetic waves.[116] It is only in 1956 that the term
“plasmon” is introduced by Pines[117] and the next year by Ritchie who first described the theory of
surface plasmon.[118] In 1968, Ritchie introduced the concept of “surface plasmon resonance”[119] and
at the same time Otto[120] and Kretschmann[121] provided two distinct configurations to excite surface
plasmon on a thin film by using a prism. These works established a simple and efficient way to exploit
surface plasmons giving rise to different possible applications.[122, 123] The main use of the plasmon
resonance (and the one we are interested in) is in biosensors.[123, 124] In the early 1990, the first
commercial biosensors appear on the market (Biacore©) following by a strong enthusiasm with the
development of many devices based on plasmonic materials.

Figure I-20. Lycurgus cup under external illumination (left) and internal illumination (right) by light.

b) Theory

To understand the mechanisms of surface plasmon resonance, it is interesting to study first

the behavior of the electronic oscillations at the surface of a metal. These oscillations (often called
surface plasmon waves - SPW or surface plasmon polaritons - SPP) can be defined by an excitation at
the interface between a metal medium and a dielectric. The description of the surface plasmons is
based on the equation of propagation of electromagnetic waves called Maxwell equations.

The detailed description of the surface plasmon which leads to the determination of the
relation of dispersion is given in (appendix B). The results obtained conduct to different properties of
the surface plasmons:

Ø Surface plasmons are described by the dispersion relation:

ijk = lm n
op oq
op roq
Equation I.18

it is true for noble metal such as gold and silver : 1s < 0

Ø Real part of the metal permittivity has to be negative in the range of wavelength used,

Ø 1: < −1s this condition is verified for an interface metal/dielectric

Ø Surface plasmons can exist only for the radial polarization TM of the light

27
After these conclusions, we know that surface plasmons can be studied in different systems based on
their dispersion relation. Moreover the existence of a metal/dielectric interface is necessary to
generate the plasmon.

To generate the resonance phenomenon, the surface plasmon has to be coupled with light. We can
distinguish several different cases which are used for biosensing, but the two mains are:

· Localized surface plasmon resonance (LSPR), when the surface plasmon is confined to a
metallic nanostructure of a size comparable to the wavelength of light. In this case, the
collective oscillation of the free electrons of the nanoparticle becomes coherent with the
electric field of the incoming light giving the resonance phenomenon. Therefore, the incident
light can excite the surface plasmon without need of specific coupling.
· Propagative surface plasmon resonance (SPR), when the surface plasmon is confined into a
thin metallic film. In this case, the incident light cannot excite directly the surface plasmon. A
specific configuration is necessary to create the resonance phenomenon.

There are different techniques to excite the surface plasmon on thin metallic layer. The excitation
can be performed by waveguide coupling when a guided mode propagating along the surface or by
grating coupling by using a diffraction grating.[125] Nevertheless, the configuration by prism coupling is
the most conventional and the most used. This configuration using a prism with a high refractive index
is detailed in the next part using optic laws.

c) Excitation condition of surface plasmons with Kretschmann

configuration

The excitation of the surface plasmon on a thin film is based on the coupling of light in a prism
by using the technique of total internal reflection (TIR). To excite the plasmon, Kretschmann geometry
is used where the metallic layer is directly deposited on the prism surface. Plasmonic resonance
requires the coupling of the surface plasmon at the metal/dielectric interface with the evanescent
wave generated by the TIR method. The total reflection phenomenon appears when an incident light
incomes at the interface between two medium with different refractive index with an angle over the
value of a critical angle θC. In this case, there is no transmitted light and only reflected light. Moreover,
close to the interface, an evanescent wave is generated which decreases rapidly with the distance to
the surface. This condition happens at the critical angle defined by the Snell-Descartes law.

28
 dielectric
βsp

y θ2 metal

x
z prism
ky θ1

kx

Figure I-21. Schematic representation of the coupling resonance phenomenon by using a prism in the Kretschmann
configuration

Considering a medium 1 (prism) with a refractive index n1 crossed by an incident light forming
an angle with the normal at the interface θ1 (Figure I-21). This incident light will be reflected with the
same angle θ1 and transmitted with an angle θ2 in the medium 2 (metal) with a refractive index n2. The
Snell Descartes law gives us:

&* sin(C* ) = &_ sin(C_ ) Equation I.19

We deduct:

sin(C_ ) = sin(C* )
yz
yc
Equation I.20

In the case where the medium 2 is more refractive than the medium 1 (n2 > n1), the equation I.19 can

Indeed, if z sin(C* ) > 1 it cannot be possible to have a transmitted light: it is the total internal
y
be true for every angles. However if n1 > n2, this relation is no longer defined for every value of θ.
yc
reflection. In our case, the refractive index of the prism is larger than the metal refractive index (gold)
so this condition is respected for angle θ given by:

|} ≥ €•‚ƒ„…( ‡ )
†
†
Equation I.21
}

In this case, the totality of the incident energy is reflected and an evanescent wave appears in the
medium 2 and can only propagate at the vicinity of the surface.

The resonance phenomenon occurs when the wave vector k transmitted in the metal is equal to the
wave vector of the surface plasmon βSP. The wave vector of the transmitted light in the medium 2 has
a tangential component and a normal component at the interface between the two medium and is
defined by:

X8 = & sin(C_ )
Œ
' _
‰‰‰⃗
Xˆ = ‹X• = Œ
& )Ž@ _ (C_ )
' _
Equation I.22
X• = 0
By using Descartes law and the property cos2θ + sin2θ = 1, we have:

29
 X8 = & sin(C* )
Œ
' *
‰‰‰⃗
Xˆ = ‹X• = Œ
+&_ _ − &* _ @A&_ (C* )
'
Equation I.23
X• = 0
We can observe than the component ky is real when 0 < n1sin(θ1) < n2 and is pure imaginary when
n1sin(θ1) > n2. In this last case, we find again the limit where the incident light is only reflected and the
transmitted wave in the medium 2 is evanescent.

The plasmon resonance condition will exist when both the propagation constant of the surface
plasmon and the tangential component of the evanescent wave are equal. We obtain this resonance
condition by equalizing the equation I.18 and 1.23:

X8 = •[‘ Equation I.24

& sin(C* ) =
Œ Œ ,’ ,U
' *
n
' ,’ r,U
Equation I.25

Or:

†} ƒ„…(|} ) = no proq
o o
Equation I.26
p q

Equation I.26 gives us the resonance condition to generate and excite surface plasmon at the
metal/dielectric interface. This condition is true for a single incident angle and for a given wavelength.
Moreover, the SPR phenomenon can be monitored since the intensity of the reflected light decrease
and give a minimum of intensity for the resonance condition. This minimum of intensity (called
“resonance peak”) is dependent either on the incident angle on the wavelength, we talk about angular
or spectral interrogation respectively.

d) Influence of parameters on plasmons excitation

Several parameters can influence the resonance of the surface plasmons such as the refractive
index of the prism, the permittivity of the metal and of the dielectric. Some of these parameters have
to be controlled in order to satisfy the versatility of the measurements.

(1) Choice of the material

The material used has a strong influence on the position of the resonance. Most of metal
present plasmonics properties (gold, silver, aluminum, copper). However, gold and silver are
preferably used since their plasmon resonance occurs in the visible which allow their characterization
with UV-visible spectroscopy and even use naked-eye in the case of colloidal suspensions. Moreover
gold surface can be easily functionalized with different molecules as seen previously.[39, 40] Nevertheless
silver suffers from a poor chemical stability and SPR devices should use a protective layer in order to
protect the metal from the oxidation and make easier functionalization.[126]

30
 Gold surface remains the perfect candidate since its functionalization is easily afforded with
self-assembled monolayers (cf. II.A.1).

(2) Metal permittivity

Different points have to be highlighted after considering the dispersion relation (equation I.18)
of the surface plasmon. First, it is important to notice that the permittivity of the metal is a function of
the pulsation εm(ω) and can be described by the Drude model for the free electron in the metal:[127]

1s (“) = 1 −
Œ”
c

Œ r•Œ–
c Equation I.27
—

with γl the damping factor and ωp the plasma pulsation given by:

“˜ = ns,
y7²
Equation I.28
-

with n the electron density of the material, e the electron charge and m the electron mass.

In the visible light domain (400 - 800 nm) the metal is considered without losses[127] and
therefore the damping factor is equal to zero and the equation I.27 becomes:

1s (“) = 1 −
Œ”
c

Œc
Equation I.29

Knowing that the pulsation is given by:

“=
_e'
™
Equation I.30

The permittivity of the metal is given for each excitation wavelength. In the plasmon dispersion

(equation I.29), we can notice the denominator is cancelled for 1: + 1s = 0. Therefore for the
relation (equation I.18), if we replace the permittivity according to the Drude model without losses

frequencies = ±
Œ”
+*r,’
, k tends to the infinite. These are the frequencies of the system allowing the

is water the asymptote equation will be “ =

Œ” [127]
determination of the asymptote from the dispersion relation (figure I-22). If the dielectric at the surface

√_
.

31
 βsp

Figure I-22. Dispersion relation of surface plasmon as function of the surrounding medium. From Bryche.[127]

As we can see from the dispersion relation (equation I.26); the coupling between the
propagation of the incident light and the surface plasmon occurs when the curves representing the
dispersion relation of the plasmon and the incident light are crossing. Here, we see how the medium
supporting the incident light is critical to excite the surface plasmon and therefore explain the use of
a prism with high refractive index.

(3) Temperature and metal thickness

We can also take in account the influence of the temperature which is dependent of the metal
conductivity. The plasma frequency is therefore dependent of the temperature and linked by the
relation:[127]
Œ”-
“˜ =
+*rš(›œ›- )
Equation I.31

with T0 the reference temperature, “˜- the plasma frequency and α the thermal dilatation coefficient.
The temperature is a parameter that could modify the resonance condition. Therefore the
temperature is critical and will be controlled during the SPR measurement (18°C-21°C in our study).

Another parameter is the thickness of the metallic layer. Obviously the thickness of the metal
should be thin enough otherwise the evanescent wave will not reach the interface with the dielectric.
In another hand, the permittivity of the metal can be impacted by the film thickness.[128] Some
variations can be observed as function of the thickness due to the losses of the metal permittivity
which is represented by the increase of the imaginary part.[129] This should be importantly considered
and therefore working with metallic films with the same thickness. In our study we will consider a
continuous and homogenous metal film with thickness of 30 nm or 50 nm.1

1
The study will be realized on different substrates, the thickness of the metallic layer will be constant for each
study, but different samples will be characterized either with a 30 nm either a 50 nm gold thickness.

32
 (4) Refractive index at the surface

The previous parameters can be settled to measure the samples in the same conditions.
However, the condition of excitation of the plasmon (equation I.26) shows a strong dependence on
the dielectric permittivity εd. Indeed, the change of permittivity at the surface of the metal will
therefore induce a change in the spectral or angular peak position. The figure I-24 represents the
resonance angle as a function of the permittivity of the dielectric when the medium 1 is a prism of BK7
with a refractive index of 1.51. Here, we can observe that the incident angle is higher for shorter
wavelength.

Figure I-23. Change of the angular response as function of the dielectric constant at the surface for different working
wavelengths

permittivity of the dielectric 1: at the metal surface. It is also observed that a change of permittivity
From the equation I.26, we can easily see that the incident angle is strongly dependent of the

creates a larger change in the incidence angle for shorter wavelengths. This better sensitivity for the
shorter wavelengths have been explained by Liedberg and al[130, 131] who showed that the material
have a larger absorption (due to the increase of the imaginary part of εm) for shorter wavelengths.

(5) Penetration depth

One property which is critical for the understanding of the surface plasmon is the penetration
depth which corresponds to the distance that the evanescent wave can reach from the metal surface.
The evanescent wave has a progressive structure on the normal direction of the metal/dielectric
interface (y axis) and its amplitude decays exponentially as:

33
 6• (•) = 6• (0)G œQž • Equation I.32

with

X• _ = X8 _ − 1
Œ²
'² :
Equation I.33

Assuming this, the decay electric field of the evanescent wave can be written:

6• (•) = 6• (0)G œŸ
ž
Equation I.34

with δ the penetration depth expressed as:

= n¡ U, ² ’ ¡ = n¡ U, ² ’¡
' , r, ™ , r,
_Πde
Equation I.35
’ ’

The penetration depth increases with the wavelength of excitation and decreases when the
dielectric constant increases at the surface. This exponential decay is the reason why the plasmon can
be detected only at the vicinity of the surface. This value is important since it determines the range
and the volume of sensing at the metal surface. In the early 90s, Liedberg et al. showed that the
sensitivity is the best, the closest to the surface by modeling dielectric matrix at different distances
from the metal (Figure I-25).[131]

Figure I-24. (a) Attenuation of the electric field for two model with a direct surface interaction (top) or extended matrix
(bottom) and (b) calculation of the SPR response as function of the distance from the metal interface of an organic layer.
From Liedberg et al.[131]

Therefore performing biosensing requires detecting the analyte at the vicinity of the metal in
order to ensure a good sensitivity and an optimization of the biosensor. We will see how changing the
parameters can influence the penetration depth and what are the consequences on the sensitivity of
the sensors.

To conclude, the surface plasmon of a gold substrate can be excited with light in the visible
range when it is coupled with a prism in the Kretschmann configuration. The variation of permittivity
(and therefore refractive index) at the surface of the gold substrate will induce a change in the
resonance condition which will be monitored in order to detect adsorbed species.

34
 3. Plasmonic biosensors

a) Generalities on plasmonics biosensors

The sensitivity to change in refractive index allows plasmonics sensors to detect the binding
events at the surface of the metal. The main goal of a biosensor is to give a response by binding an
analyte with the best sensitivity meaning the maximum response for a minimum change of refractive
index. As shown in the previous part, the signal of the SPR is strongly dependent of the dielectric
properties on the surrounding environment. Therefore, plasmonic sensors are used for biosensing
devices and require several characteristics:

Ø A sensing surface including an organic receptor to adsorb the target molecules through
molecular recognition process
Ø A microfluidic system allowing the injection in-situ and real time of the analyte
Ø An optical configuration linked to an online computer to measure the change in the
reflected light (as function of the incident angle or wavelength)

Therefore, the key parameter is the functionalization of the surface by recognition elements
which has to be highly specific to detect the target molecules. For the material, gold and silver are the
most common materials used as biosensors because of their strong resonance peak in the visible.
Moreover, these metallic surfaces present an easy and well described functionalization in literature
which allows the grafting of receptor molecules by different approaches. Indeed a significant aspect
is the way the bioreceptors are anchored to the surface. The molecules are either physisorbed,[132-134]
either chemically attached.[132] Covalent binding is preferred as it provides stronger and stable binding.

The performances of SPR sensing are driven by several characteristics which are specific to the
device and allow understanding the performances but also the limitations of the biosensors:

Ø Resolution or limit of detection (LOD); it corresponds to the lowest concentration which

can be detected and differentiated from the noise.
Ø Sensitivity; it corresponds to the ratio between the variation of the response and the
change of corresponding refractive index.
Ø Selectivity; the capacity to differentiate two different analytes.
Ø Accuracy; the difference between the measured value and the real value.

This kind of devices present multiple advantages such as the rapid and real time analysis, in-
situ detection, a low limit of detection and it does not require the labeling of the analyte. This is why
using SPR as biosensing presents a great interest. Several species such as protein, DNA, enzyme, cell,
nucleic acid, antigen–antibody, and microorganism have been detected below the pM[113, 122, 135, 136]
which makes the SPR sensing a powerful tools (table I-1).

35
Table I-1. Limit of detection of different analytes. Extract from Fan et al.[114]

Technology
Optical structures Analyte Detection limit References
platform

Surface plasmon resonance Bulk solution 10-5-10-8 RIU [137-142]

Long range SPR Bulk solution 10-7-10-8 RIU [143-145]

Surface plasmon resonance Bulk solution 10-5-10-7 RIU [146-149]

[150]
Imaging (SPRi) Protein 1 nM
[151]
Optical heterodyne SPR Protein 0.2 nM
[152]
Phase sensitive SPR Protein 1.3 nM
[153]
Wavelength modulated SPR DNA 10 pM
Surface
plasmon SPRi DNA and RNA 10 nM [154]

resonance [155]
Flow injection SPR DNA 54 fM, 1.38 fM
[156]
Angle modulated SPR Protein 0.15 ng/mL
[157]
Surface plasmon resonance Protein 66.7 unit/mL
[158]
Surface plasmon resonance Protein 50 ng/mL

Bacteria (E.coli) 106 cfu/mL [159]

Prism-based SPR
[160]
Salmonella 100 cfu/mL
[161]
Biacore 2000 SPR Bacteria 25 cfu/mL

Nevertheless, SPR biosensing have limits such as the detection of small molecules or very
diluted concentrations.[162] These limits have to be pushed in order to improve the detection this is
why, the optimization of the SPR sensing becomes necessary.

Progress and achievements of the two different kinds of sensors (propagative and localized
plasmon resonance) are showed in the following parts.

b) Localized surface plasmon resonance sensors (LSPR)

LSPR biosensors are based on the use of metallic nanostructures which can generate plasmons.
In these cases, the incident electric field will induce a uniform displacement of the conduction
electrons. LSPR can be directly excited by incident field if the nanostructure is smaller than the
wavelength of light particularly because of the geometry of the nanostructure (figure I-26).[163]

36
Figure I-25. Schematic representation of the localized surface plasmon generated by electric field on metal nanoparticle

In this case, the Mie theory[164] describes the properties of the nanostructures and explains the
adsorption and scattering phenomenon. In this case, the resonance is expressed as:

“¢[‘£ =
Œ”
+(*r_,’ )
Equation I.36

with ωp the plasma frequency of bulk material. As seen before, the LSPR frequency is strongly
dependent of the surrounding dielectric at the surface of the metallic nanostructure.

Gold, silver and aluminum are the common metals used because of their plasmonic resonance
which is set in the visible light domain. The optical extinction of the nanoparticles displays a maximum
at the plasmon resonance frequency which occurs in the visible domain for noble metals, explaining
the color of metallic nanoparticle suspensions. Thanks to these properties, nanostructures can be used
as biosensors by functionalizing their surface with various bio-receptors. Thus, the recognition can be
easily monitored with spectroscopy techniques.

The shape and size of nanostructures modify their absorption band (figure I-27) which can
allow a spectral tenability of the SPR signal.[162, 165] The surrounding medium affects also the absorption
band and the adsorption of molecules at the surface changes the spectroscopy measurements.

Figure I-26. Absorption bands of gold nanoparticles as function of their shape. From Noguez.[165]

37
 UV-vis spectroscopy is a good technique to monitor the change at the nanoparticle surface.[166]
Moreover, the use of noble metal nanoparticles such as gold or silver gives an absorption in the visible
light which makes possible the detection with naked eye. It gives raise to two different approaches to
use plasmonic nanostructures for biodetection; the first one using nanoparticles in solution
(colorimetric assays) and the second using nanostructures onto substrates to form arrays.

(1) Colorimetric assays

Noble metal nanoparticles are used in plasmonic biosensors for detection. Indeed, the
aggregation of metallic nanoparticles causes the change of the surrounding environment and creates
coupling interactions between them.[167] The stability of the suspension is therefore a key parameter
and the introduction of the target molecules, which act as a molecular linker between the
nanostructures, will lead to the formation of aggregates in solution and to the change of the solution
coloration.[168] The use of recognition elements such as the biotin-streptavidin couple has been widely
used. The gold nanoparticles can be easily functionalized by thiol groups with a biotin group at the
surface in order to detect streptavidin in a solution. The streptavidin with its four binding sites can link
the nanoparticles together and form array of nanoparticles in suspension.[169, 170] The group of Kikuchi
et al. used functionalized gold nanoparticles to detect glycosidases (Figure I-28).[171] The presence of
glycosidase in the nanoparticles suspension triggered the aggregation through electrostatic
interactions. Gold nanoparticles have also been crosslinked with DNA strands.[172, 173] The nanoparticles
are functionalized with aptamers and the presence of the complementary aptamers in the suspension
led to their aggregation.

Figure I-27. Different concentration of β-galactosidase and β-glucosidase showing the change of coloration of gold
nanoparticles caused by aggregation. From Zeng et al.[171]

These colorimetric assays are an easy and real time way to detect in solution the presence of
various analytes such as proteins, enzymes or aptamers. The controlled aggregation in suspension of
the nanostructures is the key parameter as long as it is responsible of the color change of solution
observed by naked eye.

(2) Surface nanostructured architecture

Plasmonic nanostructures can also be used on a surface in a Kretschmann configuration to

detect local changes of the refractive index with a fluidic channel. The top-down (lithography) and

38
bottom-up (assembly of nano-objects) approaches are both used to create devices using
nanostructures with localized surface plasmons onto substrates.[174] The different assembling methods
(cf. I.B.1) allow controlling the structuration of plasmonic nanoparticles onto a surface with a good
control of shape and size ranging from 2 nm until 100 nm.[166, 175]

In another hand, different lithography processes allow a fine control of the size and the shape
with a high density on large areas but cannot reach size below hundreds of nanometer. Nanostructures
on the surface[176-178] or nanopatterning popularized by Van Duyne[179] (Figure I-29) have been used to
creates nanostructured surfaces. These systems highlighted a good sensitivity because of the
confinement of the electromagnetic energy in the nanostructures and considerably reduced the
penetration depth (cf. I.C.1.d).[177, 180]

Figure I-28. Scanning electron microscopy of gold nanodiscs deposited on a substrat. From Barbillon et al.[177]

These lithography approaches are widely used but the processes are expensive and slow.
Therefore, top-down approach is not suitable to achieve fast and cheap nanostructured surfaces. The
use of nanomaterials in order to improve the performance of sensors is often more preferred.

c) Propagative surface plasmon resonance (SPR)

As seen in section (I.C.2.c) the SPR displayed by thin metallic films cannot be excited by free
space radiations and therefore has to be coupled with a prism with a high refractive index. However,
different methods exist to couple light with the surface plasmon of thin films (figure I-30):[114, 181]

Ø Waveguide coupling: the light propagates in a waveguide through total internal

reflection and generates an evanescent field at the waveguide–metal interface and
excites the surface plasmon as for the prism configuration.[128, 182]
Ø Optical fibers: in the same way, the light is carried by the fiber which can have different
configurations such as side-polished single mode or multimode, tip-polished,
polarization maintaining of D-shaped fiber.[183-186]
Ø Grating coupling: the incident light is coupled with a periodic array at the substrate
surface.[187, 188]
Ø Long/short-range surface plasmon: the plasmonic substrate is surrounded by two
dielectric layers that have similar refractive index. The surface plasmons from both
interfaces are coupled and generate two modes called long and short range surface
plasmons.[126, 145, 189-191]

39
 Figure I-29. Schematic representation of various SPR sensor configurations. (A) Prism coupling, (B) waveguide coupling,
(C) optical fiber coupling, (D) side-polished fiber coupling, (E) grating coupling and (F) long-range and short-range surface
plasmon (LRSP and SRSP). Adapted from Fan et al.[114]

The feature of the propagative SPR lies in the larger volume sensing in comparison with the
LSPR. Indeed, the large penetration depth allows sensing over few hundred of nanometers which is a
great advantage to detect very large objects such as cells or bacteria.[123] However, this large
penetration depth reduces the sensitivity. Long-range SPR, despite the large sensing depth, allows a
good sensitivity and a narrow SPR feature.[192]

In conclusion, each configuration presents different advantages and propagative and localized
SPR can even be coupled in order to combine both properties and enhanced the sensitivity.[127, 193]

4. Plasmonics biosensors optimization

As we saw previously, many techniques have been used to perform bio-sensing using
plasmonic properties of noble metal materials. The efficiency of these sensors can be increased by two
methods; improving the intrinsic optical properties and the characteristics of surface functionalization.

a) Intrinsic properties

The first method to improve the performance consists to decrease the sensing volume. Indeed,
as shown by equation I.35, the penetration depth is representing the distance that the evanescent
wave can probe and so be effective for sensing.

The response of a SPR measurement can be expressed as function of λ or θ and is dependent

of the change of refractive index at the surface of the metal. The response of a sensing system R is
given by:

¤ = ¥∆& = ¥(&7§§ − &¨ ) Equation I.37

with m the sensing factor, neff the effective refractive index of the adlayer at the surface of the
metal and ns the refractive index of the solvent. The intensity of light is the field strength squared, so

40
 ª«c
it decays with height y above the metal surface as e —’ with ld the decay length in the dielectric layer
which is linked with the penetration depth δ by:

¬: = 2 Equation I.38
ªc«
The weighting factor for determining the average refractive index is therefore e —’ . This was
indeed proven to be very accurate through Maxwell’s equations by Liedberg et al.[131] The effective
refractive index is therefore calculated with the integral:[194]
ªcž
&7§§ = ∫ &(•)G —’ ®•
_ ¯
]’ F
Equation I.39

Considering a bilayer structure we have, n(y) = nd when 0 < y < d and n(y) = ns when y > d and the
integral becomes:
ªc’
&7§§ = &¨ + (&: − &¨ )[1 − G —’
] Equation I.40

and R becomes:

ª‡p
´ = q(†p − †j ) µ} − ¶ ·p ¸ Equation I.41

The response, R, is most of the time corresponding to a spectral shift measured with a constant
incident angle (Δλ) or an angular shift measured with a constant wavelength (Δθ). By the equation I.39,
we can easily see that the sensitivity can be increased by two different factors for a given change of
refractive index at the surface:

· by decreasing the decay length ld

· by increasing the sensitivity factor m

Decreasing the decay length, and therefore the penetration depth, corresponds to decrease
the sensing volume and therefore to increase the contribution of the analyte refractive index on the
surface of the sensor. The penetration depth, given by equation I.35, is dependent of the excitation
wavelength in the metal and the permittivity of the dielectric layer εd. We can easily see that the
penetration depth increases directly with the wavelength. Therefore, the sensitivity of the detection
can be improved by changing the incident wavelength. The penetration depth can also be tuned by
the nanostructuration of the metal at the surface.[177] Indeed, the decay length (ranging from 100 nm
up to 1 μm for classic SPR system) can be confined in nanostructure and reaches values down to 20
nm.[175, 180] The main drawback we can highlight is that the confinement around the nanostructures. It
decreases the sensing volume and does not allow the detection above few nanometers from the
surface of the nanostructures.

The sensitivity factor m (or refractive index sensitivity) corresponds to the slope of the
variation of the measured shift (spectral or angular) for a change in effective refractive index at the
surface. It becomes crucial for detecting small analytes (lower than 500 Da) to have a strong sensitivity.
The sensitivity factor increases when the analyte is close to the surface but decreases with the distance
from the surface.[131] The sensitivity is also dependent of the nanostructuration of SPR material, the
operating wavelength and the instrumentation.[195]

41
 The SPR and LSPR have similar magnitude order for the sensitivity. While propagation SPR has
a large penetration depth and sensitivity factor, in contrast the LSPR has a low sensitivity factor but
short decay length. To improve the total sensitivity of such nanodevices, different configurations of
sensors are used (table I-2).
Table I-2. Summarize of SPR performances for different configurations. Extract from Roh et al.[125] References correspond
to the cited article.

Despite of the advantages that provide the nanostructures for LSPR, the sensitivity of this kind
of sensors remains low. Nanostructured layers can provide a higher sensitivity due to the electric field
enhancement between nanoparticles. Jain and al have demonstrated than the sensitivity of sensors
using gold nanoparticles is higher when the nanostructures are closer but remain below 500
nm/RIU.[196] In another hand nanostructured film based biosensors can achieve a higher sensitivity up
to 30000 nm/RIU by using metamaterials based on gold nanorods which support a guided mode onto
a porous surface (Figure I-31).[162]

42
 Figure I-30. Schematic representation of the nanostructured surface of gold nanorods. From Kabashin et al. [162]

b) Surface functionalization

The sensitivity of these sensors is tuned by controlling the nanostructuration of plasmonic

materials at the surface. Furthermore, the sensitivity can be increased by tuning the surface chemistry.
Indeed, tuning the quantity of receptor molecules can improve the detection of the target analyte. The
detection of proteins or large molecule implies a steric hindrance at the surface which can be optimized
by diluting the recognition elements.[197-200]

Figure I-31. Schematic representation of streptavidin adsorption onto mixed biotinylated-SAM. From Jung et al.[200]

The use of mixed SAMs onto a gold surface allowed the good recognition of the analyte (Figure
I-32). Controlling the spacing between functional recognition groups can therefore increase the
sensitivity by adsorbing more target molecules. Indeed, Perez et al. showed that the recognition of
streptavidin on a mixed SAM surface is optimal when the concentration of biotin groups at the surface
is about 20%.[199] The correct functionalization and homogenous spatial arrangement of receptor
molecules at the surface is critical to have efficient sensing.

43
 As we saw, the sensitivity of plasmonics biosensors is controlled by the structuration of the
plasmonic material deposited onto and the configuration and the chemical recognition at the surface.
Nevertheless, non-plasmonic materials have been also used for biosensing and can achieved a great
enhancement in sensitivity.[126, 201-204]

5. Iron oxide nanoparticles based biosensors

Nan-plasmonic materials can be of interest to enhance markedly sensitivity. Indeed, high

refractive index materials deposited onto gold thin films allow shifting strongly the resonance peak. As
seen previously, sandwich immunoassay gives rise to the sensitivity of biosensors. Indeed, the signal is
not sufficient to detect small analytes adsorbed on the surface. Nevertheless, coupling the analyte to
a nanoparticle or to an object with large dimensions will give a larger shift after the adsorption. Gold
nanoparticles which have been first used to amplify the SPR response.[205, 206] Other kinds of
nanoparticles such as latex beads or silica nanoparticles with higher refractive index also induce a
larger shift.[207, 208]

The use of iron oxide nanoparticles gives a better sensitivity thanks to its high refractive index
which generates a larger shift and acts as a detection amplifier. Sun and al were the first in 2007 to use
magnetite to improve the sensitivity of SPR sensors.[209] Therefore the magnetite nanoparticles coated
with antibody are trapped onto a gold substrate using magnetic force. The detection of the specific
antigen occurs at the surface of the nanoparticles. The main advantage of this approach is the reuse
of the sensors by removing the magnetic microbeads support when the magnetic force is withdrawn.
Soeldberg and al in 2009 used magnetic nanoparticles coated with streptavidin to bind a biotinylated
antibody at the substrate surface.[210] The recognition of antibody alone is not sufficient to generate a
large shift, but the presence of the antibody will indirectly bring 50 nm iron oxide nanoparticles which
will create a strong shift. Similar and extended studies have been reported with different sizes of
nanoparticles allowing the detection of multiple analytes at concentration below nM by using
sandwich assays.[211-213] The magnetic field can also be used for trapping aggregated nanoparticles
which are stable in solution but form aggregates on the plasmonic substrate in presence of the
targeted analyte (Figure I-33).[214-216] In this case, the SPR response is linked to the analyte
concentration in solution and the system can afford to determine the limit of detection.

44
 Figure I-32. Schematic representation of a sandwich assay to increase the sensitivity of SPR sensors by using iron oxide
nanoparticles. From Teramura et al.[216]

These techniques based on the use of iron oxide nanoparticles allow the increase of the
sensitivity and the limit of detection below nM for different proteins. In these examples, the
nanoparticles are conjugated to a recognition element for detecting the analyte and then trapped onto
the surface by applying magnetic field.[214, 215] The refractive index increases strongly at the surface and
gives a higher SPR response. Moreover the high surface/volume ratio of nanoparticles leads to a high
density of biomolecules at their surface which contributes to enhance the signal.[216] A crucial point is
the non-aggregation of the nanoparticles in suspension before starting analysis in order to monitor
precisely the variation of the refractive index at the surface.

Nanoparticles have been used in sandwich assay to enhance the detection of small molecules.
Nevertheless it has not been reported as intermediate to increase directly the sensitivity factor. Pichon
et al. showed the use of iron oxide nanoparticles to control the position of the plasmon resonance
peak by grafting iron oxide nanoparticles onto a gold surface (Figure I-34).[64] The control of the peak
position as well as the increase of the surface area is suitable to improve the sensitivity of the
detection. Furthermore, the functionalization by receptor molecules on nanoparticle surfaces with a
high curvature radius can increase the accessibility of molecules receptors and favor the molecular
recognition with the target biomolecules.[217, 218]

45
 Figure I-33. Position of the plasmon peak resonance as function of the density of magnetite nanoparticles. From Pichon
et al.[64]

The strategy developed in this manuscript is based on the assembly of magnetite nanoparticles
onto gold surfaces. Indeed, as we have seen before, gold is an excellent candidate since it fulfills the
condition to generate surface plasmon and allow an easy functionalization by thiol derivatives.
Moreover, the iron oxide presents a high refractive index in the visible range which will allows
controlling the plasmonic properties of gold substrate. Therefore, the assembly of magnetite
nanoparticles by CuAAC “click” chemistry is a good opportunity since the synthesis and assembling
processes are well controlled on gold surface which were functionalized by SAMs. The assembly of
Fe3O4 nanoparticles through covalent binding to the surface using click chemistry brings many
advantages such as:

· Versatility; terminal head groups at the surface can be chosen in a variety of different
functions.
· Specificity; crucial point for the preparation of nanoparticle and further assembly for the
grafting of bio-receptors which are essential for the SPR detection of analytes.
· Size; detection happening in the vicinity of the metal, thin layer of materials is convenient
to optimize the signal.

Synthesis and assembly processes allow the control of the size and density of the nanoparticles
to tailor the plasmonic properties of the gold substrate. The grafting of receptor molecules at the
surface of nanoparticles can be done by reusing the CuAAC “click” chemistry. Indeed molecules
functionalized with functional groups available for the “click” reaction are grafted on the assembly of
magnetite nanoparticles.

46
 D. Bibliography

[1] T. F. W. Barth, E. P., Zeitschrift für Kristallographie, (1932) 82, 325.

[2] Pauly, M., Université de Strasbourg, (2010),
[3] R. M. Cornell, U. S., The Iron Oxides: Structure, Properties, Reactions, Occurences and Uses,
VCH, W. Ed. (2003).
[4] P. Novak, H. S., J. Englich, J. Kohout, V. A. M. Brabers, Physical Review B, (2000) 61, 1256.
[5] Antonov, V. N.; Harmon, B. N.; Antropov, V. P.; Perlov, A. Y.; Yaresko, A. N., Physical Review B,
(2001) 64,
[6] Nanoparticles, V. C. d. O. o. I., JACS, (2007) 129, 10358.
[7] M. Jackson, B. M., J. Bowles, The irm Quarterly, (2011) 20, 1.
[8] Safronov, A. P.; Beketov, I. V.; Komogortsev, S. V.; Kurlyandskaya, G. V.; Medvedev, A. I.;
Leiman, D. V.; Larrañaga, A.; Bhagat, S. M., AIP Advances, (2013) 3, 052135.
[9] Santoyo Salazar, J.; Perez, L.; de Abril, O.; Truong Phuoc, L.; Ihiawakrim, D.; Vazquez, M.;
Greneche, J.-M.; Begin-Colin, S.; Pourroy, G., Chemistry of Materials, (2011) 23, 1379.
[10] Sandulyak, A. A.; Sandulyak, A. V.; Ershova, V.; Pamme, N.; Ngmasom, B.; Iles, A., Journal of
Magnetism and Magnetic Materials, (2017) 441, 724.
[11] Query, M. R., (1985),
[12] Ali, A.; Zafar, H.; Zia, M.; Ul Haq, I.; Phull, A. R.; Ali, J. S.; Hussain, A., Nanotechnology, science
and applications, (2016) 9, 49.
[13] Salata, O., Journal of Nanobiotechnology, (2004) 2, 3.
[14] F. Hasany, S.; Ahmed, I.; J, R.; Rehman, A., Nanoscience and Nanotechnology, (2013) 2, 148.
[15] Xu, J.; Sun, J.; Wang, Y.; Sheng, J.; Wang, F.; Sun, M., Molecules, (2014) 19, 11465.
[16] Gupta, A. K.; Gupta, M., Biomaterials, (2005) 26, 3995.
[17] S. Laurent, D. F., M. Port, Al. Roch, C. Robic, L. Vander Elst, R. N. Muller, Chemical reviews,
(2008) 108, 2064.
[18] Figuerola, A.; Di Corato, R.; Manna, L.; Pellegrino, T., Pharmacological research, (2010) 62,
126.
[19] Lu, A. H.; Salabas, E. L.; Schuth, F., Angew Chem Int Ed Engl, (2007) 46, 1222.
[20] J. A. Lopez-Pérez, M. A. L.-Q., J. Mira, and J. Rivas, IEEE Transactions on magnetics, (1997) 33,
4359.
[21] D. Chen, R. X., Materials Research Bulletin, (1998) 33, 1015.
[22] Wan, J.; Cai, W.; Meng, X.; Liu, E., Chem Commun (Camb), (2007), 5004.
[23] Zeng, S. S. a. H., JACS, (2002) 124, 8204.
[24] Shouheng Sun, H. Z., David B. Robinson, Simone Raoux, Philip M. Rice, Shan X. Wang, and
Guanxiong Li, JACS, (2004) 126, 273.
[25] U. Jeong, X. T., Y. Wang, H. Yang, Y. X, Advanced materials, (2006) 19, 33.
[26] Victor K. LaMer, R. H. D., JACS, (1950) 72, 4847.
[27] Baaziz, W.; Pichon, B. P.; Fleutot, S.; Liu, Y.; Lefevre, C.; Greneche, J.-M.; Toumi, M.; Mhiri, T.;
Begin-Colin, S., The Journal of Physical Chemistry C, (2014) 118, 3795.
[28] Demortiere, A.; Panissod, P.; Pichon, B. P.; Pourroy, G.; Guillon, D.; Donnio, B.; Begin-Colin, S.,
Nanoscale, (2011) 3, 225.
[29] Cotin, G., Université de Strasbourg (2017).
[30] Xiaojie LIU, B. P. P., Corinne Ulhaq, Christophe Lefevre, Jean-Marc Greneche, Dominique
Begin, and Sylvie Begin-Colin, Chem Mater, (2015) 27, 4073.
[31] Park, J.; An, K.; Hwang, Y.; Park, J. G.; Noh, H. J.; Kim, J. Y.; Park, J. H.; Hwang, N. M.; Hyeon,
T., Nature materials, (2004) 3, 891.
[32] Liese, H. C., The American Mineralogist, (1967) 52, 1198.
[33] Baaziz, W., Université de Strasbourg, (2011),
[34] Bellido, E.; Domingo, N.; Ojea-Jimenez, I.; Ruiz-Molina, D., Small, (2012) 8, 1465.

47
[35] F. C. Meldrum, N. A. K., and J. H. Fendler, J Phys Chem (1994) 98, 4506.
[36] F. Mammeri, Y. L. B., T. J. Daou, J-L. Gallani, S. Colis, G. Pourroy, B. Donnio, D. Guillon, S.
Beégin-Colin, j Phys Chem B, (2009) 113, 734.
[37] Pauly, M.; Pichon, B. P.; Panissod, P.; Fleutot, S.; Rodriguez, P.; Drillon, M.; Begin-Colin, S.,
Journal of Materials Chemistry, (2012) 22, 6343.
[38] Pichon, B. P.; Louet, P.; Felix, O.; Drillon, M.; Begin-Colin, S.; Decher, G., Chemistry of
Materials, (2011) 23, 3668.
[39] J. C. Love, L. A. E., J. K. Kriebel, R. G. Nuzzo, G. M. Whitesides, Chemical reviews, (2005) 105,
1103.
[40] Morel, A. L.; Volmant, R. M.; Methivier, C.; Krafft, J. M.; Boujday, S.; Pradier, C. M., Colloids
and surfaces. B, Biointerfaces, (2010) 81, 304.
[41] Bedford, E. E.; Spadavecchia, J.; Pradier, C. M.; Gu, F. X., Macromolecular bioscience, (2012)
12, 724.
[42] Briand, E.; Humblot, V.; Landoulsi, J.; Petronis, S.; Pradier, C. M.; Kasemo, B.; Svedhem, S.,
Langmuir, (2011) 27, 678.
[43] Francisco Palazon, P. R. R., Ali Belarouci, Céline Chevalier, Hassan Chamas, Éliane Souteyrand,
Abdelkader Souifi, Yann Chevolot and Jean-Pierre Cloarec, Nanopackaging: From Nanomaterials to
the Atomic Scale, Springer Ed. (2015).
[44] Palazon F., M. M., Chevolot Y., Souteyrand E., Cloarec J-P., Journal of Colloid Science and
Biotechnology, (2013) 2, 249.
[45] Vericat, C.; Vela, M. E.; Benitez, G.; Carro, P.; Salvarezza, R. C., Chemical Society reviews,
(2010) 39, 1805.
[46] G. E. Poirier, E. D. P., Science, (1996) 272, 1145.
[47] Onclin, S.; Ravoo, B. J.; Reinhoudt, D. N., Angew Chem Int Ed Engl, (2005) 44, 6282.
[48] Schilardi, P. L.; Dip, P.; dos Santos Claro, P. C.; Benitez, G. A.; Fonticelli, M. H.; Azzaroni, O.;
Salvarezza, R. C., Chemistry, (2005) 12, 38.
[49] Gang, T.; Yildirim, O.; Kinge, S.; Duan, X.; Reinhoudt, D. N.; Blank, D. H. A.; Rijnders, G.; van
der Wiel, W. G.; Huskens, J., Journal of Materials Chemistry, (2011) 21, 14800.
[50] Yildirim, O.; Gang, T.; Kinge, S.; Reinhoudt, D. N.; Blank, D. H.; van der Wiel, W. G.; Rijnders,
G.; Huskens, J., International journal of molecular sciences, (2010) 11, 1162.
[51] Guerrero, G.; Alauzun, J. G.; Granier, M.; Laurencin, D.; Mutin, P. H., Dalton Trans, (2013) 42,
12569.
[52] J. P. Collman, N. K. D., T. P. A. Eberspacher, C. E. D. Chidsey, Langmuir, (2006) 22, 2457.
[53] F. Tielens, E. S., J Phys Chem C, (2010) 114, 9444.
[54] S. Lee, W. L., C. Kuo, M. Karakachian, Y. Lin, B. Yu, J. Shyue, J Phys Chem C, (2010) 114, 10512.
[55] M. I. Béthencourt, L. S., P. Chinwangso, T. R. Lee, Langmuir, (2008) 25, 1265.
[56] R. K. Smith, S. M. R., P. A. Lewis, J. D. Monnell, R. S. Clegg, K. F. Kelly, L. A. Bumm, J. E.
Hutchison, P. S. Weiss, J Phys Chem C, (2001) 105, 1119.
[57] J. P. Folkers, P. E. L., G. M. Whitesides, Langmuir, (1992) 8, 1330.
[58] Yu, Q.; Chen, S.; Taylor, A. D.; Homola, J.; Hock, B.; Jiang, S., Sensors and Actuators B:
Chemical, (2005) 107, 193.
[59] Valiokas, R.; Malysheva, L.; Onipko, A.; Lee, H.-H.; Ruželė, Ž.; Svedhem, S.; Svensson, S. C. T.;
Gelius, U.; Liedberg, B., Journal of Electron Spectroscopy and Related Phenomena, (2009) 172, 9.
[60] Smith, R. K.; Lewis, P. A.; Weiss, P. S., Progress in Surface Science, (2004) 75, 1.
[61] L. Yan, C. M., A. Terfort, G. M. Whitesides, Langmuir, (1997) 13, 6704.
[62] J. Noh, T. M., K. Nakajima, H. Lee, M. Hara, J Phys Chem B, (2000) 104, 7411.
[63] F. Tielens, V. H., C-M. Pradier, M. Calatayud, F. Illas, Langmuir, (2009) 25, 9980.
[64] Pichon, B. P.; Barbillon, G.; Marie, P.; Pauly, M.; Begin-Colin, S., Nanoscale, (2011) 3, 4696.
[65] H. C. Kolb, M. G. F., K. B. Sharpless, Angew Chem Int Ed, (2001) 40, 2004.
[66] L. F. Tietze, A. S., Eur. J. Org. Chem., (1998) 189, 1629.
[67] K. V. Gothelf, K. A. J., Chemical reviews, (1998) 98, 863.
[68] J. P. Collman, N. K. D., C. E. D. Chidsey, Langmuir, (2004) 20, 1051.

48
[69] T. Lummerstorfer, H. H., J Phys Chem B, (2004) 108, 3963.
[70] Li, Y.; Wang, J.; Cai, C., Langmuir, (2011) 27, 2437.
[71] Upadhyay, A. P.; Behara, D. K.; Sharma, G. P.; Bajpai, A.; Sharac, N.; Ragan, R.; Pala, R. G.;
Sivakumar, S., ACS applied materials & interfaces, (2013) 5, 9554.
[72] Rianasari, I.; de Jong, M. P.; Huskens, J.; van der Wiel, W. G., International journal of
molecular sciences, (2013) 14, 3705.
[73] Toulemon, D.; Pichon, B. P.; Cattoen, X.; Man, M. W.; Begin-Colin, S., Chem Commun (Camb),
(2011) 47, 11954.
[74] Toulemon, D.; Pichon, B. P.; Leuvrey, C.; Zafeiratos, S.; Papaefthimiou, V.; Cattoën, X.; Bégin-
Colin, S., Chemistry of Materials, (2013) 25, 2849.
[75] Kinge, S.; Gang, T.; Naber, W. J. M.; van der Wiel, W. G.; Reinhoudt, D. N., Langmuir, (2011)
27, 570.
[76] Toulemon, D., Université de Strasbourg, (2013),
[77] Mark A. Olson, A. C., Rafal Klajn, Lei Fang, Sanjeev K. Dey, Kevin P. Browne, Bartosz A.
Grzybowski, J. Fraser Stoddart, Nano Letters, (2009) 9, 3185.
[78] Kinge, S.; Crego-Calama, M.; Reinhoudt, D. N., Chemphyschem : a European journal of
chemical physics and physical chemistry, (2008) 9, 20.
[79] V. Mahalingam, S. O., M. Peter, B. Jan Ravoo, J. Huskens, D. N. Reinhoudt, Langmuir, (2004)
20, 11756.
[80] J. Huskens, M. A. D., D. N. Reinhoudt, Angew Chem Int Ed, (2002) 47, 4467.
[81] Mulder, A.; Onclin, S.; Péter, M.; Hoogenboom, J. P.; Beijleveld, H.; ter Maat, J.; García-
Parajó, M. F.; Ravoo, B. J.; Huskens, J.; van Hulst, N. F.; Reinhoudt, D. N., Small, (2005) 1, 242.
[82] Christian A. Nijhuis, F. Y., Wolfgang Knoll, Jurriaan Huskens, David N. Reinhoudt, Langmuir,
(2005) 21, 7866.
[83] M. Mammen, S. C., G. M. Whitesides, Angew Chem Int Ed, (1998) 37, 2754.
[84] A. Mulder, J. H., D. N. Reinhoudt, Org. Biomol. Chem., (2004) 2, 3409.
[85] F. Pincet, E. P., Faraday Communications, (1995) 91, 4329.
[86] Wolfgang H. Binder, C. K., Marina Josipovic, Christoph J. Straif, Gernot Friedbacher,
Macromolecules, (2006) 39, 8092.
[87] Ronald Zirbs, F. K., Peter Hinterdorfer, and Wolfgang H. Binder, Langmuir, (2005) 21, 8414.
[88] Singh, P.; Ménard-Moyon, C.; Kumar, J.; Fabre, B.; Verma, S.; Bianco, A., Carbon, (2012) 50,
3170.
[89] Sessler, J. L.; Lawrence, C. M.; Jayawickramarajah, J., Chemical Society reviews, (2007) 36,
314.
[90] Viswanathan, K.; Long, T. E.; Ward, T. C., Langmuir, (2009) 25, 6808.
[91] Kalpana Viswanathan, H. O., Casey L. Elkins, Cheryl Heisey, Thomas C. Ward, Timothy E. Long,
Langmuir, (2006) 22, 1099.
[92] K. Viswanathan, H. O., C. L. Elkins, C. Hesey,T. C. Ward, T. E. Long, Langmuir, (2006) 22, 1099.
[93] Andrew K. Boal, F. I., Jason E. DeRouchey, Thomas Thurn-Albrecht, Thomas P. Russell &
Vincent M. Rotello, Nature, (2000),
[94] A. Csaki, R. M., J. M. Köhler, W. Fritzsche, Nucleic Acids Research, (2000) 28, 91.
[95] A. Csaki, R. M., W. Straube, J. M. Köhler, W. Fritzsche, Nucleic Acids Research, (2001) 29, 29.
[96] C. M. Niemeyer, B. C., S. Gao, L. Chi, S. Peschel, U. Simon, Colloid Polym Sci, (2001) 279, 68.
[97] Derfus, A. M.; von Maltzahn, G.; Harris, T. J.; Duza, T.; Vecchio, K. S.; Ruoslahti, E.; Bhatia, S.
N., Advanced Materials, (2007) 19, 3932.
[98] Mehmet Veysel Yigit, D. M., and Yi Lu, Bioconjugate Chem, (2008) 19, 412.
[99] Frédéric Geinguenaud, C. B., Antoine F. Carpentier, and Laurence Motte, Nanomaterials,
(2015) 5, 1588.
[100] Boal, A. K.; Frankamp, B. L.; Uzun, O.; Tuominen, M. T.; Rotello, V. M., Chemistry of Materials,
(2004) 16, 3252.
[101] Jin Seon Park, G. S. L., Yun-Jo Lee, Yong Soo Park, and Kyung Byung Yoon, JACS, (2002) 124,
13366.

49
[102] Bedanta, S.; Kleemann, W., Journal of Physics D: Applied Physics, (2009) 42, 013001.
[103] Roca, A. G.; Niznansky, D.; Poltierova-Vejpravova, J.; Bittova, B.; González-Fernández, M. A.;
Serna, C. J.; Morales, M. P., Journal of Applied Physics, (2009) 105, 114309.
[104] T. J. Daou, J. M. G., G. Pourroy, S. Buathong, A. Derory, C. Ulhaq-Bouillet, B. Donnio, D.
Guillon, and S. Begin-Colin, Chem Mater, (2008) 20, 5869.
[105] Fleutot, S.; Nealon, G. L.; Pauly, M.; Pichon, B. P.; Leuvrey, C.; Drillon, M.; Gallani, J. L.;
Guillon, D.; Donnio, B.; Begin-Colin, S., Nanoscale, (2013) 5, 1507.
[106] B. L. Frankamp, A. K. B., M. T. Tuominen, V. M. Rotello, JACS, (2005) 127, 9731.
[107] Poddar, P.; Telem-Shafir, T.; Fried, T.; Markovich, G., Physical Review B, (2002) 66,
[108] Kechrakos, D.; Trohidou, K. N., Journal of Nanoscience and Nanotechnology, (2008) 8, 2929.
[109] Dorothy Farrell, Y. C., R. William McCallum, Madhur Sachan, Sara A. Majetich, J Phys Chem C,
(2005) 109, 13409.
[110] C. Petit, Z. L. W., M. P. Pileni, J Phys Chem B, (2005) 109, 15309.
[111] Morup, S.; Tronc, E., Phys Rev Lett, (1994) 72, 3278.
[112] Trohidou, D. K. a. K. N., physical Review B, (1997) 58, 12169.
[113] Lafleur, J. P.; Jonsson, A.; Senkbeil, S.; Kutter, J. P., Biosensors & bioelectronics, (2016) 76,
213.
[114] Fan, X.; White, I. M.; Shopova, S. I.; Zhu, H.; Suter, J. D.; Sun, Y., Analytica chimica acta, (2008)
620, 8.
[115] Wood, R. W., Proceedings of the Physical Society of London, (1902) 18, 269.
[116] Fano, U., Journal of the Optical Society of America, (1941) 31, 213.
[117] Pines, D., Reviews of Modern Physics, (1956) 28, 184.
[118] Ritchie, R. H., Physical Review, (1957) 106, 874.
[119] Ritchie, R.; Arakawa, E.; Cowan, J.; Hamm, R., Physical Review Letters, (1968) 21, 1530.
[120] Otto, A., Zeitschrift ffir Physik, (1968) 216, 398.
[121] Raether, E. K. a. H., Zeitschrift für Naturforschung, (1968) 23, 2135.
[122] Zeng, S.; Baillargeat, D.; Ho, H. P.; Yong, K. T., Chemical Society reviews, (2014) 43, 3426.
[123] Ming Li, S. K. C. a. N. W., Analyst, (2015) 140, 386.
[124] Estevez, M. C.; Otte, M. A.; Sepulveda, B.; Lechuga, L. M., Analytica chimica acta, (2014) 806,
55.
[125] Roh, S.; Chung, T.; Lee, B., Sensors (Basel), (2011) 11, 1565.
[126] Szunerits, S.; Shalabney, A.; Boukherroub, R.; Abdulhalim, I., Reviews in Analytical Chemistry,
(2012) 31,
[127] Bryche, J.-F., (2016),
[128] Srivastava, T.; Das, R.; Jha, R., Plasmonics, (2012) 8, 515.
[129] A. Hoffmann, N. K., Z. Lenkefi, Z. Szentirmay, Surface science, (1996) 352-354, 1043.
[130] Stefan Löfas, M. M., Inger Rönnbergd, Esa Stenberg, Sensors and Actuators B: Chemical,
(1991) 5, 79.
[131] B. Liedberg, I. L., Sensors and Actuators B: Chemical, (1993) 11, 63.
[132] Mayer, K. M.; Hafner, J. H., Chemical reviews, (2011) 111, 3828.
[133] Amanda J. Haes, S. Z., Jing Zhao, George C. Schatz, Richard P. Van Duyne, JACS, (2006) 128,
10905.
[134] Sepúlveda, B.; Angelomé, P. C.; Lechuga, L. M.; Liz-Marzán, L. M., Nano Today, (2009) 4, 244.
[135] Zhang, D.; Liu, Q., Biosensors & bioelectronics, (2016) 75, 273.
[136] He, L.; Pagneux, Q.; Larroulet, I.; Serrano, A. Y.; Pesquera, A.; Zurutuza, A.; Mandler, D.;
Boukherroub, R.; Szunerits, S., Biosensors and Bioelectronics, (2017) 89, 606.
[137] Tao, N. J.; Boussaad, S.; Huang, W. L.; Arechabaleta, R. A.; D’Agnese, J., Review of Scientific
Instruments, (1999) 70, 4656.
[138] Radan Slavik , J. H., Jirı Ctyroky, Sensors and Actuators B, (1999) 54, 74.
[139] Monzón-Hernández, D.; Villatoro, J., Sensors and Actuators B: Chemical, (2006) 115, 227.
[140] Suzuki, A.; Kondoh, J.; Matsui, Y.; Shiokawa, S.; Suzuki, K., Sensors and Actuators B: Chemical,
(2005) 106, 383.

50
[141] Chien, F. C.; Lin, C. Y.; Yih, J. N.; Lee, K. L.; Chang, C. W.; Wei, P. K.; Sun, C. C.; Chen, S. J.,
Biosensors & bioelectronics, (2007) 22, 2737.
[142] Dostálek, J.; Homola, J.; Miler, M., Sensors and Actuators B: Chemical, (2005) 107, 154.
[143] G.G. Nenningera, P. T., J. Homolaa, S.S. Yee, Sensors and Actuators B, (2001) 74, 145.
[144] J. T. Hastings, J. G., P. D. Keathley, P. B. Kumaresh, Y. Wei, S. Law, L. G. Bachas, Optics Express,
(2007) 15, 17661.
[145] Slavík, R.; Homola, J., Sensors and Actuators B: Chemical, (2007) 123, 10.
[146] Jennifer S. Shumaker-Parry, C. T. C., Analytical chemistry, (2004) 76, 907.
[147] Alexander Zybin, C. G., Vladimir M. Mirsky, Jurgen Kuhlmann, Otto S. Wolfbeis, Kay Niemax,
Analytical chemistry, (2005) 77, 2393.
[148] Piliarik, M.; Vaisocherova, H.; Homola, J., Biosensors & bioelectronics, (2005) 20, 2104.
[149] Timothy M Chinowsky, M. S. G., Kyle S Johnston, Kjell Nelson, Thayne Edwards, Elain Fu, Paul
Yager, Biosensors & bioelectronics, (2007) 15, 2208.
[150] Hye Jin Lee, D. N., Robert M. Corn, Analytical chemistry, (2006) 78, 6504.
[151] Wen-Chuan Kuo, C. C., and Hsieh-Ting Wu, Optics Letters, (2003) 28, 1329.
[152] Law, W. C.; Markowicz, P.; Yong, K. T.; Roy, I.; Baev, A.; Patskovsky, S.; Kabashin, A. V.; Ho, H.
P.; Prasad, P. N., Biosensors & bioelectronics, (2007) 23, 627.
[153] Lin He, M. D. M., Sheila R. Nicewarner, Frank G. Salinas, Stephen J. Benkovic, Michael J.
Natan, and Christine D. Keating, JACS, (2000) 122, 9071.
[154] B.P. Nelson, T. E. G., M.R. Liles, R.M. Goodman, R.M. Corn, Analytical chemistry, (2001) 73, 1.
[155] Yao, X.; Li, X.; Toledo, F.; Zurita-Lopez, C.; Gutova, M.; Momand, J.; Zhou, F., Anal Biochem,
(2006) 354, 220.
[156] Besselink, G. A.; Kooyman, R. P.; van Os, P. J.; Engbers, G. H.; Schasfoort, R. B., Anal Biochem,
(2004) 333, 165.
[157] Chung, J. W.; Bernhardt, R.; Pyun, J. C., Sensors and Actuators B: Chemical, (2006) 118, 28.
[158] Teramura, Y.; Iwata, H., Anal Biochem, (2007) 365, 201.
[159] Percin, I.; Idil, N.; Bakhshpour, M.; Yilmaz, E.; Mattiasson, B.; Denizli, A., Sensors (Basel),
(2017) 17,
[160] Oh, B. K.; Kim, Y. K.; Park, K. W.; Lee, W. H.; Choi, J. W., Biosensors & bioelectronics, (2004)
19, 1497.
[161] J.W. WASWA, C. D. a. J. I., Journal of Food Process Engineering, (2006) 29, 373.
[162] A. V. Kabashin, P. E., S. Pastkovsky,W. Hendren, G. A.Wurtz, R. Atkinson, R. Pollard, V. A.
Podolskiy and A. V. Zayats, Nature Material, (2009) 8, 867.
[163] Sun, L.; Chen, P.; Lin, L., (2016),
[164] Mie, G., ANNALEN DER PHYSIK, (1908) 3, 377.
[165] Noguez, C., J Phys Chem C, (2007) 111, 3806.
[166] Focsan, M.; Campu, A.; Craciun, A. M.; Potara, M.; Leordean, C.; Maniu, D.; Astilean, S.,
Biosensors & bioelectronics, (2016) 86, 728.
[167] Pal, S. K. G. a. T., Chemical reviews, (2007) 107, 4797.
[168] Halas, N. J.; Lal, S.; Chang, W. S.; Link, S.; Nordlander, P., Chemical reviews, (2011) 111, 3913.
[169] Kadir Aslan, C. C. L., and Victor H. Pérez-Luna, J Phys Chem B, (2004) 108, 15631.
[170] Mei Li, K. K. W. W., and Stephen Mann, Chem Mater, (1999) 11, 23.
[171] Zeng, Z.; Mizukami, S.; Kikuchi, K., Analytical chemistry, (2012) 84, 9089.
[172] James J. Storhoff, A. A. L., Robert C. Mucic, Chad A. Mirkin, Robert L. Letsinger, and George C.
Schatz, J. Am. Chem. Soc., (2000) 122, 4640.
[173] Chad. A. Mirkin, R. L., R. Mucic, J. Storhoff, Letter to Nature, (1996) 382, 607.
[174] Szunerits, S.; Boukherroub, R., Chem Commun (Camb), (2012) 48, 8999.
[175] Stuart, D. A.; Haes, A. J.; Yonzon, C. R.; Hicks, E. M.; Van Duyne, R. P., IEE proceedings.
Nanobiotechnology, (2005) 152, 13.
[176] Barbillon, G., Micromachines, (2012) 3, 21.
[177] Barbillon, G.; Bijeon, J. L.; Plain, J.; Royer, P., Thin Solid Films, (2009) 517, 2997.

51
[178] Chen, S.; Svedendahl, M.; Kall, M.; Gunnarsson, L.; Dmitriev, A., Nanotechnology, (2009) 20,
434015.
[179] Duyne, A. J. H. a. R. P. V., JACS, (2002) 124, 10596.
[180] G. Barbillon, J.-L. B., J. - S. Bouillard, J. Plain, M. Lamy De La Chapelle, P.-M. Adam, P. Royer,
Journal of Microscopy, (2007) 229, 270.
[181] Akjouj, A.; Lévêque, G. t.; Szunerits, S.; Pennec, Y.; Djafari-Rouhani, B.; Boukherroub, R.;
Dobrzyński, L., Surface Science Reports, (2013) 68, 1.
[182] A. Shalabney, I. A., Optics Letters, (2012) 37, 1175.
[183] Hong-Yu Lin, W.-H. T., Yu-Chia Tsao, and Bor-Chiou Sheu, Applied Optics, (2007) 46, 800.
[184] Ma, Y.; Farrell, G.; Semenova, Y.; Chan, H. P.; Zhang, H.; Wu, Q., Photonics and
Nanostructures - Fundamentals and Applications, (2014) 12, 69.
[185] Yee, R. C. J. a. S. S., Sensors and Actuators B, (1993) 12, 213.
[186] Ming-Hung Chiu, S.-F. W. a. R.-S. C., Optics Letters, (2005) 30, 233.
[187] Oliner, A. H. a. A. A., Applied Optics, (1965) 4, 1275.
[188] Lakowicz, J. R., Plasmonics, (2006) 1, 5.
[189] Méjard, R.; Dostálek, J.; Huang, C.-J.; Griesser, H.; Thierry, B., Optical Materials, (2013) 35,
2507.
[190] Burke, J. J.; Stegeman, G. I.; Tamir, T., Physical Review B, (1986) 33, 5186.
[191] Slavík, R.; Homola, J., Optics Communications, (2006) 259, 507.
[192] Shalabney, A.; Abdulhalim, I., Laser & Photonics Reviews, (2011) 5, 571.
[193] Sarkar, M.; Besbes, M.; Moreau, J.; Bryche, J.-F.; Olivéro, A.; Barbillon, G.; Coutrot, A.-L.;
Bartenlian, B.; Canva, M., ACS Photonics, (2015) 2, 237.
[194] Linda S. Jung, C. T. C., Timothy M. Chinowsky, Mimi N. Mar, and Sinclair S. Yee, Langmuir,
(1998) 14, 5636.
[195] Huang, D.-W., Optical Engineering, (2010) 49, 054403.
[196] P. K. Jain, M. A. E.-S., Nano Letter, (2008) 8, 4347.
[197] Rios, F.; Smirnov, S., ACS applied materials & interfaces, (2009) 1, 768.
[198] J. Spinke, M. L., H-J. Guder, L. Angermaier, W. Knoll, Langmuir, (1993) 9, 1821.
[199] V. H. Pérez-Luna, M. J. O. B., K A. Opperman, P. D. Hampton, Gabriel P. Lopez, L A. Klumb, P.
S. Stayton, JACS, (1999) 121, 6469.
[200] L. S. Jung, K. E. N., C. T. Campbell, P. S. Stayton, S. S. Yee, V. Pérez-Luna, G. P. Lopez, Sensors
and Actuators B, (1999) 54, 137.
[201] Chiu, N. F.; Fan, S. Y.; Yang, C. D.; Huang, T. Y., Biosensors & bioelectronics, (2017) 89, 370.
[202] Sabine Szunerits, R. B., Langmuir, (2006) 22, 1660.
[203] Lahav, A., Journal of Nanophotonics, (2009) 3, 031501.
[204] Wijaya, E.; Lenaerts, C.; Maricot, S.; Hastanin, J.; Habraken, S.; Vilcot, J.-P.; Boukherroub, R.;
Szunerits, S., Current Opinion in Solid State and Materials Science, (2011) 15, 208.
[205] L. Andrew Lyon, D. J. P., and Michael J. Natan, J. Phys. Chem. B, (1999) 103, 5826.
[206] L. Andrew Lyon, M. D. M., and Michael J. Natan, Anal. Chem, (1998) 70, 5177.
[207] Besselink, G. A. J.; Kooyman, R. P. H.; van Os, P. J. H. J.; Engbers, G. H. M.; Schasfoort, R. B.
M., Analytical Biochemistry, (2004) 333, 165.
[208] Ryken, J.; Li, J.; Steylaerts, T.; Vos, R.; Loo, J.; Jans, K.; Van Roy, W.; Stakenborg, T.; Van Dorpe,
P.; Lammertyn, J.; Lagae, L., Sensors and Actuators B: Chemical, (2014) 200, 167.
[209] Sun, Y.; Bai, Y.; Song, D.; Li, X.; Wang, L.; Zhang, H., Biosensors & bioelectronics, (2007) 23,
473.
[210] Scott D. Soelberg, R. C. S., Ajit P. Limaye, and Clement E. Furlong, Analytical chemistry, (2009)
81, 2357.
[211] Wang, J.; Zhu, Z.; Munir, A.; Zhou, H. S., Talanta, (2011) 84, 783.
[212] Sadrolhosseini, A. R.; Naseri, M.; Kamari, H. M., Optics Communications, (2017) 383, 132.
[213] Zamfir, L.-G.; Geana, I.; Bourigua, S.; Rotariu, L.; Bala, C.; Errachid, A.; Jaffrezic-Renault, N.,
Sensors and Actuators B: Chemical, (2011) 159, 178.
[214] Lee, K. S.; Lee, M.; Byun, K. M.; Lee, I. S., Journal of Materials Chemistry, (2011) 21, 5156.

52
[215] Garibo, D.; Campbell, K.; Casanova, A.; de la Iglesia, P.; Fernández-Tejedor, M.; Diogène, J.;
Elliott, C. T.; Campàs, M., Sensors and Actuators B: Chemical, (2014) 190, 822.
[216] Teramura, Y.; Arima, Y.; Iwata, H., Anal Biochem, (2006) 357, 208.
[217] Park, J.-W.; Min, H.; Kim, Y.-P.; Kyong Shon, H.; Kim, J.; Moon, D. W.; Lee, T. G., Surface and
Interface Analysis, (2009) 41, 694.
[218] Kim, Y.-P.; Hong, M.-Y.; Shon, H. K.; Chegal, W.; Cho, H. M.; Moon, D. W.; Kim, H.-S.; Lee, T.
G., Applied Surface Science, (2008) 255, 1110.

53
54
II. CHAPITRE II - Structuration of Iron Oxide Nanoparticle
Assemblies

55
 The aim of this chapter is to study the synthesis and the assembly of iron oxide nanoparticles
onto gold substrates. The synthesis of nanoparticles allows a good control on the size and the shape
of the nanoparticles in order to tune their intrinsic properties. Nanoparticles were assembled onto a
surface through covalent binding by performing the CuAAC “click” chemistry which presents the
advantages to be robust, highly selective, versatile and to create stable assemblies. Finally, the
magnetic properties of nanoparticles were studied as function of their size and of their spatial
arrangement.

A. Nanoparticle synthesis

Iron oxide nanoparticles have been synthesized by thermal decomposition. This method allows
controlling a narrow size distribution, a precise control on the morphology and a good stability in
suspension. Spherical 10 nm-sized nanoparticles have been used as reference since their stability in
organic solvent is very good and their synthesis and purification are well controlled. Furthermore, 5
nm and 20 nm-sized nanoparticles were synthesized by adapting carefully the experimental conditions.

1. Experimental details

a) Nanoparticles synthesis conditions

The nanoparticles synthesis is well mastered in our laboratory.[1, 2] An iron stearate precursor
was decomposed (2.2 mmol, 1.384 g) in 20 mL of dioctylether (T boil = 288°C) in the presence of oleic
acid as a surfactant (4.4 mmol, 1.25 g). The temperature was set thanks to a thermocouple dipped in
the solution and controlled by a computer. The mixture was heated at 110°C without a condenser and
stirred to solubilize the reactants and to remove the water traces. Then, the mixture was heated up
with a condenser with a ramp of 5°C per minute until the boiling temperature was reached. The
mixture remained at this temperature for 2 hours and then was cooled down around 100°C. Figure II-
1.b shows the target temperature and the temperature measured by the thermocouple. The resulting
synthesized nanoparticles coated with oleic acid are noted NP10.

56
 a) Condenser b)

Computer
control

Heating

Figure II-1. a) Schematic representation of the thermal decomposition process and b) temperature profile of the setpoint
and of the temperature of the reaction media

The thermal decomposition method allows controlling the size of nanoparticles by varying
synthesis parameters such as solvent, temperature ramp or quantity of surfactant.[2] Therefore,
spherical nanoparticles with a diameter of about 20 nm and 5 nm, and denoted NP20 and NP5
respectively, were also synthesized. The table II-1 summarizes the synthesis parameters which allow
controlling the size of the nanoparticles.
Table II-1. Summarize of the synthesis parameters as function of the desired size of nanoparticles

NP5 NP10 NP20

Squalane +
Solvent Dioctylether Dioctylether
dibenzylether
(boiling temperature) (288°C) (288°C)
(470°C)

Oleic acid No yes yes

The synthesis of 5 nm nanoparticles follows the same procedure given for the 10 nm
nanoparticles except that no oleic acid was added in the reaction medium. Only the stearate chains
present in the iron precursor will act as capping agent. Without oleic acid, which is used as surfactant
but is also known to stabilized the iron stearate precursor (leading to its decomposition at higher
temperature), the iron precursor will decompose faster and create more nuclei in the solution, thus
giving smaller nanoparticles.[3]

The synthesis of 20 nm nanoparticles follows a similar protocol as reported previously for NP10
in the exception of squalane (Tboil = 470°C) which was used instead of dioctylether. The iron stearate
(2.9 mmol, 1.832 g) were decomposed in 19.5 mL of squalane and 0.5 mL of dibenzylether in presence
of oleic acid (6.6 mmol, 1.89 g). The temperature ramp was slightly different with a first step of 1 hour
at 120°C followed by a ramp of 5°C per minute and a last step at 330°C for 1 hour. The mixture was
then cooled down to 90°C.

57
 b) Purification of the nanoparticles

The purification of the nanoparticles is an essential step to remove the excess of oleic acid, unreacted
iron stearate and side products resulting from the thermal decomposition process. The control of free
molecules in solution is also a crucial point to ensure a good stability of the suspension. Indeed, oleic
acid molecules are characterized by an equilibrium between grafted molecules at the surface of
nanoparticles and free molecules in the solvent.[4] The stability of the nanoparticle suspension is
therefore dependent of this equilibrium.

The purification was realized by centrifugation to precipitate the nanoparticles. A mixture of

solvents with different polarities was used to separate the nanoparticles from the free molecules.
Chloroform was used because of its good affinity to the oleic acid coated nanoparticles. In contrast,
acetone favored the precipitation of nanoparticles. The chloroform/acetone mixture (1:5, v:v) was
added to the reaction medium and was centrifuged at 14 000 rpm for 5 minutes in order to separate
nanoparticles from the other products which remained in the supernatant. This purification step was
performed several times and was monitored by infrared spectroscopy until the disappearance of free
oleic acid and stearate molecules. FT-IR spectra were recorded for the NP10 sample in order to monitor
removal of oleic acid (figure II-2).

Figure II-2. Infrared spectra of NP10 after 3 washes (black curve), 10 washes (red curve), 17 washes (blue curve) and 22
washes (green curve)

Infrared spectra show the vibrational bands localized in NP10 coated with oleic acid. The broad
band localized between 500 cm-1 and 800 cm-1 is characteristic of the stretching vibrational modes of
Fe-O bonds. The bands corresponding to the symmetric and asymmetric stretching of the C-H bonds
of the oleic acid and the stearate are localized at 2850 cm-1 and 2920 cm-1 respectively.[5] The rocking
and bending modes of the C-H bonds are localized at 1460 cm-1 and 720 cm-1, respectively. The
vibrational band at 2960 cm-1 corresponds to the stretching of the C-H binding of the methylene (CH3)
terminal groups. At 1710 cm-1 a narrow band is representative of the C=O bonds of the free oleic acid

58
in solution. The vibrational band at 1660 cm-1 and 1420 cm-1 are representative of the oleic acid grafted
on the nanoparticle surface and correspond to the asymmetric νas(COO-) and symmetric νs(COO-)
stretching modes, respectively.

Figure II-3. Infrared spectra of NP10 after washing zoomed in the 400 cm-1 - 2000 cm-1 area

The purification step is evaluated as function of the peak area ratios between the different
vibrational bands (table II-2). The nanoparticles coated with oleic acid were purified until the
disappearance of the band corresponding to the free oleic acid C=O (1710 cm-1). The intensity ratio
between the ν(Fe-O) band of iron oxide (580 cm-1) and the ν(C-H) band of the alkyl chains (2920 cm-1)
has also to be close to one.
Table II-2. Surface area ratios between ν(C-H), ν(C=O), ρ(C-H) and ν(Fe-O) vibrational bands.

ν(C=O)/ν(Fe-O) ν(C-H)/ν(Fe-O) ρ(C-H)/ν(Fe-O)

3 washes 0.15 0.94 0.04
10 washes 0.07 0.68 0.02
17 washes 0.02 0.30 0.01
22 washes < 0.01 0.20 <0.01

The disappearance of the ν(C=O) and the ρ(C-H) bands after 22 washes and the different ratios
indicates that most of the free molecules in solution were removed. This purification step is different
for all synthesis and the number of washes can be strongly dependent of the environmental conditions
(hygroscopy, local temperature) in the laboratory. Moreover, the used of warm acetone (pre-heated
before the centrifugation) led to a faster removal of the organic species which reduce considerably the
number of washes.[3]

59
 Infrared spectra were also recorded for the NP5 and NP20 after several washes to get “clean”
nanoparticles. The washing step consists in removing the maximum of free molecules while avoiding
the aggregation. These washing steps are strongly dependent of the size of the nanoparticles since the
surface composition and therefore the binding of the carboxylic acid at the surface of iron oxide is
different (cf. II.A.2.a). Infrared spectra of nanoparticles after washing was completed and shown in the
figure II-4.

Figure II-4. Infrared spectra of NP5 (black curve), NP10 (red curve) and NP20 (blue curve) after the washing step was
completed

The NP5 and NP20 were washed 11 times and 6 times respectively until the ν(Fe-O)/ν(C-H)
ratio was close to one. The NP5 and NP20 still present the band of the free oleic acid at 1710 cm -1 in
low proportion. However, more washes of these nanoparticles lead to their aggregation.

The strongest dipolar interactions in the NP20 led to the formation of aggregates. An excess of
oleic acid is needed and therefore, the washing steps have to be stopped much earlier than for NP10.
In the case of NP5, the smaller diameter of the nanoparticles is correlated to a lower surface. Given
the weak interactions between the carboxylic acid group and the iron oxide surface, the quantity of
free oleic acid in solution should be higher to preserve the equilibrium between the ligand free and
adsorbed at the nanoparticle surface which rule the stability of NP5 suspension.

Furthermore, the ν(Fe-O) band of iron oxide is located at 590 cm-1 for each size of
nanoparticles. A closer look shows that the NP5 and NP10 are closer from the value of the maghemite
(638 cm-1) which shows an evolution of the composition. Indeed, smaller nanoparticles exhibit a higher
surface/volume ratio which favors the oxidation of magnetite in maghemite (cf. II.A.2.c).[1] Moreover,
a component at 630 cm-1 can be observed especially for the NP5 and NP20 which show the presence
of maghemite. Nevertheless, all the samples present a layer of maghemite due to the surface
oxidation[1] and the determination of the quantity of maghemite with infrared spectroscopy is
complicated because of the method of preparation which can promote the oxidation of magnetite.

60
 2. Structural characterizations

a) Size and morphology of the nanoparticles

The size and the morphology of the nanoparticles were characterized by transmission electron
microscopy (TEM). The size distribution was calculated by measuring more than 200 nanoparticles
using the ImageJ software and was fitted by a Gaussian function. The analysis was done on NP5, NP10
and NP20 (figure II-5).

a)

b)

c)

Figure II-5. TEM images of oleic acid coated nanoparticles (NP@OA) and the corresponding size distribution. a) NP5, b)
NP10 and c) NP20

TEM images showed spherical morphologies for each size of nanoparticles. High resolution

narrow for each size of nanoparticles and are centered to 5. 1 ± 0.8 nm (16 % deviation), 10.1 ± 1.1
images showed not completely spherical objects, but faceted nanoparticles. The size distributions are

61
nm (11% deviation) and 21.2 ± 1.2 nm (7% deviation) for NP5, NP10 and NP20, respectively. All
nanoparticles present a rather low standard deviation (below 16%). The 20 nm nanoparticles present
more facets which have been previously observed.[6, 7] Indeed, the crystallographic planes do not
present the same surface energy and, therefore, grow at different rates which induce facets which are
more visible for the largest nanoparticles.

b) Colloidal stability of nanoparticle suspension

The preparation of a stable suspension of nanoparticles is an essential step to assemble them

onto a surface. The colloidal stability of the suspension has been investigated by granulometry
measurements by dynamic light scattering (DLS) in order to determine their distribution of
hydrodynamic diameters (figure II-6).

a) b)

c)

Figure II-6. Hydrodynamic diameter distributions in volume measured by granulometry and histogram of size
distributions determined by TEM of a) NP5, b) NP10 and c) NP20

The hydrodynamic diameters were measured in tetrahydrofuran (THF) solvent which will be
used for the assembly step. In each case, the hydrodynamic diameter distribution is monomodal with
a low polydispersity index which agree with a single population of nanoparticles in suspension. The
hydrodynamic diameters centered at 8.7 nm, 12.5 nm and 23.8 nm for NP5, NP10 and NP20,
respectively, shows that the nanoparticles were not aggregated since these values are similar to TEM
sizes. A slight shift of 2-3 nm can be observed between the size distribution measured on the TEM

62
images and the peak position of the value given by DLS. Indeed, the measurement of the size by TEM
gives the diameter of the inorganic core of the nanoparticles whereas the DLS measurement gives the
hydrodynamic diameter which includes the inorganic core, the organic shell corresponding to the oleic
acid molecules and the solvation sphere. This analysis shows the good colloidal stability of the
nanoparticles in THF.

c) Structural characterization of the nanoparticles

Nanoparticles were characterized by X-ray diffraction to determine the crystalline phase and
their composition. This characterization allows also calculating the size of diffracting domains and the
lattice parameter. Figure II-7 shows the XRD patterns of nanoparticles of different sizes.

Figure II-7. XRD patterns of NP5 (black curve), NP10 (red curve) and NP20 (blue curve). Calibration silicon peaks are
noted with a star
The narrow peaks (marked with a star) are ascribed to silicon powder which was added to
calibrate XRD patterns in order to perform Rietveld refinements.2 The diffraction peaks observed for
each size of nanoparticle corresponds to the spinel structure. The peak positions between magnetite
and maghemite being very close, the refinement with Fullprof software allowed determining the lattice
parameters and the mean crystallite sizes (table II.3).
Table II-3. Structural parameters of the nanoparticles

NP5 NP10 NP20

Lattice parameter (Å) 8.360 ± 0.002 8.379 ± 0.002 8.391 ± 1.000
Crystallite size (nm) 3±1 10 ± 1 13 ± 1

2
Collaboration with C. Lefevre (IPCMS)

63
 The lattice parameters of the three sizes of nanoparticles are intermediate between the lattice
parameters of the stoichiometric magnetite (8.396 ± 0.001 Å, JCPDS file n°00-019-0629) and
maghemite (8.338 ± 0.001 Å, JCPDS file n°00-013-0458) phases. However, the larger nanoparticles
(NP20) present a lattice parameter closer to the magnetite than the maghemite. In contrast, the
smallest nanoparticles display a cell parameter much closer to the one of maghemite. Indeed, NP5 and
NP10 are more sensitive to the surface oxidation[6]. As seen in chapter I, the thickness of the
maghemite layer at the surface of the nanoparticle is about 2-3 nm and therefore, for small
nanoparticles, the contribution of the oxidized layer is the most important.

The XRD patterns show that for larger nanoparticles, the diffraction peaks are narrower
meaning larger diffracting domains. Indeed, the crystallite sizes increase with the size of nanoparticles.
For NP5 and NP10 the crystallite sizes are similar to the measured sizes with TEM which confirms that
the nanoparticles are single crystal. For the NP20, the crystallite size is about 13 nm which is lower
than the measured size at TEM (21.2 nm). This has been attributed to the presence of lattice
dislocations induced by the oxidation of nanoparticles and explains the smaller structural coherence
length compared to TEM size.[3, 8, 9]

3. Magnetic characterizations

a) Magnetic measurements as function of the applied field

Magnetic characterizations were performed on NP5, NP10 and NP20 by using a SQUID
(Superconducting QUantum Interference Device) magnetometer3. Magnetization cycles were
obtained by varying an external field between -5 tesla and 5 tesla. The measurements were performed
at 300 K and 5 K on samples in the powder state (figure II-8). Saturation magnetization was calculated
by measuring the concentration of Fe3O4 in solution by atomic absorption spectroscopy (AAS) (see
appendix A).

3
Collaboration with Lise-Marie Lacroix, LPCNO, Toulouse

64
a) b)

Figure II-8. Magnetization curves as function of applied field for different sizes of nanoparticles at a) 300 K and b) 5 K

No hysteresis loops (HC = 0 Oe and MR/MS =0) were observed at 300 K for the different sizes of
nanoparticles which agree with a superparamagnetic behavior. At low temperature (5 K), the cycles
present an opening indicating than magnetic moments are blocked. The saturation magnetization, the
coercive field and the relative remanent magnetization are extracted from the magnetization curves
and summarized in table II-4.
Table II-4. Coercive fields HC, relative remanent magnetization MR/MS and saturation magnetization MS for the different
sizes of nanoparticles at 5 K

Sample HC (Oe) MR/MS MS (emu/g)

NP5 240 ± 30 0.36 ± 0.05 55 ± 5
NP10 495 ± 30 0.35 ± 0.05 55 ± 5
NP20 330 ± 30 0.16 ± 0.05 75 ± 5

The saturation magnetization is lower than the magnetite in the bulk state (92 emu/g). This
lower magnetization can be explained by the presence of the oxidized layer at the nanoparticle surface.
Indeed, the saturation magnetization of the maghemite (74 emu/g) is lower than the magnetite. The
lower MS for smaller nanoparticles is due to the composition closer from the maghemite as shown by
the structural characterizations. Moreover, defects and surface effect may cause the lower
magnetization. A disorder of the magnetic moments can exist at the nanoparticle surface. This
phenomenon is called spin canting and induces a diminution of the saturation magnetization.[1, 7] For
smaller nanoparticles, the contribution of the spin canting is more important which explains the lower
MS for NP5 and NP10.

The remanent magnetization is the same between NP10 and NP5 (MR/MS = 0.35) and still
below the theoretical value for a random orientation (MR/MS = 0.5). However this ratio is lower for the
NP20 (MR/MS = 0.16) which results from the strong dipolar interactions which tends to align the
magnetic moments. Indeed, the theoretical value corresponds to a random orientation, but the
increase of the anisotropy (with the increase of the size for example) can be responsible of the
decrease of the MR/MS ratio.

65
 The coercive field of NP10 (495 Oe) is larger than for the NP20 (330 Oe) and the NP5 (240 Oe).
This increase of the coercive field with the size of the nanoparticles but decrease for size larger than
10 nm is expected from the literature.[10-12] Moreover, the values of HC are in accordance with the
values found for similar sizes.[6, 7, 10, 13] This behavior is explained by the fact that we consider here
nanoparticles with a single magnetic domain and with increase of the size the required field to reverse
the magnetization increases. Nevertheless, the decrease of the coercive field for NP20 may be related
to structural defects in NP20.

The magnetic and structural characterizations of the NP20 can show an ambiguity to their
composition and can be relevant of a core/shell structure. The presence of a wüstite core has already
been observed on large or cubic nanoparticles (e.g. FexO@Fe3-δO4 core/shell nanoparticles).[6, 7, 14]
Moreover, XRD being sensitive to composition above 5 % in mass, we cannot rule out the presence of
small amount of wüstite within the core of nanoparticles. Wüstite is antiferromagnetic (AFM) and its
coupling with ferrimagnetic (FiM) material may induce exchange bias coupling between interfacial
AFM and FiM spins.[15] The exchange bias can be observed easily by the shift of the hysteresis loop
measured after cooling down under a magnetic field which correspond to the exchange field:

¹¶ =
|¹ª
» |œ¼¹» ¼
½

‡
Equation II.1

Therefore, magnetization cycles were performed after cooling down to 5 K under an external
field of 7 T. This measurement was compared with the cycle at 5 K without external field (figure II-9).

Figure II-9. Magnetization curves as function of applied field recorded at 5K after zero field cooling (ZFC) (blue curve) and
field cooling (FC) (green curve) at 7 tesla.

The exchange field measured on the sample is He = 20 Oe. This low exchange field is not
significant of a wüstite/magnetite core/shell structure as shown previously.[14] Nevertheless, this small
field can be induce by the coupling between the magnetite and the external spin canting layer.[1]

b) Magnetic measurements as function of the temperature

66
 Magnetization measurements were also performed as a function of the temperature (figure II-
10.a). First, magnetization was recorded from 5 K to 350 K after cooling down the sample without
applying any magnetic field (Zero Field Cooled, ZFC). Second, magnetization was also recorded from 5
K to 350 K but after cooling the sample under an external magnetic field of 75 Oe (Field Cooled, FC).
The ZFC and FC curves allow identifying the blocking temperature (TB) of the samples in the powder
state. The blocking temperature is often assimilated to the maximum of the ZFC curve. However, this
value is overestimated because the ZFC curve reflects a distribution of magnetic anisotropies which
depend on the size distribution of the nanoparticles and of the dipolar interactions. Therefore, it is
much accurate to consider that the blocking temperature corresponds to the inflection point of the
ZFC curve. This inflection point can be determined by the maximum of the derivative of the difference
between ZFC and FC curve (figure II-10.b).[14] Therefore the maximum of the ZFC curve is named TMax.

a) b)

Figure II-10. a) Magnetization as function of temperature (ZFC/FC curves) for different sizes of nanoparticles and b)
temperature derivative of the difference between ZFC and FC

Table II-5 summarizes TMax and TB values of the nanoparticles. These values increase with the
size of the nanoparticles. Indeed, nanoparticles with larger sizes increase the magnetocrystalline
energy and, therefore, a higher thermal energy (kBT) is required to pass the energy barrier KV (see
chapter I).[6, 13] In the case of NP20, the M(H) curve recorded at 300 K agree with a blocking temperature
below 300 K. Therefore, TMax which is larger than 350 K is not coherent with a superparamagnetic
behavior.
Table II-5. TMax and TB determined by ZFC/FC curves for different sizes of nanoparticles

Sample TMax TB
NP5 50 K 20 K
NP10 110 K 75 K
NP20 > 350 K 185 K, 90 K, ≈20 K

The values of TB determined with the derivative of the difference between ZFC and FC curves
are values lower than the TMax. The approximation of TB corresponding to the maximum of the ZFC
values here show clearly an error on its value. Here, the TB for each nanoparticle size is below 300 K
and is coherent with the superparamagnetic behavior at room temperature showed by M(H) curves.
The TB for NP5 and NP10 (20 K and 75 K respectively) are in accordance with their sizes whereas the

67
NP20 at 185 K is slightly lower than expected.[6] Once again, this behavior can be caused by defect in
the crystal structure of the nanoparticles. Moreover, NP20 present different contributions in the ZFC
curve. A first one can be observed around 90 K and can be attributed at the Verwey transition which
corresponds to a structural phase transition and can modified the physical properties of the magnetite
such as its conductivity, calorific capacity and magnetization. This transition is visible for large
nanoparticles in the ZFC curve.[16] A second contribution is observed around 20 K and can be attributed
to the canted layer at the surface of the nanoparticles.[3]

4. Conclusion

Iron oxide nanoparticles with different sizes have been synthesized by thermal decomposition.
Their structural and magnetic properties have been studied. The structural characterizations showed
that the size of the nanoparticles have an influence on their composition. Small nanoparticles favor
the maghemite phase which has for main consequence to reduce saturation magnetization. In
contrast, large nanoparticles favor the contribution of magnetite and larger magnetization saturation.
Nevertheless, nanoparticles of 5 nm, 10 nm and 20 nm in the powder state display the
superparamagnetic behavior at room temperature. The saturation magnetization and the blocking
temperature increases with the size of the nanoparticles. Therefore, larger nanoparticles favor
magnetic anisotropy collective and stronger dipolar interactions which may alter their stability as
suspensions.

68
 B. Nanoparticle and substrate functionalization

To produce highly stable nanoparticle assemblies, iron oxide nanoparticles were covalently
bound onto planar substrates. The CuAAC “click” reaction which is well known for its versatility and
selectivity is well suited to assemble the iron oxide nanoparticles into stable single monolayer.[17-19]
Moreover, this assembling method allows the control of the assembly structuration.[20, 21] Assemblies
of nanoparticles by “click” chemistry involve the functionalization of nanoparticles and substrates by
alkyne and azide complementary groups, respectively. These groups have to be available at the surface
to ensure the complete reaction. Preliminary studies in our team showed that the best system consist
in a gold substrate functionalized with alkyne groups while nanoparticles exhibit azide groups at their
surface.[6]

1. Nanoparticle functionalization

In order to graft a specific group required for the CuAAC “click” reaction, a ligand exchange has
to be performed to replace oleic acid on the nanoparticle surface with a molecule carrying an azide
functional group. Functionalized nanoparticles are then characterized in order to control the ligand
replacement at their surface as well as the suspension stability.

a) Experimental section

A direct ligand exchange was performed to replace the oleic acid by 12-azido-dodecyl-
phosphonic acid (AP12N3). A phosphonic acid is used because the interactions with the iron oxide are
stronger than with the carboxylic acid.[22-26] Therefore, phosphonic acid will spontaneously replace the
oleic acid which results in the production of azido-terminated nanoparticles (NP@N3). Moreover, the
long alkyne chains allow the stability of the nanoparticle suspensions in THF while the azide groups are
directly available from the surface. The ligand exchange was performed in solution (figure II-11). 20
mg of AP12N3 was solubilized in THF before adding a volume of 10 mL at 5 mg/mL of NP@OA. Then,
the suspension was stirred for 16 hours.

69
Figure II-11. Schematic representation of the preparation of NP@N3 by performing the ligand exchange process

NP@N3 were purified in order to remove the excess of phosphonic acid molecules which could
interfere during the click reaction and the oleic acid in solution desorbed from the nanoparticle
surface. The purification step was realized by ultrafiltration which consists in the filtration of the
nanoparticle suspension through a cellulose membrane (30 kDa). The suspension is pushed by putting
pressure with argon flux in order to eliminate the free molecules and to remain the nanoparticles
which are too large to go through the membrane. The purification step is monitored by FT-IR
spectroscopy to ensure the disappearance of the free molecules in the suspension.

b) Characterization after functionalization

It is essential to be certain that the nanoparticles have still a good stability and do not present
any aggregation prior to their assembly through CuAAC “click” chemistry. Granulometry
measurements were performed to ensure having non aggregated nanostructures in suspension after
the ligand exchange. Characterizations here are presented for NP10 which are the reference
nanoparticles, NP5 and NP20 characterizations are presented in appendix C and the values of
hydrodynamic diameter are summarized in table II-6. Figure II-12 shows a monomodal hydrodynamic
diameter after the exchange with the phosphonic acid meaning a good stability of the suspension in
THF. The average value of the hydrodynamic diameter (11.8 nm) is slightly smaller than before the
ligand exchange (13.5 nm). It can be explained by a shorter length of the alkyl chains of AP12N3 (12
carbons) than oleic acid (18 carbons).

70
 Figure II-12. Granulometry measurements of the nanoparticles in THF before (black curve) and after (blue curve) ligand
exchange

Table II-6. Summarize of the hydrodynamic diameter for nanoparticles before (NP@OA) and after functionalization
(NP@N3)

NP5 NP10 NP20

NP@OA 8.7 nm 13.5 nm 23.8 nm
NP@N3 6.5 nm 11.8 nm 21.2 nm

To confirm the correct grafting of the AP12N3 on the nanoparticles, FT-IR spectroscopy
measurements were performed before and after the ligand exchange and on AP12N3 (figure II-13).

71
 Figure II-13. Infrared spectra of the nanoparticles with oleic acid (black curve), the free molecule of AP12N3 (red curve)
and the NPs functionalized with the azide phosphonic acid (blue curve)

Infrared spectra of nanoparticles functionalized with oleic acid (black curve) present the
characteristic peaks of iron oxide (ν(Fe-O) - 600 cm-1), alkyne chains (νas(CH2) - 2920 cm-1 and νs(CH2) -
2850 cm-1) and oleic acid grafted at their surface (νas(COO-) 1660 cm-1 and νs(COO-) 1420 cm-1) as
described previously. After the functionalization step (blue curve), the infrared spectra of
nanoparticles do not show the carboxylic vibrational bands anymore. Moreover, a singular and
recognizable band at 2100 cm-1 is attributed to the vibrational stretching mode of the azide group
(ν(N≡N)).[5] The apparition of a wide band between 900 cm-1 and 1150 cm-1 is representative of the
phosphonic acid bound to the iron oxide vibration (ν(Fe-O-P)).[23] Indeed, the signal of the free
molecule of phosphonic acid (red curve), shows characteristic bands of the stretching mode of the
azide groups at 2100 cm-1. Moreover, the stretching modes within the P-OH binding (ν(P-O)) at 1008
cm-1 and at 945 cm-1 and the supplementary vibrational band observed at 1110 cm-1 corresponding to
the P=O stretching mode are representative of the phosphonic acid function.[6] The replacement of
the peaks at 945 cm-1 and 1010 cm-1 by the wide band located at 1015 cm-1 indicates the correct
grafting of the AP12N3 on the nanoparticle surface and the absence of free molecules of phosphonic
acid in solution.

The infrared spectroscopy confirmed the replacement of the oleic acid by the AP12N3.
Moreover, the purification step was done successfully with the ultrafiltration method and allowed
preparing NP@N3 with functional groups available at their surface. The DLS measurements confirmed
the good colloidal stability after the ligand exchange, meaning suspensions ready for the “click”
chemistry reaction.

2. Substrate functionalization

72
 The gold surface has to be functionalized with the complementary groups for the “click”
chemistry reaction. Azide groups have been localized on the nanoparticle surface, alkyne groups have
to be grafted on the substrate surface. The use of a gold surface allows the formation of self-assembled
monolayers (SAM) of thiol molecules. The arrangement of thiol derivatives with alkyl chains will allow
having ordered and functional head groups directly available from the surface.

a) Synthesis of 11-mercapto-undecyn

In order to create SAMs terminated with alkyne groups, 11-mercapto-undecyn (HCC(CH2)9SH)

was synthesized following the protocol of Colmann (figure II-14).[27] The detailed procedure can be
found in appendix C. The compound was then characterized by 1H NMR and FT-IR.

b c

Figure II-14. Scheme of the 11-mercapto-undecyn synthesis in 3 steps a) mesylation of the alcohol group, b)
nucleophile substitution of mesyle groups by thioacetate group and c) acid hydrolysis of the thioacetate group in thiol

Starting from the 11-hydroxy-undecyn, the first step is a mesylation of the alcohol function of
the molecule to obtain the 11-methylsulfonate-undecyn. This reaction takes place in THF under reflux
with mesyl chloride (CH3SO2Cl) with triéthylamine to activate the reaction (Figure II-15, step a).

The methylsulfonate group was substituted by a thioacetate group to form the 11-thioacetate-
undecyn. This reaction takes place with potassium thioacetate in methanol under argon for 3 hours
(Figure II-15, step b).

The thiol is synthesized by acid hydrolysis of the acetate group by reacting the 11-thioacetate-
undecyn with hydrochloric acid in methanol refluxed under argon for 5 hours (Figure II-15, step c).

73
 Figure II-15. Infrared spectra of the 11-mercapto-undecyn

Figure II-15 presents the infrared spectra of the synthesized compounds. The characteristic
vibrational bands at 2850 cm-1, 2920 cm-1 and 1460 cm-1 indicate the presence of the long alkane chains
of the molecules. The terminal alkyne group is characterized by the specific vibrational band of the
stretching mode within the C-H binding at 3300 cm-1, the rocking mode in the C-H binding at 635 cm-1
and the stretching mode of the C≡C binding at 2115 cm-1.[5]

b) Self-assembled monolayer formation

The formation of self-assembled monolayer is then realized by the following procedure. The
gold substrates were first cleaned by hydrogen and oxygen plasma for 2 minutes in order to remove
organic elements from the surface to activate the surface. The substrate was then immersed in an
ethanolic solution at 10 mM of 11-mercapto-undecyn. The substrates remained 24 hours in the
solution and were stored in the dark at room temperature before rinsing with ethanol (figure II-16).
The formed self-assembled monolayer terminated with alkyne is noted SAM-CC.

Figure II-16. Schematic representation of the formation of self-assembled monolayer. Adapted from Toulemon.[6]

74
 c) Characterization of self-assembled monolayers

Different surface characterizations were used to determine the arrangement and the presence
of terminal head groups at the SAM surface.

The topography of the gold substrate was investigated with atomic force microscopy (AFM)
before and after functionalization (figure II-17). AFM gives indications on the roughness and the
thickness of the substrate.

a) b)

Figure II-17. AFM images of the substrates and cross-section profile measured on a) gold naked surface and b) SAM-CC
surface

The gold substrate consists in gold grains with different sizes with a roughness of 1.6 nm. The
image of the SAM-CC presents a similar structure. However, height profiles show a larger average
thickness which are 4.5 ± 1.5 nm and 6.9 ± 1.3 nm for gold and SAM-CC, respectively. The height
increase of about 2 nm was ascribed to the grafting of 11-mercapto-undecyn at the gold surface and
to the formation of the SAM-CC.

The water contact angle (WCA) measurement was realized with the deposition of a 5 μL water
droplet on the surface to determine its wettability (figure II-18). The picture and the measured angle
values were performed one minute after the droplet deposition in order to stabilize the system.

75
 Figure II-18. Water contact angle measurements on a a) gold substrate and b) SAM-CC

The gold surface shows a contact angle of 21.7° which agree with a highly hydrophilic surface.
After functionalization with the thiol groups, the substrate presents a contact angle of 84.1° which is
characteristic of a hydrophobic surface. Such a significant variation of the contact angle agrees with
the grafting of 11-mercapto-undecyn onto the gold surface. Indeed, the thiol molecules with their long
alkyl chains and alkyne head groups induce a hydrophobic behavior of the surface.

Polarized Modulated Infrared Reflection Absorption Spectroscopy (PM-IRRAS) measurements

were performed in order to characterize the self-assembled monolayer of thiol alkyne (figure II-19).
The antisymmetric and symmetric stretching bands at 2927 cm-1 and 2856 cm-1 respectively indicates
the presence of organic chains at the surface of the gold. These positions suggest a slight disorder
between alkyl chains due to a gauche configuration.[28]

Figure II-19. PM-IRRAS spectra of the SAM surface

A characteristic band at 3 320 cm-1 is observed and correspond to the ν(C-H) vibration in the
terminal methylene group CΞC-H.[29] Therefore, the presence of terminal alkyne groups is confirmed at
the SAM surface.

3. Conclusion

On one hand, the nanoparticles have been successfully functionalized by azide groups.
Moreover, the good stability of the suspension allows having nanoparticles available for the CuAAC

76
“click” chemistry. On the other hand, the gold substrate has been functionalized with the
complementary alkyne groups required for the reaction.

77
 C. Nanoparticle assembly prepared by “click” chemistry

The assembly method used here is based on the interaction between organic groups localized
on the surface of both nanoparticles and substrate. The CuAAC “click” reaction allows the strong
binding of the nanoparticles to the substrate and presents the advantage of being highly selective and
versatile. The CuAAC reaction usually takes place in an aqueous solution between an alkyne and an
azide groups and is catalyzed by copper (I).[30] However, copper (I) has poor stability in aqueous
medium and the nanoparticles are stable in organic solvent. Therefore, we performed the reaction in
an organic solvent by using a copper catalyst stabilized by ligands: [CuBr(PPh3)3]. The “click” reaction
was performed by using the protocol described by Xavier Cattoën[31] in THF in presence of
triethylamine to activate the reaction (figure II-20).

Figure II-20. Schematic representation of the "click" chemistry assembly of nanoparticles onto gold surface

The functionalization of both nanoparticles and substrate with complementary groups will
allow the preparation of assemblies. The process of this reaction for different size of nanoparticles, the
influence of reaction parameters and the characterizations of these assemblies will be studied in this
part.

1. Assembly of 10 nm nanoparticles

The protocol and control experiment will be detailed in this part as reference assembly. The
concept will be presented here and the study of different sizes and parameters will be discussed
afterwards.

a) Assembly protocol

The click chemistry reaction was performed in a reactor of 10 mL (figure II-21). A suspension
of NP@N3 in THF was prepared with a concentration of 1 mg/mL. A CuBr(PPh3)3 catalyst (3% of
nanoparticle weight) and triethylamine (0.5 mL) were added to foster the reaction by increasing the
lability of the C-H binding in the alkyne group.[6, 31]

78
Figure II-21. Schematic representation of the apparatus required to perform nanoparticle assembling by "click" chemistry
reaction

The SAM-CC was then immersed in this solution and heated at 70°C for 48 hours with a
condenser under argon. The reactor was put in a silicon bath and the temperature was controlled by
a thermocouple. After the reaction was completed, the substrate was removed, rinsed with THF and
put in ultrasonic bath for 1 minute in order to remove the nanoparticles which are not covalently
bound to the surface. The substrate was then dried under an air stream.

b) Scanning electron microscopy characterization

Scanning electron microscopy (SEM) is a powerful technique which allows determining the
structure of the nanoparticle assemblies.

a) b)

Figure II-22. SEM images of 10 nm-sized nanoparticle assemblies at magnifications a) X100000 and b) X50000

79
 On the SEM picture showed in figure II-22, a fully covered surface of nanoparticles can be
observed. The image with low magnification (X50000) shows a homogenous surface without
aggregates neither holes in the assembly on a large area. The image with a higher magnification
(X100000) allows determining the density of nanoparticles on the surface. The number of
nanoparticles per surface unit area was measured with ImageJ software by counting nanoparticles in
five different areas which allow determining an average with a standard deviation. For this sample, the
density is 4 950 ± 95 NPs/μm². In order to compare the density for different sizes of nanoparticles, it
was normalized to a maximal theoretical value which corresponds to the hexagonal close packing of
nanoparticles onto a surface (figure II-23).

Rhombus area: 2√3¿²

Spheres occupation (colored area): À. ¿²

= = m, ÅmÆÅ
‡√Á² ‡√Á
ò Ã
Filing factor:

Figure II-23. Schematic representation of the nanoparticle arrangement on a hexagonal close packing

This maximal theoretical value was calculated for a hexagonal close packing of 10.1 nm spheres
surrounded with an organic layer about 2 nm which is 5 810 NPs/μm². Considering this, the sample
showed on the figure II-22 presents 85% of the maximum theoretical density. This value is in
accordance with previous studies of nanoparticle films.[4, 6] These values for 5.1 nm and 21.2 nm
nanoparticles are respectively 13 940 NPs/μm² and 1 820 NPs/μm².

2. Study of the kinetics of the assembly

The considerable advantage of the “click” chemistry method is to control the spatial
arrangement of the assembly as function of the reaction time.[20] It is a real advantage to control the
magnetic collective properties and specifically the plasmonic properties carried by the substrate.

The kinetic of the reaction allows controlling the density of the nanoparticles at the surface. A
typical assembly of 10 nm of nanoparticles in a concentration of 0.8 mg/mL in Fe3O4 has been realized
by varying the reaction time between 1 hour and 48 hours. SEM analysis showed the evolution of the
surface coverage and allowed measuring the nanoparticle density as function of the reaction time
(figure II-24).

80
 1h 2h 4h

8h 24 h 48 h

100 nm

Figure II-24. SEM images of the NP10 assembly after different reaction times. Magnification X100000

After 1 hour of reaction, some azido-terminated nanoparticles were already grafted onto the
alkyne-terminated SAM. Isolated nanoparticles can be observed at the gold grain boundaries where
the energy necessary for the assembly should be the lowest. Alignments or packing from 3 to 6
nanoparticles promoted by magnetic dipolar interactions were also observed.[6, 32] The density
increased with the reaction time until reaching a full monolayer of nanoparticles after 48 hours. After
24 hours, some aggregates were observed. This aggregation phenomenon may be caused by the
evaporation of the solvent during the reaction.

The density was measured and reported to the maximal theoretical value for NP10 (5 810
NPs/μm²) as function of the reaction time (figure II-25).

81
 Figure II-25. Density of nanoparticles onto the gold surface as a function of the reaction time (calculated from SEM
images)

The data were fitted with a logarithm function and the scale bars correspond to measurements
on different areas. The evolution of the density is nonlinear and a rapid evolution is monitored until
reaching 80% of the maximal coverage after 24 hours. The film of nanoparticles does not reach 100 %
of the maximal theoretical density value even after 48 hours which can be explained by the random
grafting of the nanoparticles at the surface which can let holes on the surface smaller than the
nanoparticles size. Indeed, the maximum value was calculated for a hexagonal close packing, which is
not the case for the random arrangement of the nanoparticles in these samples.

3. Variation of assembly parameters

The structuration of the nanoparticle assembly is crucial to control the collective and individual
properties of these films. The size of nanoparticles directly influence the properties of the assembly
and therefore has to be studied. Moreover, the presence of aggregates on long reaction times may
question us on the influence of some parameters of the assembly. Different parameters have been
studied previously in our team which showed the influence on the nanoparticle density.[6] However,
the suspension concentration has not been studied yet and can influence the presence of aggregates.

a) Nanoparticle sizes

The kinetics of assembly reaction was studied for each nanoparticle size. Figure II-26 shows
the SEM analysis for the assembly of 5 nm nanoparticles, with a concentration of 0.45 mg/mL, by
varying the reaction time between 1 hour and 48 hours.

82
 1h 2h 4h

8h 24 h 48 h

Figure II-26. SEM images of the assembly of NP5 after different reaction times. Magnification X100000

The presence of nanoparticles is observed after one hour of reaction. The density is very low
until 8 hours of reaction. The nanoparticles are isolated on the surface and no alignments or clusters
are observed. Contrary to the 10 nm nanoparticles, after 48 hours of reaction, the surface is not fully
covered. The density as function of the reaction time is reported on figure II-27.

48 h

Figure II-27. Density of the NP5 at the gold surface as function of reaction time

The same behavior than for the NP10 is observed with a fast increase of the density until 24
hours of reaction. The kinetics of assembly for the NP5 is slower than for the NP10. The density value,
even after 48 hours, remains low below 40 % of maximal theoretical value.

83
 In a colloidal suspension, smaller nanoparticles have a higher mobility and therefore statically
generate more encounters with the gold surface. However, the kinetic of the assembly reaction is
slower in this case which disagrees with this theory. In the case of assembly of nanoparticles with
“click” chemistry, the kinetic of the reaction is slow since the azide and alkyne groups have to be in
presence of the copper catalyst to form the reactional intermediate. Therefore the reaction requires
more time and a fast mobility of nanoparticle can reduce the kinetic of reaction. Moreover, the NP5
due to their lower surface, present less functional groups which can interact at the SAM surface. The
reduction of active groups on the nanoparticle surface can also decrease the recognition process.
Finally, the dipolar interactions can drive the assembly by inducing nanoparticles to get closer (see part
C.3.b).[6] The smaller nanoparticles present less dipolar interactions due to their size[13] and, therefore,
may contribute to such a slow kinetic.

The same study was performed with NP20 with a concentration of 0.49 mg/mL. The SEM
analysis is shown on figure II-28.

1h 2h 4h

8h 24 h 48 h

Figure II-28. SEM images of the assembly of NP20 after different reaction times. Magnification X100000

The nanoparticles can be observed onto the gold surface after one hour of reaction. After 4
hours of reaction, the density increases faster and at 8 hours the surface is almost fully covered.
Moreover, the presence of clusters after 4 hours seems to indicate that the nanoparticles which were
already immobilized onto the substrate favored the assembly of others next to them. The assembly
after 24 hours and 48 hours of reaction present a great order of the nanoparticles which confirm the
directed assembly. This is caused by the stronger dipolar interactions between 20 nm nanoparticles
which induces attraction of the nanoparticles. The density of nanoparticles as function of the time is
reported on figure II-29.

84
 Figure II-29. Density of the NP20 onto the gold surface as function of reaction time

With the NP20, the density onto the gold surface reaches almost 100% of the maximum
theoretical coverage. Such a high coverage may be explained by dipolar interactions which promote
the tight packing of nanoparticles.

The assembly of nanoparticles with three different sizes showed different behaviors. The
smallest nanoparticles seem to assemble the slowest and are isolated on the surface, neither clusters
nor alinements were observed. In contrast, 10 nm and 20 nm sized nanoparticles assemble much faster
into clusters of chains of nanoparticles. The strongest dipolar interactions between these nanoparticles
favor their assembly and, in the case of NP20, create an arrangement with higher density. The
magnetic properties seem to influence directly the spatial arrangement of the nanoparticles onto
substrates.

b) Concentration of the nanoparticle suspensions

The concentrations of the nanoparticle suspensions were varied in order to study the influence
on the kinetics of the reaction (figure II-30). For each size of nanoparticles, two different
concentrations were studied: one low concentration with few nanoparticles in suspension which
induce fewer encounters of the nanoparticles at the surface; and one with high concentration where
the dynamic promotes the encounters between nanoparticles and functional groups at the surface.
The concentrations are given in mg of Fe3O4 per mL.

85
a) b)

c)

Figure II-30. Density of the a) NP5, b) NP10 and c) NP20 assembly as function of reaction time for two different
concentrations

The increase of nanoparticle concentrations resulted in higher densities for the same reaction
time, whatever their size. These results are ascribed to the increase of probability of nanoparticles to
hit the substrate. Moreover, the highest concentrations lead to stronger dipolar interactions between
nanoparticles which foster the assembly kinetic. Longer reaction times may improve the density of
nanoparticles onto the substrate, but the stability of the suspensions decreased with the time.
Therefore, high concentration of nanoparticles for short times of reaction is expected to increase the
density of nanoparticles without aggregation at the surface.

The film thickness was also measured by performing ellipsometry. The measurement was
performed on the SAM-CC to know the thickness of the organic layer. The thickness measured was 1.4
± 0.2 nm and this value was used thereafter to measure the thickness of the iron oxide layer. The
refractive index used for the determination of the thickness is 2.42. This refractive index corresponds
to a thin film of iron oxide. Here, the nanoparticle assembly does not form a complete film and the
refractive index is overestimated. The measurements were performed on different spots of the surface
of the sample to give average values with standard deviation. The thickness was measured for the two
different concentrations studied and the results are show on figure II-31.

86
a) b)

c)

Figure II-31. Ellipsometry measurements of the a) NP5, b) NP10 and c) NP20 assembly as function of reaction time for
two different concentrations

The thickness of the nanoparticle assembly increases with the reaction time for each size of
nanoparticles. The trends are following the densities measured by SEM with apparition of a plateau
for the NP20 and slight increase of the thickness for NP5 and NP10 after 24 hours of reaction. The
values after 48 hours of reaction are close of the theoretical maximum values (which correspond to
91% of the size of the nanoparticles in the case of a close packed hexagonal arrangement) for NP10
(8.9 ± 0.2 nm) and NP20 (18.1 ± 2.5 nm). However for NP5, the value is unexpected with a thickness
reaching 6.4 ± 0.5 nm after 48 hours of reaction for the highest concentration. This overestimation can
be due to the refractive index used (2.42) which is too high with regard to the low density of
nanoparticles.

c) Stability of the assembling solution in time

The assembly is promoted by the dipolar interactions between the nanoparticles as shown
with the assembly of NP20 which form an array on the surface. However, the stability of the suspension
evolves with the reaction time and the nanoparticles are aggregated at the end of the reaction
(especially in the case of NP20). This aggregation is visible, but it can happen earlier during the reaction.

87
Therefore, to get a better understanding, the stability during the reaction was studied. For each size
of nanoparticles, 1 mL of the solution was taken off for different reaction times and was characterized
by granulometry (figure II-32).

Figure II-32. Hydrodynamic diameter in volume as function of the reaction time of a) NP5, b) NP10 and c) NP20

The hydrodynamic diameter shows two phases for each size of nanoparticles. For short
reaction times, the mean hydrodynamic diameter corresponds to the size of the nanoparticles,
meaning a good stability of the suspension. However, for longer times, the suspensions present a huge
value of hydrodynamic diameters which corresponds to aggregation phenomenon. This aggregation
appears at different times as function of the size of the nanoparticles. For NP20, the aggregation is
observed after 4 hours of reaction only. This aggregation appears later for smaller nanoparticles: after
8 hours for NP10 and 24h for NP5 (we can observe a slight aggregation for NP5 after the first hours of
reaction). The largest nanoparticles present an aggregation earlier than the smallest nanoparticles.

This aggregation can be caused by the desorption or degradation of the ligand at the
nanoparticle surface.[33] Infrared spectra were performed on the assembly solution after the “click”
reaction of NP10 in order to control the stability of the phosphonic acid molecules (figure II-33).

88
 Figure II-33. Infrared spectra of the assembly solution after 48 h of reaction of NP10

The assembly solution with the presence of triethylamine, nanoparticles and the copper
catalyst presents many bands which are difficult to interpret. However, the presence of azide
vibrational band at 2100 cm-1 allows to conclude that there is no degradation of the azide groups.
However, the desorption of the phosphonic acid is still possible and may induce aggregation.
Moreover, dipolar interactions may favor the formation of clusters in suspension which may explain
the easiest aggregation in the case of the largest nanoparticles.

4. Conclusions

Iron oxide nanoparticles have been assembled onto gold substrates through “click” chemistry
reaction which provides a stable assembly thanks to covalent bonds. The control of the reaction time
showed that the spatial arrangement can be tuned. Moreover, the density at the surface is dependent
of the concentration of the nanoparticle suspension. By following the rate of coverage, dipolar
interactions have been highlighted to influence the assembly. For small nanoparticles, the assembly
structure consists in isolated nanoparticles. In contrast, for larger nanoparticles such as NP20, strong
dipolar interactions accelerate the assembly kinetics and lead to clusters and alinements. The study of
the stability of the suspensions as function of the reaction time has shown that the assembly may also
be driven by the aggregation of nanoparticles in solution.

89
 D. Characterization of the nanoparticle assemblies

1. Composition and structural characterizations

The structural and composition characterizations were performed on a reference sample of a

fully covered assembly of 10 nm-sized nanoparticles (NP10) obtained after 48 hours of reaction.

a) Atomic force microscopy

Atomic force microscopy (AFM) was performed on the NP10 assembly to have complementary
information to the SEM images on the structure of the assembly. Figure II-34 shows the AFM images
2x2 μm of the surface of an assembly after 48 hours. Dense packed nanoparticles were observed on
the surface. The sample is homogenous and the density of nanoparticles is constant on the image. A
topographic profile has been done using WSXM software.[34] The measured height on the assembly
reaches an average of 9.1 nm which agree with the nanoparticle size.
22.92 nm

a) b)

400nm

1.95 nm

Figure II-34. a) AFM image of the 10 nanometer-sized nanoparticle assembly and b) topographic profile following the
blue line

The roughness of the sample can be also extracted from these images and is evaluated to 2.6
nm. This larger value than SAM-CC can be explained by the presence of nanoparticles at the surface
which increase the roughness of the assembly.

b) Water contact angle

Water contact angle (WCA) measurements have been performed to evaluate the
hydrophobicity of the surface. The measurements have been done by depositing a 5 μL droplet on the
surface and take a picture of the sample after 1 minute in order to stabilize the system. The nature of
this hydrophobicity is driven by the functional groups at the surface. Here, long alkane chains do not

90
present polar groups which can induce a hydrophilic surface. Indeed, figure II-35 shows a contact angle
value of 99.8° relevant of a hydrophobic surface.

99,8 ± 0.4°

Figure II-35. Water contact angle characterization of the 10 nm-sized nanoparticle assembly

c) Polarized Modulated Infrared Reflection Absorption Spectroscopy

Polarized Modulated Infrared Reflection Absorption Spectroscopy (PM-IRRAS) allows

determining the vibrational bands of compounds grafted onto a substrate by using infrared with a
grazing (< 5°) incidence (figure II-36). The presence of interferences reduces the spectral range
between 3500 cm-1 and 900 cm-1 and avoids the observation of the iron oxide band. Nevertheless, it
can be possible to observe functional groups at the surface of the nanoparticle assembly.

Figure II-36. PM-IRRAS spectrum of the assembly of nanoparticles of 10 nm with full density

The vibrational bands characteristics of the alkane chains are visible at 2850 cm-1 and 2920 cm-
1
, meaning the presence of the ligands at the surface of the nanoparticles. The peak of the stretching
mode (ν(Fe-O-P)) at 1040 cm-1 and the azide vibrational band (ν(N≡N)) at 2100 cm-1 show that AP12N3
was still grafted onto the nanoparticle surface. Moreover, it confirms the stability of AP12N3 at the
nanoparticle surface after the CuAAC reaction was completed.

91
 d) X-Ray Photoelectron Spectroscopy

In order to confirm the presence of azide groups at the surface and to validate PM-IRRAS
results, XPS measurements have been performed (figure II-37). The Fe2P region shows multiple peaks
which can be attributed to 2P1/2 and 2P3/2 core level. Each peak can be deconvoluted in two
components corresponding to Fe2+ and Fe3+ from the magnetite.[35] The satellite peaks, especially at
720 eV can be attributed at the Fe3+ present in the maghemite phase which indicates the oxidation of
the nanoparticles.[36] However, the XPS is a surface characterization and cannot show a rigorous
stoichiometry in the film which allow to estimate the quantity of maghemite. The P2P core level region
shows a low signal at 133.5 eV. The phosphonic acids are localized at the nanoparticle surface while
they are between nanoparticle and the substrate and cannot be observed since the XPS signal does
not probe beyond 9-10 nm.[37]

a) b)

c) d))

Figure II-37. XPS spectra of the NPs assembly on the region a) Fe2P b) P2P, c) C1S and d) N1S

N1S core level region shows both characteristic peaks at 401,3 eV and 405,4 eV of the azide
groups. The intense peak at 401,3 eV corresponds to the two lateral nitrogen atoms (-N=N=N) whereas
the peak at 405,4 eV correspond to the central nitrogen atom (-N=N=N). It is confirmed by the ratio of
the peak area close to 2:1.[38] These measurements confirm the presence of the azide groups at the
nanoparticle assembly surface after the click reaction. The C1S core level presents a large component
at 284.8 eV which is attributed to the C-C and C-H bonds and a smaller one at 286.6 eV representing
the C-N bonds in the triazole bridge.[39, 40]

92
 The surface characterizations showed that the azide groups are still present at the surface of
nanoparticle assembled onto substrate and are therefore available for further grafting.

2. Magnetics characterization

The magnetic properties of nanoparticle assemblies are strongly dependent of their spatial
arrangement. Indeed, magnetic dipolar interactions are dependent on the intrinsic properties of
nanoparticles (magnetization saturation) and also on the distance between nanoparticles. In the first
part of this chapter, the magnetization saturation of the nanoparticles was measured for the NP20
which presents the highest MS (75 emu/g) and that of NP5 and NP10 was lower (55 emu/g). The dipolar
energy can be evaluated by knowing the inter-particle distance which is dependent of the density.
Therefore, the collective and individual properties can be studied as function of the spatial distribution
of assemblies. Samples with different sizes and densities were measured to evaluate the influence of
magnetic dipolar interactions.

a) Cycles of magnetization as function of external field

Magnetization was measured as function of the external applied magnetic field for assemblies
of NP5, NP10 and NP20 with two different densities. Assemblies with low and high density of
nanoparticles were prepared which represent two systems with weak and strong dipolar interactions
respectively. The inter-particle distance is lower than 7 nm. The inter-particle distances were also
calculated (see Appendix A for detailed calculus) in order to evaluate the strength of dipolar
interactions. The inter-particle distance is an average value considering nanoparticles randomly
dispersed onto a surface. The samples with low densities of nanoparticles present a low coverage
(below 25% of the maximal theoretical value) which corresponds to inter-particle average distances
larger than 10 nm. In the case of high density, the NP10 and NP20 present coverage close to 80%
whereas the NP5 presents coverage of 41% of the maximal theoretical value. However, we saw
(especially in the case of NP20), that the nanoparticles can form clusters and alignments onto the
substrate. Therefore, this value only represents the idea of how far a first neighbor can be found. The
density values are summarized in table II-7 and the SEM images are shown in figure II-38.

Table II-7. Structural characteristics of nanoparticle assemblies. density and the corresponding average inter-particle
distance

Inter-particle distance
Density (NPs/μm²)
(nm)
2000 ± 95
Low density 11 ± 2
(20 ± 1 %)
NP5
4019 ± 430
High density 7±2
(41 ± 4%)
1240 ± 255
NP10 Low density 14 ± 4
(24 ± 5%)

93
 3890 ± 95
High density 5±1
(76 ± 2%)
320 ± 45
Low density 27 ± 5
(17 ± 3 %)
NP20
1505 ± 60
High density 6±1
(81 ± 3%)

a) b) c)

d e) f)
)

Figure II-38. SEM images of NPs assembly of a) and b) 5 nm, c) and d) 10 nm, e) and f) 20 nm at low density a),c) and e)
and high density b), d) and f). Scale bar: 100 nm

The inter-particle distances here play an important role since the dipolar interactions between
nanoparticles influence the collective magnetic properties of these assemblies. If we consider the
macro-spin approximation and a triangular array of nanoparticles, the dipolar energy Ed can be defined
as (see chapter I):

6: = 2.8 9f
.²
Equation II.2

with μ the nanoparticle magnetization and a the distance between center of the nanoparticles.
Considering the magnetization of a nanoparticle as the saturation magnetization MS multiplied by its
volume (V = πD3/6). The distance between centers of nanoparticles corresponds to the sum of the
diameter of the nanoparticle and the inter-particle distance (a = D + s), which give:
c
ÉÊf
Ç!È Ì
6: = 2.8
Ë
(ar¨)f
Equation II.3

with MS the nanoparticle magnetization, D the nanoparticle diameter and s the distance
between nanoparticles. Here, we can easily see the strong influence of the size of the nanoparticles on
this interaction. The dipolar energy of bulk magnetite normalized with the Boltzmann constant is
determined with an inter-particle distance of 4 nm which corresponds to twice the organic layer (figure

94
 II-39.a). A strong increase of the dipolar energy can be observed for larger nanoparticles, whereas
nanoparticles smaller than 10 nm exhibit a lower dipolar energy. The distance between nanoparticles
influences also the dipolar energy. This effect of the distance was calculated for the three sizes of
nanoparticles considered and normalized with their anisotropy energy defined by:

69 = <> = <
ea f
Í
Equation II.4

with K anisotropy constant of the nanoparticle. The ratio between dipolar and anisotropy
energies can be expressed as:
b
= Ñar¨Ò
Î’ _.Ð!È a
ÎÏ ÍO
Equation II.5

This ratio is determined with the saturation magnetization calculated for the size of each
nanoparticle and with the magnetocrystalline constants of magnetite (K = 1.1x105 erg/cm3) and
presented in figure II-39.b.

a) b)

Figure II-39. a) Dipolar energy as function of the nanoparticle diameter and b) ratio between dipolar energy and
anisotropy energy as function of the inter-particle distance for different sizes of nanoparticle

The dipolar energy decreases fast with the inter-particle distance especially for NP5 and NP10.
In the case of NP20, the dipolar interactions stay strong even for the sample with low density and large
inter-particle distance (27nm) where the ratio Ed/Ea is larger than sample NP10 with high density. These
curves show the strong influence of the size of the nanoparticles on the dipolar interactions and explain
the structuration of the assemblies where larger nanoparticles form alignments. These informations
will help to understand the magnetic measurements performed on these assemblies.

Figure II-40 presents the hysteresis loop of the assemblies of nanoparticles from different sizes
for the low (red curve) and the high density coverage (blue curve). The coercive field HC and ratio of
remanent magnetization on saturation magnetization at 5 K are summarized in table II-8.

95
Table II-8. Coercive field and remanent magnetization for nanoparticle assemblies of different sizes and densities
compared with powder at 5 K

HC (Oe) MR/MS
HC (Oe) MR/MS Ed/kB (K)
powder powder
NP5 Low density 1100 0.38 30
240 0.36
NP5 High density 210 0.42 70
NP10 Low
1660 0.40 330
density
500 0.36
NP10 High
370 0.42 1250
density
NP20 Low
480 0.28 3300
density
330 0.16
NP20 High
550 0.31 18000
density

a) b)

c)

Figure II-40. Magnetization as a function of an applied magnetic field recorded at 5 K for nanoparticle monolayers with
high (blue curve) and low (red curve) densities a) NP5, b) NP10 and c) NP20. Insets represent cycles at 300

The interpretation of the evolution of the coercive field and the remanent magnetization as
function of the dipolar interactions and the anisotropy of assemblies is not trivial. The literature is
conflicting in the evolution of HC and MR/MS as function of the dipolar interactions.[13, 41-44] The coercive
field and remanent magnetization should decrease when the dipolar interactions are stronger which
are mostly explained by the collective properties of the nanoparticles. The force needed to reverse the
magnetization of an assembly of nanoparticles will be weaker than the force to reverse an isolated
nanoparticle because of the dipolar interactions will participate to reverse the magnetic moments. If

96
we consider the case of NP5 and NP10 which present both weak dipolar interactions, a similar behavior
is observed. Indeed, a larger coercive field is observed for samples with low densities which is coherent
with a high HC for isolated nanoparticles. The dipolar energy for the NP10 sample with high density
(1250 K) is almost four times higher than for the low density sample (330 K). It is correlated to the
coercive field which is four times weaker (370 Oe versus 1660 Oe). In the case of the NP5 assembly
with low density, the coercive field is five times larger than for high density whereas the dipolar energy
is only twice weaker. This behavior can be explained by the calculated average distance which does
not correspond exactly to the spatial arrangement of nanoparticles. In the case of NP10, a significant
amount of clusters of nanoparticles can be observed which strengthen the dipolar interactions.
Moreover, the hysteresis loop of NP5 with low coverage is not symmetric, meaning all the magnetic
moments are not reversed at the same times. Different populations of nanoparticles can be
responsible of this behavior such as isolated nanoparticles and clusters of nanoparticles.

In the case of NP20, this enlargement of the hysteresis loop is not observed for assembly with
low density of nanoparticles. The explanation may come from the strong dipolar interactions even with
large inter-particles distances. Indeed, the Ed/kb ratio for NP20 is much larger than for NP5 and NP10
assemblies with high densities. Moreover, as shown previously, the 20 nm nanoparticles tend to easily
form clusters or alinements.

The slight decrease of HC in the dense assemblies compared to the powder state can be
attributed of the increase of the anisotropy in 2D which favors in-plane magnetization reversal.
However, for NP20, HC is the largest in the nanoparticle assembly. The strong dipolar interactions can
be responsible of this behavior which has been already observed.[13]

The increase of the MR/MS ratio in assemblies compared to powder may be due to the loss of
order in the powder state. Indeed, a 2D array shows a higher order and, therefore, presents more
anisotropy than powder. A random orientation is expected in the powder state which conducts to
decrease the MR/MS value.[13]

b) Magnetic measurements as function of the temperature

The measure of the magnetization as function of the temperature was performed on the
nanoparticle assemblies of different sizes for low and high coverage (figure II-41).

97
 a) b)

c) d)

Figure II-41. a,b) ZFC/FC temperature dependent magnetization curves and c,d) temperature derivative of the difference
between ZFC and FC for assemblies of nanoparticles of different sizes at a,c) low density and b,d) high density

TMax were extracted from the maximum of the ZFC curves. Values for the assemblies with
different densities and the nanoparticles in powder state are summarized in table II-9.

Table II-9. TMax values for nanoparticle assemblies of different sizes and densities and for the powder state

NP5 NP10 NP20

TMax TB TMax TB TMax TB
Low density 30 K 10 K 90 K 260 K
High density 35 K 15 K 100 K 60 K 270 K 170 K
Powder
50 K 20 K 110 K 75 K > 350 K 185 K
state

TMax measured from the ZFC curves and TB from the temperature derivatives (figure II-41).
However, for low density film, the measure of TB is complicated due to the low signal and to high noise
on the measurement. Nevertheless, the values of TB are following the same trend than TMax. For all
nanoparticle assemblies, TMax increases with nanoparticle density which is correlated to higher
magnetic anisotropy of nanoparticles.

98
 These results have been already shown in the case of nanoparticle assemblies prepared by the
Langmuir-Blodgett[13] and click chemistry[6] techniques which showed the 2D enhances collective
properties by favoring anisotropy of dipolar interactions in plane.

The lower dipolar interactions for non-interacting nanoparticles explain the slightly decrease
of the blocking temperature in the case of low density assemblies.[45] Nevertheless, the assemblies of
nanoparticles in a 2D layer increase the anisotropy of the film and therefore require a higher energy
to change the behavior of nanoparticles and to make them becoming superparamagnetic.[6]

A higher blocking temperature could be expected for high density films in comparison with
powder state. Here, the TMax is lower for the film than for the powder which goes against what have
already been observed.[6, 13] This behavior could have been explained by the presence of aggregates or
inhomogeneous density of nanoparticles assemblies, but the SEM images do not show such
inhomogeneities. Nevertheless, NP5 and NP20 high density assemblies present some “holes” at the
surface where no nanoparticles are grafted and so can be responsible of the smaller value of TMax.

Another observation is the non-saturation of the FC curve at low temperatures for the NP5
assemblies. The dipolar interactions are weaker with the decrease of the size of nanoparticles[13].
Therefore, the FC curve will not present a plateau for the low temperature. Isolated nanoparticles or
with low dipolar interactions are typical representative for this non-saturation. The combination of the
small size and the low coverage on this sample makes very low dipolar interactions and the non-
saturation even for low temperatures.

3. Conclusion

The magnetic properties of assemblies of nanoparticles with different sizes and different
densities have been studied. The investigations of the dipolar interactions have highlighted two distinct
behaviors between the NP5 and NP10 which present weak dipolar interactions and NP20 which
presents strong dipolar interactions even in the case of the assembly with large inter-particle distance.
The non-interacting nanoparticles showed a higher coercive field which is explained by the loss of
collective properties. In contrast, larger nanoparticles present collective behaviors even for low
coverages confirmed by similar coercive fields. The temperature of transition between ferrimagnetism
and superparamagnetism for dense assemblies is surprisingly close to the powder state. The presence
of defects in the films could explain this behavior.[32, 45] However, the decrease of TMax for the lowest
densities is coherent with non-interacting nanoparticles.[45, 46]

99
 E. General conclusion

The thermal decomposition method has allowed synthesizing iron oxide highly stable
nanoparticle suspension in organic solvent. Three different sizes have been successfully synthesized
and characterized: 5.1 ± 0.8 nm, 10.1 ± 1.1 nm and 21.2 ± 1.2 nm.

Nanoparticles were post-functionalized by direct ligand exchange with a phosphonic acid

derivative carrying azide group in order to complete the assembly through CuAAC “click” chemistry.
Moreover, gold substrates were functionalized with thiol carrying alkyne complementary groups
required for the “click” reaction.

The assemblies of the azido-terminated nanoparticles onto the alkyne-terminated SAM were
performed by CuAAC “click” reaction and the conditions of reaction were studied. The variation of
reaction time allowed controlling the spatial distribution of the nanoparticles on the gold surface from
few isolated nanoparticles to a dense monolayer. The kinetic of the assembly was studied for two
different concentrations and showed a faster kinetic for larger nanoparticles in contrast with small
nanoparticles which do not reach the full monolayer. The investigation on the colloidal stability of the
nanoparticle suspension after different reaction time showed that the aggregation of nanoparticles is
favored for larger nanoparticles. The assembly kinetics seems driven by the stability of the suspension
and the dipolar interactions stronger in the case of 20 nm-sized nanoparticles.

Finally, the magnetic properties of assemblies prepared with different sizes and different
densities were performed. The measure of the magnetization as function of an applied field and the
temperature highlighted that the NP20 present stronger dipolar interactions than the NP5 and NP10
whatever the density of nanoparticles. This behavior explains the assembly is driven by these dipolar
interactions and explains the faster kinetic and stronger tight-packing of the NP20.

100
 F. Bibliography

[1] Baaziz, W.; Pichon, B. P.; Fleutot, S.; Liu, Y.; Lefevre, C.; Greneche, J.-M.; Toumi, M.; Mhiri, T.;
Begin-Colin, S., The Journal of Physical Chemistry C, (2014) 118, 3795.
[2] Demortiere, A.; Panissod, P.; Pichon, B. P.; Pourroy, G.; Guillon, D.; Donnio, B.; Begin-Colin, S.,
Nanoscale, (2011) 3, 225.
[3] Cotin, G., Université de Strasbourg (2017).
[4] Benoit P. Pichon, A. D., Matthias Pauly, Karin Mougin, Alain Derory, and Sylvie Bégin-Colin, J
Phys Chem, (2010) 114, 9041.
[5] K, N., Wiley: New york, (1986),
[6] Toulemon, D., Université de Strasbourg, (2013),
[7] Baaziz, W., Université de Strasbourg, (2011),
[8] Walter, A.; Billotey, C.; Garofalo, A.; Ulhaq-Bouillet, C.; Lefèvre, C.; Taleb, J.; Laurent, S.; Vander
Elst, L.; Muller, R. N.; Lartigue, L.; Gazeau, F.; Felder-Flesch, D.; Begin-Colin, S., Chemistry of Materials,
(2014) 26, 5252.
[9] Erik Wetterskog, C.-W. T., Jekabs Grins, Lennart Bergström, and German Salazar-Alvarez, ACS
Nano, (2013) 7, 7132.
[10] Park, J.; An, K.; Hwang, Y.; Park, J. G.; Noh, H. J.; Kim, J. Y.; Park, J. H.; Hwang, N. M.; Hyeon, T.,
Nature materials, (2004) 3, 891.
[11] Park, J.; Lee, E.; Hwang, N. M.; Kang, M.; Kim, S. C.; Hwang, Y.; Park, J. G.; Noh, H. J.; Kim, J. Y.;
Park, J. H.; Hyeon, T., Angew Chem Int Ed Engl, (2005) 44, 2873.
[12] Zhang, Q. S. a. Z. J., J. Phys. Chem. B, (2006) 110, 11205.
[13] Pauly, M., Université de Strasbourg, (2010),
[14] Pichon, B. P.; Gerber, O.; Lefevre, C.; Florea, I.; Fleutot, S.; Baaziz, W.; Pauly, M.; Ohlmann, M.;
Ulhaq, C.; Ersen, O.; Pierron-Bohnes, V. r.; Panissod, P.; Drillon, M.; Begin-Colin, S., Chemistry of
Materials, (2011) 23, 2886.
[15] Nogués, J.; Sort, J.; Langlais, V.; Skumryev, V.; Suriñach, S.; Muñoz, J. S.; Baró, M. D., Physics
Reports, (2005) 422, 65.
[16] E. J. W. Verwey, E. L. H., The Journal of Chemical Physics, (1947) 15, 174.
[17] Upadhyay, A. P.; Behara, D. K.; Sharma, G. P.; Bajpai, A.; Sharac, N.; Ragan, R.; Pala, R. G.;
Sivakumar, S., ACS applied materials & interfaces, (2013) 5, 9554.
[18] Rianasari, I.; de Jong, M. P.; Huskens, J.; van der Wiel, W. G., International journal of molecular
sciences, (2013) 14, 3705.
[19] Kinge, S.; Gang, T.; Naber, W. J. M.; van der Wiel, W. G.; Reinhoudt, D. N., Langmuir, (2011) 27,
570.
[20] Toulemon, D.; Pichon, B. P.; Cattoen, X.; Man, M. W.; Begin-Colin, S., Chem Commun (Camb),
(2011) 47, 11954.
[21] Toulemon, D.; Pichon, B. P.; Leuvrey, C.; Zafeiratos, S.; Papaefthimiou, V.; Cattoën, X.; Bégin-
Colin, S., Chemistry of Materials, (2013) 25, 2849.
[22] T. J. Daou, J. M. G., G. Pourroy, S. Buathong, A. Derory, C. Ulhaq-Bouillet, B. Donnio, D. Guillon,
and S. Begin-Colin, Chem Mater, (2008) 20, 5869.
[23] Basly, B.; Popa, G.; Fleutot, S.; Pichon, B. P.; Garofalo, A.; Ghobril, C.; Billotey, C.; Berniard, A.;
Bonazza, P.; Martinez, H.; Felder-Flesch, D.; Begin-Colin, S., Dalton Trans, (2013) 42, 2146.
[24] T. J. Daou, S. B.-C., J. M. Grenèche, F. Thomas, A. Derory, P. Bernhardt, P. Legare, and G.
Pourroy, Chem Mater, (2007) 19, 4494.
[25] Daou, T.; Buathong, S.; Ung, D.; Donnio, B.; Pourroy, G.; Guillon, D.; Begin, S., Sensors and
Actuators B: Chemical, (2007) 126, 159.
[26] Hyeon, T., Chemical Communications, (2003), 927.
[27] James P. Collman, N. K. D., and Christopher E. D. Chidsey, Langmuir, (2004) 20, 1051.
[28] Ralph G. Nuzzo, B. R. Z., and Lawrence H. Dubois, JACS, (1986) 109, 733.

101
[29] F. Tielens, V. H., C-M. Pradier, M. Calatayud, F. Illas, Langmuir, (2009) 25, 9980.
[30] H. C. Kolb, M. G. F., K. B. Sharpless, Angew Chem Int Ed, (2001) 40, 2004.
[31] Moitra, N.; Moreau, J. J.; Cattoen, X.; Wong Chi Man, M., Chem Commun (Camb), (2010) 46,
8416.
[32] Toulemon, D.; Rastei, M. V.; Schmool, D.; Garitaonandia, J. S.; Lezama, L.; Cattoën, X.; Bégin-
Colin, S.; Pichon, B. P., Advanced Functional Materials, (2016) 26, 2454.
[33] Maalouli, N.; Barras, A.; Siriwardena, A.; Bouazaoui, M.; Boukherroub, R.; Szunerits, S., The
Analyst, (2013) 138, 805.
[34] Horcas, I.; Fernandez, R.; Gomez-Rodriguez, J. M.; Colchero, J.; Gomez-Herrero, J.; Baro, A. M.,
The Review of scientific instruments, (2007) 78, 013705.
[35] Nan, A.; Turcu, R.; Liebscher, J., Journal of Polymer Science Part A: Polymer Chemistry, (2012)
50, 1485.
[36] Cuenca, J. A.; Bugler, K.; Taylor, S.; Morgan, D.; Williams, P.; Bauer, J.; Porch, A., Journal of
physics. Condensed matter : an Institute of Physics journal, (2016) 28, 106002.
[37] Gilbert, J. B.; Rubner, M. F.; Cohen, R. E., Proceedings of the National Academy of Sciences of
the United States of America, (2013) 110, 6651.
[38] Gouget-Laemmel, A. C.; Yang, J.; Lodhi, M. A.; Siriwardena, A.; Aureau, D.; Boukherroub, R.;
Chazalviel, J. N.; Ozanam, F.; Szunerits, S., The Journal of Physical Chemistry C, (2013) 117, 368.
[39] Baio, J. E.; Weidner, T.; Brison, J.; Graham, D. J.; Gamble, L. J.; Castner, D. G., Journal of electron
spectroscopy and related phenomena, (2009) 172, 2.
[40] Vermette, P.; Gengenbach, T.; Divisekera, U.; Kambouris, P. A.; Griesser, H. J.; Meagher, L.,
Journal of Colloid and Interface Science, (2003) 259, 13.
[41] B. L. Frankamp, A. K. B., M. T. Tuominen, V. M. Rotello, JACS, (2005) 127, 9731.
[42] Poddar, P.; Telem-Shafir, T.; Fried, T.; Markovich, G., Physical Review B, (2002) 66,
[43] Kechrakos, D.; Trohidou, K. N., Journal of Nanoscience and Nanotechnology, (2008) 8, 2929.
[44] Dorothy Farrell, Y. C., R. William McCallum, Madhur Sachan, Sara A. Majetich, J Phys Chem C,
(2005) 109, 13409.
[45] Klughertz, G.; Manfredi, G.; Hervieux, P.-A.; Pichon, B. P.; Begin-Colin, S., The Journal of Physical
Chemistry C, (2016) 120, 7381.
[46] Fleutot, S. N., G. L.; Pauly, M.; Pichon, B. P.; Leuvrey, C.; Drillon, M.; Gallani, J.-L.; Guillon, D.;
Donnio, B.; Begin-Colin, S, Nanoscale, (2013) 5, 1507.

102
III. CHAPITRE III –Nanoparticle self-assembly by multiple
hydrogen binding between nucleosides

103
 A. Nanoparticle assembly through nucleosides

The self-assembly through multiple hydrogen bonding allow more flexible and dynamic in
geometry than rigid covalent bonds. Moreover, they are directional and can be rendered highly
selective by exploiting multiple hydrogen bonding molecules. Here, the strategy consists in the
functionalization of nanoparticles and substrates by complementary group for click chemistry and graft
subsequently modified nucleosides.

The recognition process will use the self-complementary Watson-Crick multiple hydrogen-
bonding interactions nucleobases pairing. This process involves the recognition between nucleosides
from the DNA: guanine-cytosine (GC) and adenine-thymine (AT)

An essential part of the assembly through nucleobase pairing is to understand the assembly
process through multiple hydrogen bonds. The interaction energy for hydrogen bonds is usually
defined between 10 and 50 kJ.mol-1[1], however this determination is not trivial since the recognition
involves a high number of groups between the surfaces. Theoretical investigations have been
performed to study the interaction energies between nucleobases and showed energies around 20
kJ.mol-1 (for the A-T pairs) and 50 kJ.mol-1 (for the G-C pairs).[2-4] Nevertheless, many parameters can
influence the strength of the binding:

- The solvent can induce a competition to the pairing. Indeed, protic and polar solvents will favor
recognition with the nucleobases and therefore not favor the base pairing in contrary to
aprotic solvents which do not generate competition.
- π-stacking between close packed rings decreases the interaction energy and therefore reduce
the strength of the binding between pairs.
- The temperature can weaken the interactions between nucleobases. These binding usually
break around 55°C.

These parameters shows that some requirements are needed to perform the assembly. Investigations
on the solvent used for the assembly showed reversibility on the attachment of polymer onto substrate
surface functionalized by complementary nucleobases.[5-8]

The couple adenine-thymine will be use in this study by functionalizing both nanoparticle and substrate
surfaces with complementary nucleobases. The recognition happens by dipping the functionalized
substrate in the nanoparticle suspensions at room temperature. The influence of experimental
parameters such as the solvent, the pi-stacking and the temperature, was studied.

1. Substrate post-functionalization

104
 In order to perform the recognition process, the surfaces have to be functionalized with
nucleosides. The use of modified nucleosides with azide and alkyne groups will allow their grafting
onto surfaces by CuAAC cycloaddition “click” reaction. Adenine modified by an azide group and
thymine modified with an alkyne group are used (figure III-1).4 The atom indexation in the ring will
facilitate the infrared interpretation.

a) b)

Figure III-1. Schematic representation of a) alkyne-modified thymine and b) azide-modified adenine

a) Experimental procedure

As see in the previous chapter, the most efficient system is the alkyne-terminated SAM. The
adenine modified with azide group is used to be grafted through click chemistry reaction on a SAM
alkyne surface.

1-Dodecanethiol (DDT) was used to dilute alkyne groups at the surface prior to control the
quantity of adenine groups at the surface. During the gold functionalization, substrates were immersed
in a thiol solution with different molar ratios of DDT and 11-mercapto undecyn (MUY) with a total
concentration in thiol of 10 mM. The ratio MUY/DDT was varied from 0% to 100% and SAM were called
SAM-MUY/DDT 0%, 20%, 50%, 80% and 100%. Azide-modified adenine (Ade-N3) was then grafted to
the alkyne on the surface of SAM prepared with different MUY/DDT ratios (figure III-2).

Figure III-2. Schematic representation of the substrate functionalization assisted by micro-wave (MW) irradiation

The CuAAC “click” reaction was performed by the same protocol than for the assembly of
nanoparticles (chapter II) by using copper I catalyst. A large excess of modified nucleosides are used in
order to promote the kinetics and diffusion of the molecules to the surface. In order to increase the

4
Collaboration with X. Cattoën, Institut Néel, Grenoble.

105
efficiency of the process, the reaction was performed under microwave irradiations to decrease
strongly the time of reaction. Indeed, the activation by micro-wave irradiation presents a different way
of heating since the activation energy will directly stimulate the solvent molecules instead of classic
heating process where the vessel is heated and the heat diffuse in the liquid.[9-12] The assembly solution
was prepared with 6.5 mg CuBr(PPh3)3, 5 mL of THF, 0.5 mL of triethylamine and 10 mg (49 μmol) of
azide-modified adenine (Ade-N3). The reagents were mixed in a vial, settled in a microwave reactor
and heated until 120°C with a constant power of 50 W at a 2.45 GHz frequency. Then, the temperature
was maintained at 120°C for 30 min before cooled at room temperature.

b) Characterization

The SAM surface modified with adenine was characterized in order to study the presence of
the nucleosides and to evaluate its quantity.

(1) Water contact angle

Contact angle measurements were performed on SAM with different MUY/DDT ratios after
grafting of the adenine (figure III-3).

Figure III-3. Water contact angle values as function of the MUY/DDT molar ratio in solution

The water contact angle value decreases when the MUY/DDT molar ratio increase. The higher
hydrophilicity of the substrate for high MUY/DDT ratio suggests a higher quantity of adenine. Indeed,
the adenine presents a hydrophilic structure with the presence of polar groups. Therefore, the
substrate becomes more hydrophilic with the increase of adenine. The contact angle runs from 105°
when the surface is covered at 100% with DDT, to 33 % for a full coverage with adenine. The value
decreases quasi-linearly which means that the quantity of adenine roughly corresponds to the ratio in
solution.

106
 (2) X-Ray photoelectron spectroscopy

XPS measurements were performed on the SAM-Ade 100% in order to control the presence of
adenine on the SAM surface. The adenine molecule presents nitrogen atoms. Thus, the N1S region was
probed to attest the presence of adenine groups (figure III-4).

Figure III-4. XPS spectrum of the N1S region for the SAM-Ade 100%

The large and intense peak around 400 eV shows the significant presence of nitrogen at the
surface. The peak can be deconvoluted in two components at 399.6 eV and 400.8 eV. The first
component can be attributed to the N-H and the nitrogen in aromatic ring whereas the second
correspond to the N-C binding present in the position 9 of the adenine.[13-15] The component at 400.8
eV can also be attributed for the N-C binding in the triazole bridge.

(3) Phase Modulated Infrared reflection absorption spectroscopy

PM-IRRAS were also performed on the SAM-Ade 100%. The results are compared to adenine
deposited on a gold surface by drop casting (figure III-5).

107
 Figure III-5. PM-IRRAS spectra of adenine dropcasted on a surface (red curve) and grafted through click chemistry on
alkyne SAM (blue curve)

The red curve shows the azido-modified adenine molecule deposited by drop casting on the
substrate. Vibrational bands of the nucleosides can be identified in the range 800 - 2500 cm-1. The
nucleosides present a large variety of vibrational bands between 800 cm-1 and 1500 cm-1 which have
been listed in the literature.[16, 17] However, we will focus on the main and more intense bands at 1660
cm-1 and 1580 cm-1 which are attributed to the bending vibrational mode in amine δ(N-H) and the
stretching mode ν(N3-C4). Moreover, the nucleoside has been modified by an azide group which can
be noticed thanks to the characteristic vibrational band at 2100 cm-1. The blue curve shows the surface
after the click chemistry reaction between the azido adenine and the alkyne SAM. The presence of
adenine is confirmed by the strong vibrational band at 1660 cm-1. There are not anymore two fine
bands, but a broad one which may be explained by the fact that the adenine groups are stacked at the
nanoparticle surface in different orientations. Moreover, the characteristic band at 2100 cm-1 is not
visible which agree with the reaction of azide and alkyne groups. This last information confirms the
grafting of adenine through triazole bridge to the SAM.

The XPS and PM-IRRAS showed the apparition of signature of functions present in the adenine
and the disappearance of azide groups. It confirms the grafting of the adenine onto the SAM surface.
Moreover, the WCA showed that different MUY/DDT ratios conduct to the quasi-linear change of the
hydrophobicity of the SAM after grafting of azido-modified adenine grafting. It confirms the control of
the quantity of adenine at the surface of the SAM. The substrates with different ratio in adenine were
prepared in order to realize self-assembly of the nanoparticles.

2. Nanoparticle post-functionalization

a) Control of the quantity of azide groups at the surface

The 10 nm-sized nanoparticles functionalized with azide groups (cf. chapter II) were used to
graft alkyne-modified thymine (figure III-1.a). In order to control the pi-stacking, as on the substrate,
thymine groups were diluted at the surface of the nanoparticles. In order to change de density of
thymine groups at the surface, the nanoparticles were functionalized with different phosphonic acid

108
molecules. The purpose was to lead to mixed nanoparticles functionalized with azide groups and
methyl groups which are inactive for the CuAAC click reaction. The functionalization was performed in
THF with different ratios of 12-azido-dodecyl-phosphonic acid (AP12N3) and decyl-phosphonic acid
(DPA) represented on figure III-6; 20%, 50%, 80% and 100%. The nanoparticles were then washed by
ultrafiltration. The functionalization and the stability were monitored by FT-IR spectroscopy and
granulometry.

a)

b)

Figure III-6. Schematic representation of the a) 12-azido-dodecyl-phosphonic acid (AP12N3) and b) decyl-phosphonic acid
(DPA)

The alkyl chain is slightly longer for AP12N3 than DPA in order to favor the accessibility of azide
groups at the nanoparticle surface.

(1) FT-IR spectroscopy measurements

The infrared spectroscopy measurements were performed in order to control the quantity of
biotin groups at the nanoparticle surface (figure III-7). The spectra were normalized on the vibrational
band of FeO at 590 cm-1.

Figure III-7. FT-IR spectra of the nanoparticles functionalized with different AP12N3/DPA ratios

The FT-IR spectroscopy measurements show the vibrational bands of iron oxide, phosphonic
acid and azide groups at 590 cm-1, 1080 cm-1 and 2100 cm-1 respectively. This indicates the correct

109
functionalization of the nanoparticles. The intensity ratio between iron oxide and phosphonic acid is
very close whatever the AP12N3/DPA ratio which agree with similar grafting rate for both AP12N3 and
DPA molecules. However, the azide vibrational band decreases with the quantity of DPA molecules
which were introduced in the reaction medium. Moreover, the vibrational band corresponding to the
stretching modes of the CH3 groups (2950 cm-1) increases with DPA molecules. The ratios of the
different intensity peaks are summarized in table III-1.
Table III-1. Summarize of the intensity ratios between the vibrational bands ν(C-H) (2950 cm-1), ν(Fe-O) (580cm-1) and
ν(N3) (2100 cm-1)

Peak intensity ratios

CH3/Fe-O N3/Fe-O
Azide phosphonic ratio in
solution
100 % 0,36 1,42
80 % 0,37 0,95
50 % 0,78 0,62
20 % 0,83 0,45

The ratio between azide and iron oxide vibrational bands increased with the ratio of
AP12N3/DPA molar ratio in solution. In contrast, the ratio between methylene and iron oxide
vibrational bands decreased. These results showed that azide groups at the nanoparticles surface can
be modulated as function of the molecular ratio in solution.

(2) DLS measurements

DLS measurements were performed in order to study the colloidal stability of the suspensions
in THF after the functionalization with different molecular ratios. The results in THF are shows in figure
III-8.

Figure III-8. DLS measurements of the nanoparticle suspensions with different AP12N3/DPA ratios in solution

110
 The nanoparticles are stable for the four different ratios studied. The size distribution is
monomodal, meaning no aggregation of the suspension and a unique size population. Some
differences can be observed (from 10.1 nm to 15.5 nm) which are not discrediting the stability of these
suspensions.

b) Post-functionalization with thymine groups

The mixed-nanoparticles were then post-functionalized with the alkyne modified thymine
groups. The CuAAC reaction was performed by using the nanoparticle suspensions with different
quantities of azide groups at their surface.

As for the grafting of adenine, a large excess of modified thymine was used to promote the
kinetics and diffusion of the molecules to the surface. Moreover, the reaction was performed under
microwave irradiation for only 5 minutes. Indeed, the motion of nanoparticles promotes the encounter
with molecules and fosters the reaction kinetics. Then, the suspension was purified by ultrafiltration
in order to remove the excess of molecules and all the reagents present in the solution assembly (figure
III-9).

Figure III-9. Schematic representation of the “click” reaction to graft the thymine group at the nanoparticle surface

The nanoparticles after purification were characterized in order to study the presence of
thymine at their surface and to control their colloidal stability. The nanoparticles functionalized with
the thymine for the different AP12N3/DPA ratios are called NP@Thym 20%, 50%, 80% and 100%.

c) Characterization of the NP@Thym

As for the substrate, the functionalized nanoparticles have been characterized in order to
confirm the presence of the thymine species at their surface.

(1) TEM analysis

Transmission electronic microscopy was performed to study the size and morphology after the
multiple functionalization processes (figure III-10).

111
 Figure III-10. TEM images of nanoparticles after functionalization with alkyne-modified thymine 100%

The morphology of the nanoparticles was not altered by the functionalization. The
nanoparticle mean diameter was centered to 8.5 ± 0.8 nm whereas the NP@OA was at 9.1 ± 0.6 nm.
The size remains constant (according to the standard deviation) even if a slight decrease can be
noticed.

(2) FT-IR measurements

FT-IR measurements were performed on the samples with different AP12N3/DPA molar ratios
after the reaction with alkyne-modified thymine (figure III-11).

Figure III-11. Infrared spectra of NP@thym with different thymine ratio at the surface

The infrared spectrum of the free molecule of alkyne-modified thymine presents a large variety
of vibrational bands in the range 500-1700 cm-1. The specific and weak band at 2115 cm-1 is
characteristic of the stretching mode of the alkyne binding ν(C≡C). The more intense bands which allow
identifying the thymine are located at 1710 cm-1 and 1660 cm-1. These bands correspond to the

112
stretching mode of the carboxylic acid in the thymine ν(C2=O) and ν(C4=O), respectively.[17, 18] The
spectra of the functionalized NP@thym with different ratios exhibit for all this a large band centered
around 1680 cm-1 which agree with the presence of the thymine at the nanoparticle surface. In the
case of the highest ratios of thymine, a single band at 1680 cm-1 was observed. For the lowest ratios,
two bands at 1710 cm-1 and 1660 cm-1 were observed as for the free molecule. The conformation of
the C=O bonds is different for the high quantity of thymine at the surface, probably due to the stacking
of the thymine groups at the nanoparticle surface. For low quantities of thymine, such a stacking
disappears and the binding modes are closer from the molecule in solution.

The disappearance of the band at 2115 cm-1 shows that the free thymine molecules in solution
were totally removed and the purification step has been well performed. Moreover, the disappearance
of the band at 2100 cm-1 reveals that all the azide groups at the nanoparticle surface have totally
reacted to form triazole bonds.

(3) DLS measurements

Granulometry measurements have been performed on the NP@Thym in order to study the
colloidal stability after the post-functionalization. The control of the aggregation in the following part
becomes a crucial point. Therefore, the NP@Thym 100% will be distinguish from the others and their
stability will be investigated first (figure III-12).

Figure III-12. DLS measurements of the NP@Thym 100% in THF (blue curve) and DMF (red curve)

The click reaction and the purification were both performed in THF. The DLS measurements
(figure III-12, blue curve) exhibit a hydrodynamic diameter centered to 100 nm. Considering the
average size of nanoparticles at 8.5 nm, this value indicates their aggregation in the suspension. A first
hypothesis could be that the nanoparticles aggregate during the “click” reaction even after very short
times (5 min). However, the DLS measurement performed after evaporation of the THF and dispersion
in dimethylformamide (DMF) showed a single peak centered to 12 nm, meaning no aggregation and a

113
good stability of the suspension (figure III-12, red curve). This last information proves that the
aggregation does not come from the “click” reaction but from the interaction with the solvent.

It is important to understand how the solvent interacts with the nanoparticle surface to
manage the assembly thereafter. Here, the nanoparticles are covered with thymine which presents
few polar groups which are able to create hydrogen bonding with electronegative atoms. The thymine
groups at the nanoparticle surface may create bonds with another thymine group which favors the
aggregation in solution. With protic solvents, bonds also exist with the thymine groups. Therefore,
solvent-thymine interactions compete with thymine-thymine interactions.[8, 19, 20]

To ensure the stability of the suspension, a solvent which promotes the solvent-thymine
interaction (thus limit the thymine-thymine interaction) is required. The DMF can generate easily
hydrogen bonds because of the presence of hydrogen and electronegative atoms in its structure (figure
III-13). Therefore, this solvent is more favorable to avoid aggregation. This aggregation is reversible by
changing the solvent of the nanoparticle suspension. Indeed, the solvent exchange was performed
with the same sample and the stability of the nanoparticles was the same, aggregated in THF and stable
in DMF.

a) b)

Figure III-13. Schematic representation of the molecules of a) THF and b) DMF

The stability has been studied for each ratio of thymine groups at the surface of the
nanoparticles. Table III-2 summarizes the stability of the suspension as function of the solvent.
Table III-2. Stability of the NP@Thym with different ratio of azide groups at the surface as function of the solvent

As expected, the nanoparticles with different quantities of thymine groups at the surface have
stability which varies with the solvent polarity. Nanoparticles are aggregated in water whatever the
quantity of thymine at their surface. The DMF allow a good dispersion of the NP@Thym 100%, 80%
and 50%, but below this value, the NP@Thym 20% are not stable. For these nanoparticles, the THF and
chloroform allow a good dispersion. The DLS measurements were performed in the solvent where the
nanoparticle suspension are the more stable (figure III-14).

114
 Figure III-14. DLS measurements for nanoparticles with different ratio of thymine at the surface in DMF for 100%, 80%
and 50% and CHCl3 for 20% and 0%

The suspensions of NP@thym are stable and do not present aggregation as long as they are in
the right solvent. The small differences of the hydrodynamic diameters may come from the formation
of small clusters in solution due to the solvent.

FT-IR and DLS allow a first understanding of the mechanisms of aggregation of the
nanoparticles covered with different quantities of thymine groups. The control of the stability by using
the solvent is an essential point that will be useful for the assembly through the multiple hydrogen
bonding.

3. Molecular recognition through nucleosides

The surfaces of both substrates and nanoparticles have been functionalized with
complementary nucleosides adenine and thymine groups. The recognition can be established to
assemble the nanoparticles at the surface. Here, the influence of the parameters such as the solvent,
the density of functional groups and the temperature are investigated to understand the mechanisms
of assembly through the nucleosides.

a) Experimental protocol

The assembly process is quite simple: substrates with different ratios of adenine were dipped
in 5 mL of a suspension of nanoparticles with different ratios of thymine (figure III-15). The substrates
were removed after one hour at room temperature. Then, they were extensively rinsed with DMF and
exposed for 15 seconds in ultrasonic bath in order to remove physisorbed nanoparticles before being
dried under air stream.

115
 Figure III-15. Schematic representation of the assembly process through multiple hydrogen bonding

b) Influence of experimental parameters

(1) Assembly time

The influence of the time where the thymine-terminated nanoparticles are in contact with the
adenine-terminated SAM is important to know the kinetics of the assembly. Therefore, different
immersion times of the SAM-Ade 100% in a suspension of NP@Thym 100% in DMF at the same
concentration were performed. The SEM images are shown in figure III-16 and the density values were
compared to maximum theoretical values calculated for a hexagonal close packed nanoparticle
assembly with a mean diameter of 8.5 ± 0.8 nm (table III-3).

a) b) c) d)

30 sec 1 min 10 min 60 min

Figure III-16. SEM images of assembly of NP@Thym on SAM-Ade for different assembly times: a) 30 sec, b) 1 min, c) 10
min and d) 60 min. Magnification X100 000

The SEM images show the increase of densities of nanoparticles at the SAM surface as function
of the immersion time. After 30 seconds, the recognition does not happen and there is no
nanoparticles at the surface (<1% of the maximal theoretical value). After 1 minute, some
nanoparticles are observed with a very low density (2.4%). After 10 minutes of immersion, the density
of nanoparticles stops increasing around 13% of the maximal theoretical density.

116
Table III-3. Density of nanoparticles as function of the time dipping time

30 sec 1 min 10 min 60 min

Density of nanoparticles
15 ± 15 115 ± 85 600 ± 45 625 ± 55
(NP/μm²)
Surface coverage (%) <1 2,4 ± 0,2 12,6 ± 1,0 13,2 ± 1,2

The density for a long reaction time remains far below the theoretical full coverage of the
surface. The concentration should have an impact on the diffusion of the nanoparticles on the surface
and therefore on the kinetics. However, due to the time-consuming ultrafiltration step to purify the
nanoparticles, the influence of the concentration could not be studied and remains low (≈ 0.5 mg/mL).

Moreover, isolated nanoparticles were observed on each sample. No clusters or chains of

nanoparticles were observed in contrast with the assembly via “click” chemistry reaction. This
observation shows that the assembly mechanism driven by nucleosides is different from the “click”
strategy.

(2) Influence of the quantity of functional groups at the

nanoparticle and SAM surfaces

One of the interactions which can prevent the recognition between nucleosides is the π-
stacking.[2, 3, 21-23] Indeed, the stacking interaction between rings of the nucleobases, generally weakens
the hydrogen bonds.[2] To study the influence of the π-stacking on the assembly, nucleobases on the
nanoparticles and substrates surfaces have been diluted.

(a) On the SAM surface

Self-assembled monolayers were realized with different CC/DDT ratios in solution in order to
mix adenine groups at the gold surface with methylene-terminal group which are non-active for the
base pairing. The assemblies have been prepared by dipping SAM-Ade 0%, 20%, 50%, 80% and 100%
in a NP@Thym 100% DMF solution for 30 min. The nanoparticle densities were measured from the
SEM images (figure III-17) and are summarized in table III-4.

117
 Figure III-17. SEM images SAM-Ade with different ratios: a) 0%, b) 20%, c) 50% d) 80% e) 100% after dipping in a
suspension of NP@Thym 100% in DMF. Magnification X100 000

Table III-4. Surface coverage of NP@Thym 100% assemblies on SAM-Ade with different ratios of adenine at the surface.

SAM-Ade 0% 20 % 50 % 80 % 100 %
NP@Thym
<1% 6.8 ± 1.1 % 8.7 ± 1.6 % 11.2 ± 1.2 % 9.1 ± 0.9 %
100 %

The surface coverage is very low for each sample (below 15% of the maximal theoretical value).
Nevertheless, a trend can be observed. Indeed, the number of nanoparticles at the surface increases
up to 11% with the quantity of adenine at the surface of the SAM until 80%. This trend does not seem
to attest a strong influence of π-stacking. However, a slight decrease is observed for the SAM-Ade
100% in comparison with the 80%.

Decreasing the adenine at the surface does not allow increasing the density of nanoparticles
which can be explained by reducing the number of nucleobases at the surface and so decreasing the
recognition rate. The number of nanoparticles is still low at the surface and the π-stacking effect is not
responsible for such low density.

Considering the previous results, SAM-Ade 80% will be used in the following section in order
to optimize the density of nanoparticles at the surface.

(b) On the nanoparticle surface

In order to control the influence of π-stacking at the nanoparticle surface, thymine groups
were diluted on the surface. The assembly took place for 30 min by dipping a SAM-Ade 80%. The
densities values are summarized in table III-5.

118
 Table III-5. Density of NP@Thym assembly for different ratios of thymine on SAM-Ade 80%

NP@Thym 20 % 50 % 80 % 100 %

SAM 80% <1% 42,1 ± 2,4 % 12,3 ± 1,0 % 11 %

The results measured on the SEM images show that the best density was obtained for
nanoparticles covered with 50% of thymine groups. The investigation of the influence of π-stacking
showed that the optimal system is the one with NP@Thym 50% grafted on SAM-Ade 80%. The density
of these assembly can be modulated by the quantity of nucleosides at the surface, however, other
parameters can contribute to improve the recognition such as solvent and temperature.

(3) Solvent influence

As shown previously, the solvent has a strong influence on the stability of the nanoparticles
covered with thymine groups. Indeed, the solvent interacts by hydrogen bonding with the nucleosides
localized on the substrates and nanoparticle surfaces. In order to improve the recognition between
adenine and thymine, a solvent which does not interferes with the multiple hydrogen bonding has to
be used. Two solvents have been used with different polarity to observe the effect on the assembly.
Different DMF/CHCl3 volume ratios were used: [4:1], [1:1] and [1:4]. The SAM-Ade 80% were dipped
for 30 min at room temperature in NP@Thym 50% suspension. SEM images are shown in figure III-18.

Figure III-18. SEM images of assemblies of NP@Thym 50% on SAM-Ade 80% with different ratios of DMF/CHCl3 a) [4:1],
b) [1:1] and c) [1:4]

As observed previously, the nanoparticles are isolated and do not present any clusters or
alignments at the SAM surface. With the information obtained earlier, we can assume that the
nanoparticles prevent the π-stacking effect and cannot be too close from each other because the
energy required to create a binding is higher.

The density increases up to the DMF/CHCl3 ratio [1:1] at 53 % before decreasing when the
quantity of chloroform increases. This phenomenon can be explained with a complementary
characterization method such as granulometry. DLS measurements were performed on each
suspension with different ratios of DMF/CHCl3 and are presented on figure III-19. The hydrodynamic
diameters are summarized in table III-6.

119
 Figure III-19. DLS measurements of NP@Thym 50% in different DMF/CHCl3 ratios

The hydrodynamic diameter of the nanoparticles evolves with the ratio DMF/CHCl3. The
suspension of nanoparticles has a very good stability in DMF. Therefore, thymine-DMF interactions
can compete with the thymine-thymine interactions. In CHCl3 the suspension is totally aggregated with
a single peak around 1300 nm which agrees with weak interactions between the solvent and the
thymine group and thus, which promoting the recognition thymine-thymine between nanoparticles.
Table III-6. Density of NP@Thym 50% assemblies on SAM-Ade 80% for different DMF/CHCl3 ratios

[4:1] [1:1] [1:4]

Density 19% 53% 18%

Hydrodynamic
13,6 nm 21,3 nm 1274 nm
Diameter

The stability with ratios containing few CHCl3 [4:1] is similar to pure DMF without aggregation
of the nanoparticles. In contrast, the one containing few DMF [1:4] is similar to CHCl3 with total
aggregation of the suspension. In the case of a [1:1] ratio, nanoparticles start to aggregate with a
hydrodynamic diameter centered to 21 nm. The stability can be directly influenced by the polarity of
the solvent which can control the hydrogen bonding between the nucleosides at the surface of the
nanoparticles and gold substrate.

The stability of the nanoparticle suspension can be directly correlated to the structure of
nanoparticles assembly. The energies of the different interactions in the system can be noted as
follows:

- ET-T : the energy necessary to form thymine-thymine bonding

- ET-S : the energy necessary to form interaction between thymine and solvent molecule
- ET-A : the energy necessary to form thymine-adenine bonding

The system will always minimize the energy and allow the formation of bonds which requires
the lowest energy interaction. However, the energy to form thymine-adenine bonds will always be

120
lower than the energy to form thymine-thymine bonds (figure III-20).[2] Immersion of a SAM-Ade in
NP@Thym would technically lead to the formation of adenine-thymine bonds before thymine-thymine
bonds (ET-A < ET-T). Therefore, assembly of nanoparticles at the SAM surface is favored against
nanoparticle aggregation in solution.

a) b)

Figure III-20. Schematic representation of the hydrogen bonds between a) adenine-thymine and b) thymine-thymine

However, the interactions between the solvent and the thymine at the nanoparticle surface
may compete the assembly process. Three cases have been studied as function of the solvent used:

Ø ET-A < ET-T < ET-S

The use of a solvent with a low polarity and thus, not able to form hydrogen bonds with
nucleobases will allow the formation of bonds between nucleosides. However, in suspension, the
nanoparticles will be aggregated because of the non-competition of the solvent to avoid the formation
of thymine-thymine bonds. It leads to nanoparticle aggregation for low DMF/CHCl3 ratios ([1:4] and
[0:1]). Moreover, the immersion of SAM-Ade leads to a low density of nanoparticles. This can be
explained by the fact that aggregates do not have the energy to create bonds with the surface and lead
to low nanoparticle density (18%).

Ø ET-S < ET-A < ET-T

In another hand, using a solvent which can create bonds with nucleosides and compete with
the formation of thymine-thymine bonds does not lead to the aggregation of the suspension.
Unfortunately, the use of DMF does not allow having a high density of the nanoparticles after
immersion of the SAM-Ade. The explanation comes from the solvent interactions with nucleosides is
too strong and disturb the recognition adenine-thymine. In this case, nanoparticles cannot be grafted
on the surface because their stability in suspension is too “good”.

Ø ET-A < ET-S < ET-T

In order to avoid a solvent which compete too much with the recognition process between
adenine and thymine and another one which promotes aggregation of the suspension, a right solvent
has to be found to allow the recognition and to favor the stability of the suspension. A mixture of
chloroform and DMF in ratio [1:1] shows the highest density at the surface (53%). The energy necessary
to form adenine-thymine bonds is lower than the energy to form bonds with the solvent and allow the
grafting of the nanoparticles. The DLS measurements show a small shift of the hydrodynamic diameter
to 21.3 nm meaning the start of the aggregation. The system is slightly destabilized which allows the
formation of bonds of the nanoparticles at the surface of the SAM-Ade.

The solvent has a strong influence on the recognition. Indeed it will promote or not the
interactions between nucleosides localized at the surface of both nanoparticles and gold substrates.[24]

121
 c) Influence of the temperature

Hydrogen bonds can be easily broken with thermal energy. Therefore, the influence of the
temperature on the structuration of the assembly was studied. A sample corresponding to an
optimized density of nanoparticles (57%) was dipped in DMF which favors the solvation of the
nanoparticles and heated at 60°C for 2 hours. The SEM images before and after heating are presented
on the figure III-21.

a) b)

Figure III-21. SEM images of a NP@Thym 50% assembly on SAM-Ade 80% a) before and b) after heating at 60°C in DMF
for 2 h

The density of nanoparticles decreases after heating in DMF and pass from 57% to 18%. It
shows that the hydrogen bonds were broken with the increase of the thermal energy. However, some
nanoparticles remain at the substrate surface, meaning there is not a fully desorption and shows a
certain stability of the system.

d) Influence of the rinsing step

The assembly of nanoparticles onto SAMs is dependent of the solvent as seen previously. The
rinsing step should also have an influence on nanoparticle removal after recognition by rinsing with a
solvent promoting the interactions with the nucleosides. In order to study the influence of the solvent
during the rinsing step, assemblies of NP@Thym 50% on SAM-Ade 80% performed with DMF/CHCl3
[1:1] mixture were rinsed with different solvents. The different solvents used and the density of
nanoparticles are summarized in table III-7.
Table III-7. Density of nanoparticles as function of the rinsing solvent

DMF/CHCl3 [1:0] [1:1] [0:1]

Density
2015 ± 260 2150 ± 80 2840 ± 165
(NPs/μm²)
Density 42.5 ± 5.5 % 45.3 ± 1.7 % 59.8 ± 3.5 %

Rinsing the assembly using CHCl3 which does not interfere in the formation of adenine-thymine
binding should not remove nanoparticles. In contrast, DMF should favor the removal of the

122
nanoparticles. Considering the density measured on the SEM image a trend can be observed with a
slight increase of the density with the rinsing in CHCl3. However, the reproducibility and the errors on
the measurements show that we have to be careful with these results. The results in CHCl3 were
expected as long as this solvent is not able to break the hydrogen binding between the nanoparticles
and the substrate. In another hand, for the DMF and the mix DMF/CHCl3 [1:1] which are able to break
the binding as shown with the DLS measurements, a removal of the nanoparticles could be expected.
Instead, except a slightly decrease in density, the nanoparticles remain at the SAM surface. The system
shows a certain stability since the nanoparticles stay at the surface after rinsing with DMF. This can be
explained by the fact that the nanoparticles are not grafted with a single nucleoside at the surface but
with few which increase the stability.

B. Conclusion

By taking advantage of the “click” chemistry, the nanoparticles and substrates were post-
functionalized with complementary nucleobases. The use of micro-wave irradiations allowed
decreasing strongly the reaction time down to few minutes. The hydrogen bonds formed during the
assembly were sensitive to external parameters such as the π-stacking, the solvent and the
temperature. Understanding the influence of these parameters allow controlling the structure of
nanoparticle assembly.

This work leads to perspectives. Indeed, the influence of the chain length at the surface of
nanoparticle and substrate or the presence of hydrophilic groups such as PEG in their structure can be
investigated.

Moreover, the control of stability of two types of nanoparticles functionalized with thymine
and adenine can be investigated. The control of the stability can be done by playing with the nature of
the solvent or the temperature. Such a kind of nanoparticle suspension incorporating both
nucleobases may allow the control of self-assembly in multilayers.

The control of the nanoparticle assembly through base pairing could allow to go further and
envisioned the assembly through more specific interactions such as the recognition through DNA
strands.[25-28]

123
 C. Bibliography

[1] Freek J. M. Hoeben, P. J., E. W. Meijer, and Albertus P. H. J. Schenning, Chem. Rev, (2005),
[2] Villani, G., J Phys Chem B, (2014) 118, 5439.
[3] Villani, G., Physical chemistry chemical physics : PCCP, (2013) 15, 19242.
[4] Cerny, J.; Hobza, P., Chem Commun (Camb), (2010) 46, 383.
[5] Viswanathan, K.; Long, T. E.; Ward, T. C., Langmuir, (2009) 25, 6808.
[6] K. Viswanathan, H. O., C. L. Elkins, C. Hesey,T. C. Ward, T. E. Long, Langmuir, (2006) 22, 1099.
[7] A. Sanyal, T. B. N., O. Uzun, V. M. Rotello, Langmuir, (2004) 20, 5958.
[8] Binder, W. H.; Lomoschitz, M.; Sachsenhofer, R.; Friedbacher, G., Journal of Nanomaterials,
(2009) 2009, 1.
[9] Toulemon, D.; Pichon, B. P.; Leuvrey, C.; Zafeiratos, S.; Papaefthimiou, V.; Cattoën, X.; Bégin-
Colin, S., Chemistry of Materials, (2013) 25, 2849.
[10] Erumpukuthickal Ashok Anumol, P. K., Parag Arvind Deshpande, Giridhar Madras, and
Narayanan Ravishankar, ACS Nano, (2011) 5, 8049.
[11] Oliver Kappe, C., Chemical Society Reviews, (2008) 37, 1127.
[12] MINGOS, D. M. P., Res. Chem. lntermed, (1994) 20, 85.
[13] Mohtasebi, A.; Chowdhury, T.; Hsu, L. H. H.; Biesinger, M. C.; Kruse, P., The Journal of Physical
Chemistry C, (2016) 120, 29248.
[14] Pagliai, M.; Caporali, S.; Muniz-Miranda, M.; Pratesi, G.; Schettino, V., The Journal of Physical
Chemistry Letters, (2012) 3, 242.
[15] Furukawa, M.; Yamada, T.; Katano, S.; Kawai, M.; Ogasawara, H.; Nilsson, A., Surface Science,
(2007) 601, 5433.
[16] Yoshimasa Kyogoku, R. C. L., and Alexander Rich, JACS, (1967) 89, 496.
[17] Pina Colarusso, K. Z., Bujin Guo, Peter E Bernath, Chemical Physics Letters, (1997) 269, 39.
[18] Singh, J. S., Journal of Molecular Structure, (2008) 876, 127.
[19] Kalpana Viswanathan, H. O., Casey L. Elkins, Cheryl Heisey,Thomas C. Ward, and Timothy E.
Long, Langmuir, (2006) 22, 1099.
[20] Andrew K. Boal, F. I., Jason E. DeRouchey, Thomas Thurn-Albrecht, Thomas P. Russell &
Vincent M. Rotello, Nature, (2000),
[21] Goodman, R. S. P. a. J. M., J. Chem. Inf. Model., (2009) 49, 944.
[22] Marlon N. Manalo, L. M. P., and Andy LiWang, JACS, (2006),
[23] Mignon, P.; Loverix, S.; Steyaert, J.; Geerlings, P., Nucleic acids research, (2005) 33, 1779.
[24] Stolarczyk, J. K.; Ghosh, S.; Brougham, D. F., Angew Chem Int Ed Engl, (2009) 48, 175.
[25] Chevolot, Y.; Laurenceau, E.; Phaner-Goutorbe, M.; Monnier, V.; Souteyrand, E.; Meyer, A.;
Gehin, T.; Vasseur, J. J.; Morvan, F., Current opinion in chemical biology, (2014) 18, 46.
[26] James J. Storhoff, A. A. L., Robert C. Mucic, Chad A. Mirkin, Robert L. Letsinger, and George C.
Schatz, J. Am. Chem. Soc., (2000) 122, 4640.
[27] A. Csaki, R. M., J. M. Köhler, W. Fritzsche, Nucleic Acids Research, (2000) 28, 91.
[28] Chad. A. Mirkin, R. L., R. Mucic, J. Storhoff, Letter to Nature, (1996) 382, 607.

124
CHAPITRE IV - Biomolecules Recognition mediated by iron oxide
nanoparticle assemblies supported onto gold thin films

125
 The assembly of iron oxide nanoparticles can be used to design a very efficient detection
platform by taking advantage of the surface plasmon resonance (SPR) properties of the gold substrate.
Receptor molecules can be easily grafted at the nanoparticle surface in order to detect target
molecules through specific interactions.

We previously saw that the “click” reaction is a powerful tool for tuning the structuration of
the assembly. Moreover, the versatility of this method ensures the grafting of a large variety of
receptor molecules. Furthermore, the covalent binding of the receptor molecules is preferred since it
provides strong and stable binding. The analyte can also be removed to regenerate the sensors without
removing the receptor molecules which is strongly anchored at the surface. Therefore, receptor
molecules can be modified with alkyne groups in order to perform the CuAAC “click” reaction with
azide groups at the surface of nanoparticles (Figure IV.1).

Figure 0-1. Schematic representation of the grafting of an alkynated derivative bioreceptor molecules on the azide
terminated nanoparticle assembly

The grafting of biomolecules on the surface can be characterized by surface analysis

techniques in order to control the chemical change at the surface of nanoparticles. This platform is
then available for the detection of the recognition element. The gold surface provides a SPR signal and
allows the detection of the analyte. The high surface/volume ratio of nanoparticles is expected to
improve the density of bioreceptors at the surface.[1] Moreover the control of the structure of
nanoparticle assembly enables the tuning of the plasmon properties[2] in order to enhance the
sensitivity.

This platform will allow the study of the sensitivity of the SPR sensors, the limit of detection
and the recognition mechanisms between an analyte and a receptor molecule. Indeed, the SPR
measurements can be performed with microfluidic channel in order to investigate the kinetics of
biomolecular recognition process between bioreceptors and analytes.

In this chapter, the efficiency of our platform will be studied by the detection of two different
bio-molecules. The two couples of receptor/target molecules are:

126
 · Biotin/streptavidin
· Iminosugar/α-glycosidase

In one hand, the biotin/streptavidin was studied as a model system because this couple is very
well known. Streptavidin is easy to detect thanks to its high binding affinity Kd = 10-15 mol/L and its high
specificity. These molecules are widely used for detection systems since the mechanisms of association
are known and well described.[3, 4]

In another hand, the glycosidase was studied because it is involved in several essential
biological processes. Irreversible inhibitors such as iminosugar were used to covalently modify the
glycosidases which is involved in many biomedical processes (Gaucher disease, cystic fibrosis,
diabetes).[5, 6] The study of the immobilization of this enzyme at the surface of nanoparticles gives the
opportunity to bring a better understanding of the recognition process with iminosugars in addition to
studies reported from solutions.[7-10]

127
 D. Biotin-Streptavidine

The immobilization of proteins with maximum retention of activity and minimized nonspecific
interactions is a key goal in the development of biosensing technologies. Molecular recognition
processes of streptavidin with biotin supported onto modified surfaces have been importantly studied
as a model system for protein immobilization. The high binding affinity allows high stability and
robustness which make this couple very well used in many different applications in biotechnology. The
use of self-assembled monolayer (SAM) of organic molecules is highly suited for biomolecular
recognition because it allows the versatile grafting of molecular receptors onto solid surfaces. The
group of Knoll in the early 90s studied the adsorption of streptavidin onto a gold surface functionalized
with biotin terminated SAMs. They used the SPR technique to monitor the molecular recognition
process.[3, 4, 11] The structuration of the SAM was studied to understand the mechanisms of recognition.
These studies showed that the recognition mechanisms was dependent of the geometry and the
orientation of biotin groups.[11, 12] More recently, different approaches have been used to detect small
concentrations of streptavidin on surfaces.[13, 14] The biotin/streptavidin couple was further used to
reach limit of detection below nM.[12, 15]

1. Design of the bio-platform

The assembly of nanoparticles represents the base of the bio-platform because of the presence
of terminated azide groups at the nanoparticle surface. In order to graft irreversibly biomolecules
receptors, an alkynated derivative of biotin was synthesized. The molecular receptor was then grafted
on the surface of the nanoparticle film using the CuAAC “click” chemistry reaction which was used to
build nanoparticle assemblies.

a) Grafting of an alkyne derivative of biotin onto nanoparticle assembly

The biotin molecule displays a terminal carboxylic acid group which allows its easy
functionalization with an amine group. Therefore, alkynated biotin was synthesized from biotin and
propargyl-amine in DMF for 24 h at room temperature by performing the carboxyl-to-amine
crosslinking using the carbodiimide EDC and Sulfo-NHS (figure IV-2). See appendix A for experimental
details.

Figure 0-2. Schematic representation of the synthesis of alkynated biotin

128
 Fourier transform infrared spectroscopy was performed to confirm the synthesis of the
alkynated biotin (appendix C). The vibrational band localized at 1680 cm-1 is characteristic of stretching
mode of C=O bond in the amide group present in the biotin. The bands at 3300 cm-1 and 2115 cm-1
correspond respectively to the stretching modes of the C≡C-H and C≡C bands. The band at 3300 cm-1
is not visible because of the strong and wide O-H band ascribed to water. Moreover, 1H NMR display
signals which agree with the desired product (appendix C).

Alkynated biotin was grafted at the surface of azido-terminated nanoparticles supported on

the gold substrate by performing CuAAC “click” reaction in a similar way as reported in Chapter II. The
nanoparticle assembly was dipped for 24 h under reflux into a THF solution of alkynated molecules
which contains Cu-catalyst and trimethylamine. After the reaction, the substrate was rinsed with THF
and exposed for 1 minute to ultrasounds in order to remove any adsorbed species. The substrate was
then dried under air stream and packed for further characterization.

b) Characterizations

The grafting of biotin on the nanoparticle surface was controlled by using several
characterization techniques. The microscopy techniques give information on the spatial arrangement
of the nanoparticles and on the topography of the surface while spectroscopy techniques and water
contact angle measurement allow the chemical characterization of the surface.

(1) Microscopy characterizations

SEM images (figure IV-3) show the surface of the nanoparticle film before and after grafting
the biotin groups. The image becomes smoother after grafting the biotin, despite the same acquisition
parameters. It can be due to the presence of larger amount of organic species at the surface which
disrupts the detection. The contamination by organic species leads to a loss of the conduction of the
electron beam which induces a strong heating. Such an increase of temperature alters locally the
sample and results in blurry images.

a) b)

100 nm 100 nm

Figure 0-3. SEM images of the film of nanoparticles (a) before and (b) after grafting the biotin. Magnification X50000

129
 The surface coverage by nanoparticles can be evaluated before and after grafting the biotin
groups. The density of nanoparticles is estimated by counting the nanoparticles on twelve different
areas for different samples and averaged at 3385 ± 205 NPs/μm² and 3035 ± 215 NPs/μm² before and
after grafting respectively. Such a slight decrease about 10% can be attributed by the removal of
physisorbed nanoparticles.

36.6 nm 39.7 nm

a) b)

400nm 400nm

3.53.34nm
nm
3.53.45nm
nm

c) d)
d)

Figure 0-4. AFM images of the surface topography and profile cross-sections along the line of the surface before (a,c) and
after (b,d) grafting of biotin

AFM images show similar topographies of the surface before and after the grafting of the biotin
groups (figure IV-4). The topographic profiles (blue line) display similar average heights about 11.8 nm
and 11.4 nm before and after grafting, respectively. The average roughness slightly increases from 3.1
nm to 4.0 nm which can be related to the removal of nanoparticles.

SEM images do not show a remarkable change at the surface. The immobilization of biotin groups
does not significantly affect the structuration of the nanoparticle film which agrees with the formation
of covalent bonds between nanoparticles and the SAM onto gold substrates.

(2) Water Contact Angle

An easy and efficient way to determine the correct functionalization of the nanoparticle
surface was to measure water contact angle (WCA). The grafting of biotin groups induces a change in
the surface hydrophobicity. The measurement was done by depositing a 5 μL water droplet on the
substrate. Pictures of both samples were taken after one minute of stabilization. An alkyne chain
terminated with azide groups shows a water contact angle of 100° (figure IV-5.a) which is coherent

130
with a hydrophobic surface. The grafting of biotin derivatives is demonstrated by the decrease of the
contact angle down to 50° which agree with a hydrophilic surface (figure IV-5.b). It is consistent with
the replacement of alkyne groups by biotin.[16, 17]

a) 100° b) 50°

Figure 0-5. Pictures of a drop of water deposited onto a) the azido-terminated and b) the biotin-terminated nanoparticle
assemblies supported onto a gold substrate

(3) X-Ray Photoelectron Spectroscopy

X-Ray photoelectron spectroscopy (XPS) was performed in order to identify the chemical
elements at the surface of nanoparticle assembly. Figure IV-6.a shows the N1S binding energy region,
binding energies at 405.3 eV (N=N=N) and 401.8 eV (N=N=N) with a peak area ratio close to 1:2 was
attributed to azide groups.[10, 18, 19] After performing CuAAC, the signal at 405.3 eV totally disappeared
while a component a new signal at 400.5 eV was attributed to amine groups present in the biotin.[17]
The component at 400.8 eV was attributed to (N-C) bonds which agreed with the formation of the
triazole bridge.[20] The C1S region (figure IV-6.b) showed components at 285.3 eV and 286.9 eV
corresponding respectively to the C-H binding in the aliphatic carbon chains and the C-N in the triazole
bridge, respectively. The component at 288.7 eV can be attributed to the C=O bond in the biotin
groups.[21] These results confirmed the total replacement of azide groups on the nanoparticle surface
by biotin groups.

131
 a) b)

c) d)

Figure 0-6. XPS measurements onte a,c) the azido-terminated and b,d) the biotin-terminated nanoparticle assembly
supported onto a gold substrate. N1S (a,c) and C1S (b,d) core level

(4) Phase Modulation Infrared Reflection Absorption

Spectroscopy measurements

Phase Modulation Infrared Reflection Absorption Spectroscopy (PM-IRRAS) was performed to

measure low amount of organic compounds such as a monolayer of biotin groups at the nanoparticle
surface. Figure IV-7.a shows the spectra recorded before (red curve) and after (blue curve)
functionalization with alkynated biotin. Given the SEM and AFM results, the quantity of phosphonic
acid groups localized at the nanoparticle surface is expected to be unchanged after the
functionalization step. Therefore, both spectra were normalized to the area of the ν(Fe-O-P) at 1082
cm-1.

132
 a)

b)

Figure 0-7. a) PM-IRRAS spectra recorded before (red) and after (blue) biotin functionalization. b) FT-IR spectrum of the
derivative alkynated biotin molecule

The disappearance of the azide band ν(N3) at 2100 cm-1 and the presence of a supplementary
contribution of the carboxylic acid band ν(C=O) at 1660 cm-1,[22] agree with the reaction of the azide
terminal group with the alkynated biotin derivative. The presence of biotin groups was also correlated
to the broad band (1750-1550 cm-1) corresponding to amide I and II (ν(C=O) and ν(N-H)) vibrations as
observed for the alkynated biotin molecule (figure IV-7.b).

To summarize, the large panel of characterization techniques we have used, shows the
replacement of azide groups by biotin groups at the surface of nanoparticle assembly. Furthermore,
our building strategy based on the CuAAC reaction allows preserving the structure of nanoparticle
assemblies as a monolayer of nanoparticles.

133
 c) Monitoring of the bio-platform build-up by surface plasmon resonance
measurements

The SPR signal of the gold substrate being highly sensitive to the variation of the local refractive
index at its surface (see chapter I for details), the SPR measurements were performed to monitor the
construction of the film.

The film construction includes several steps which increase the refractive index at the surface
of the sample:

· Gold naked substrate (Au)

· Gold substrate functionalized with alkyne-SAM (SAM-CC)
· Assembly of nanoparticles functionalized with azide groups onto the SAM (NP@N3)
· Assembly of nanoparticles functionalized with alkynated biotin (Biotin)

The SPR measurements were performed at two different wavelengths by using Bionavis©
apparatus. The reflectivity measured as function of the incident angle at 670 nm and 785 nm is shown
in figure IV-8.

Figure 0-8. SPR reflectivity measurements versus angular response θ with an excitation wavelength of a) 670 nm and b)
785 nm

As explained in chapter I, the angular position of the resonance peak is dependent of the
wavelength at which the measurement is performed. As mentioned before, for the shortest
wavelength, the position of the resonance peak and the sensitivity are overestimated due to the
largest imaginary part of the metal permittivity which is correlated to the absorption of light by the
gold thin film. It results in the enlargement of the resonance peaks recorded at 670 nm.[23] The
theoretical positions of the peaks are determined by the resonance condition equation:

& sin(C) = n
,’ ,U
,’ r,U
Equation VI.1

If we consider the refractive index of the water nwater = +εÔÕÖ×Ø = 1.33, the prism refractive index and
the metal permittivity given for the working wavelength,[24] we can calculate the theoretical value of
the resonance angle. These calculations are made with the approximation that the gold film is thick

134
enough to not disrupt the peak position. This calculated value can then be compared to the
experimental value of the resonance angle. The measurements performed with the Bionavis©
aparatus allow an error of 0.001° for the experimental values. These values are summarized in the
table IV-1 for a naked gold substrate if we consider water as the dielectric medium.
Table 0-1. Values of theoretical and experimental resonance angle for naked gold film (Au)

670 nm 785 nm
Prism refractive index
1.520 1.516
np
Metal permittivity5
-14.358 -22.855
εm
Theoretical resonance angle
69.574° 66.326°
θth
Experimental resonance angle
68.627° 65.598°
θexp

The experimental and theoretical values of the resonance angle are close, the deviation
normalized to the experimental value measured at 670 nm is about 1.4% whereas at 785 nm, it is
below 1.1%.

These values were calculated by considering that the dielectric at the gold surface is water
occupying the whole sensing volume. In our case, this model becomes more complicated because the
refractive index combines organic and inorganic parts with different ratios after each step. Therefore,
a model with multilayers of different permittivities has to be taken into account. Nevertheless, the
position of the resonance angle can be determined experimentally. The values of the minima for each
step of the platform build-up are summarized in the table IV-2.
Table 0-2. Angular peak positions measured for each step of the platform build-up for both operating wavelengths

Wavelength Au SAM NPs@N3 Biotin

670 nm 68.627° 69.260° 74.029° 72.845°
785 nm 65.598° 65.960° 68.410° 67.832°

After each step, the grafting of an organic or inorganic element increases the refractive index

in the largest shift of the resonance angle because of the largest change of refractive index (nÙØÚÕSÛÜ =
at the gold surface which results in the increase of the resonance angle. Iron oxide nanoparticles result

1.42 and nÞ×f ßà = 2.36 at 670 nm) and the largest quantity deposited at the gold surface.

However, the grafting of the biotin derivative on the nanoparticle surface resulted in a negative
shift of the resonance angle (-1.184° for 670 nm and -0.578° for 785 nm). This change is obviously
correlated to the decrease of the refractive index at the gold surface. In order to verify the possibility
of the material removal, the film of nanoparticles was refluxed in the same condition of the “click”
reaction for 24 hours without addition of biotin derivative. The film before and after refluxing was
characterized in order to see the potential desorption of nanoparticles at the surface.

5
According to the value measured experimentally from Johnson et al.[21]

135
 SEM and AFM measurements did not show significant changes of the nanoparticle density and
the topography as observed in the presence of biotin (IV.A.1.b). However, these microscopy analyses
are not accurate to determine precisely such a small variation of nanoparticle density. The SPR
measurements are more sensitive of any change at the surface. Therefore, the film of nanoparticles
was characterized by SPR measurement before and after refluxing in the condition of click reaction
without adding alkynated biotin in the medium.

Figure 0-9. SPR measurements performed on an azido-terminated nanoparticle assembly supported onto a gold
substrate before (solid curve) and after (dotted curve) being exposed to CuAAC reaction conditions without alkynated
biotin derivative.

Figure IV-9 shows the resonance angle recorded for the nanoparticle assembly before and after
refluxing for 24 hours. For each wavelength, the resonance angle decreases (-0.226° and -0.145° for
670 nm and 785 nm respectively). It is lower than for the sample with biotin; however, this change can
be attributed to the difference of density in nanoparticles and the presence of more or less aggregates
at the surface which can induce different removal of the nanoparticles. Nevertheless, these results
proved the change of refractive index at the surface which can be explained by the desorption of the
nanoparticles on the surface during the refluxing process. This desorption can be caused by the
removal of physisorbed nanoparticles although the sample was exposed to ultrasounds after
nanoparticles were assembled.

d) Performing SPR measurements to determine nanoparticle density

Here we focus on a new way to measure nanoparticle density onto which is usually measured
by SEM. Although SEM provides directly information on the spatial arrangement of nanoparticles, it
may by time consuming in the case of a large number of samples to study. Furthermore, as we showed
above, it may not be accurate to detect small variations in nanoparticle density on large surface areas.
Therefore, we decided to use SPR measurement to measure nanoparticle density

136
 Knowing the refractive index of iron oxide nanoparticles for a given wavelength, the position
of the resonance peak allows determining the density of nanoparticles. Samples with different
densities of nanoparticles (with a mean diameter of 10.4 nm) were prepared by performing “click”
reaction for different times. The position of the SPR peak measured at 785 nm and the nanoparticle
density measured from SEM micrographs were plotted as function the reaction time (figure IV-10).

Figure 0-10. Density and SPR angle position for samples prepared after different CuAAC reaction times

Both curves are rather similar and show two different regions which correspond to the fast
assembly of nanoparticles for reaction times below 3 hours which levels off but continue to increase
until 48 hours. The slower kinetic of the assembly reaction after 3 hours results from the reduction of
the available surface by 60 %. We can see that SPR provides a much better precision on nanoparticle
density than SEM in the case of the fast kinetics. The larger value of the SPR peak position after 48
hours in comparison with density can be attributed to aggregates on the surface which make the
counting of nanoparticles difficult.

The SPR peak can then be plotted as function of the density of these samples and confronted
to a theoretical model (detailed in appendix A) shown on figure IV-11.

137
 Figure 0-11. SPR peak position at 785 nm plotted with density of nanoparticles

The experimental points show a linear dependence between the SPR and the density of
nanoparticles at the gold thin film surface. The values measured after 48 hours of reaction are not
shown because of the presence of aggregates of nanoparticles at the surface. This linear fit shows a
direct correlation which allows us determining the nanoparticle density by knowing the SPR peak
position. Moreover, the simulated data for the considered size (10.4 nm, blue dotted curve) shows a
similar behavior of the experimental points. However, the simulated curve exhibits an offset of few
millidegres. This offset may result from the size distribution of nanoparticles which can generate a
distribution of angle. Thus, simulated data were performed with different values of D between 10 nm
and 12 nm and can represent a fictive size distribution (colored area). Moreover, this simulation for
several sizes shows that the evolution is not perfectly linear. Another reason of the offset may be due
to the approximation of the model where the nanoparticle assembly constitutes a layer with an
effective dielectric constant. This dielectric constant is determined by using the effective thickness of
the nanoparticle layer and therefore using the filling factor f. The filling factor is defined by the volume
fraction of nanoparticles:

â=
ãäåj
ãæ
Equation VI.2

with VNPs the volume occupied by the nanoparticles (determined from the density and the
volume of a nanoparticle) and VT the volume total with a surface of 1 μm² and a height corresponding
to the penetration depth of the evanescent wave δ (see chapter I). This approximation implies to
consider a total volume with a homogenous permittivity. However, the nanoparticles are not spread
all over the sensing volume but localized at the vicinity of the metal surface which influences the value
of θ.[25] Nevertheless, this calibration curve allows determining the density of nanoparticles onto the
gold thin film by knowing the SPR peak position.

In this part, a nanostructured recognition platform has been elaborated and the different
characterization techniques demonstrated unambiguously the grafting of biotin derivatives at the
nanoparticle surface. Moreover, SPR was demonstrated to be very efficient to monitor the build-up of
our detection platform. Furthermore, the high sensitivity of SPR properties to change of refractive

138
index can be used as an alternative to microscopy techniques to quantify precisely nanoparticles
assembled onto a gold thin film.

2. Detection of streptavidin

In order to prove the efficiency of our sensor, we studied the adsorption of streptavidin by
taking advantage of the high tunability of our detection platform. Several structural parameters
(nanoparticle density, nanoparticle size, amount of bioreceptors) were investigated to build the most
efficient detection platform. We used two different techniques to measure the SPR properties of the
gold film which are the most reported in the literature: the angular interrogation (at fixed wavelength
of 785 nm) and the spectral interrogation (at fixed angle of 74°).

a) Influence of nanoparticle assembly on the detection

The detection will be studied with two different SPR systems presenting the same
Kretschmann configuration to couple incident light with surface plasmons at the interface between
metallic substrate and dielectric (see chapter I). The incident wave goes through the prism made of
BK7 glass and hit the gold substrate. The minimum of reflection on the surface can be measured as
function of the incident wavelength or the incident angle. Therefore, the two systems are defined and
used as function of their response:

· A home-made SPR experiment allowing different configurations which was used to study
the spectral interrogation to get information on the incident resonance wavelength. In this
case, the substrate was deposited directly onto the prism and to ensure the refractive
index continuity between the prism and the sample, a coupling oil (n = 1.51) is deposited.
A gold thin film is deposited by sputtering onto a glass plate with an anchoring layer of
titanium (2-3 nm). The thickness of the gold film used is 30 nm or 50 nm.6
· A Bionavis© commercial system working at a fixed wavelength allowed studying the
angular interrogation by changing the resonance angle position.7 Here, there was no
control on the prism, commercial substrates were inserted inside the system after building
up of the detection platform. This system was combined to a microfluidic device which was
also used to study the kinetics of the recognition process. Commercial substrates consist
in a gold layer of 50 nm deposited onto a glass plate with an adhesion layer of chromium
about 2-3 nm.

In the following measurements, streptavidin was extracted from the Streptomyces avidinii
bacteria. A buffer solution is commonly used for the protein stabilization and to avoid denaturation or
conformation change during the storage. Here, the streptavidin was used in pure water to avoid
unspecific adsorption of salt on the surface which may hamper the recognition process. The

6
Collaboration with J-F. Bryche,
G. Barbillon C2N, Orsay
J. Moreau, IOGS, Orsay
7
Collaboration with F. Boulmedais, ICS, Strasbourg

139
recognition process will be performed on biotin terminated assemblies of 10 nm-sized nanoparticles
with different densities.

(1) Detection of streptavidin by performing angular interrogation

Angular interrogation was performed by using the Bionavis© system. The sample was settled
in the SPR device and the streptavidin solution was injected by using a microfluidic channel. The
incubation process was performed for a given time and concentration of streptavidin. Preliminary
studies showed that the angular shift reached its maximum value only few minutes after injection of a
streptavidin solution with a concentration of 100 μg/mL (see part VI.A.4.1). 250 μL of streptavidin
solution were injected with a flow rate of 50 μL/min. The sample was rinsed further by injecting pure
water.

The effect of the nanoparticles on the adsorption of the streptavidin was studied by measuring
the SPR signal at 785 nm (figure VI-12). Two samples were studied: i) a gold thin film without
nanoparticles was prepared by grafting azide-PEG3-biotin onto an alkyne terminated SAM by
performing CuAAC “click” reaction (appendix C for characterizations) ii) a gold thin film fully covered
by 10 nm-sized nanoparticles decorated with biotin groups at their surface was prepared as mentioned
in the previous section of this chapter.8

Figure 0-12. SPR measurements at 785 nm before (red curve) and after (green curve) adsorption of streptavidin on a) a
gold thin film b) a gold thin film covered by 10 nm nanoparticles.

Figure IV-11 shows the adsorption of streptavidin onto a gold thin film with and without
nanoparticles. As expected, the injection of streptavidin onto both biotinylated surfaces induces a shift
of the resonance angle corresponding to the adsorption of streptavidin on the surface. The main
difference between both samples is the larger SPR angle shift observed in the presence of
nanoparticles. The shift passes from 0.303 ± 0.002° to 0.582 ± 0.002°, meaning an increase about twice
by adding the nanoparticles on the surface. The broadening of the peak for the surface covered by
nanoparticles can be explained by the heterogeneity of the film caused by the nanostructures. The

8
The measurements were performed at both wavelengths. We chose to study the 785 nm since the resonance
peak is sharper. Moreover, the position at lower angle is more convenient for the grafting of nanoparticles.

140
small change in the minimum reflectivity is also probably due to the absorption of the light by the iron
oxide.[26] Furthermore, we can notice that the specific detection of streptavidin by using biotin receptors
was not affected by the presence of iron oxide nanoparticles. The adsorption is followed by a rinsing step with water. The
streptavidin was not desorbed (see part VI.A.4.1) which demonstrates that our sensing platform preserves strong biomolecular interactions.

(2) Detection of streptavidin by performing spectral interrogation

The same experiment was conducted by using a different SPR system which is based on
spectral interrogation. The fixed angle was settled at 74° which was the optimized position to monitor
the adsorption of streptavidin. Here, the streptavidin was not injected with a microfluidic channel. The
substrate was immersed in the streptavidin solution at 100 μg/mL for 30 minutes to conduct
measurements in “static” mode. The substrates were then rinsed with pure water and immersed 10
seconds in an ultrasonic bath in order to avoid non-specific deposition of the streptavidin on the
nanoparticle surface. The SPR measurements before and after streptavidin adsorption were
performed for different surface coverages with nanoparticles. Three samples were considered; i)
without nanoparticles (0%), ii) an intermediate coverage of the gold thin film which mainly consists in
isolated nanoparticles (15%) and iii) a fully covered gold thin film by nanoparticles (100%)(figure IV-
13). The intermediate coverage of nanoparticles was controlled by adjusting the reaction time of
CuAAC (cf. chapter II).

a) b)

100 nm 100 nm

Figure 0-13. SEM images showing the coverage of the gold substrate by nanoparticles corresponding to a) 15% and b)
100 % of a full monolayer after performing CuAAC "click" reaction. Scale bar: 100 nm

The SPR measurements are shown on figure IV-14 the curves are corrected because the
minimum of reflectivity changes with the presence of nanoparticles as observed for the angular
measurements. This loss in intensity was observed in the case of deposited material on a gold thin
film[27, 28] and is caused by the absorption of the iron oxide nanoparticles. The sample fully covered
with nanoparticles exhibits the largest shift of the resonance peak up to 41 nm. In contrast, a lower
density in nanoparticles corresponding to 15 % coverage of the gold substrate resulted in an
intermediate shift of 7 nm. The sample without any nanoparticles at the surface presents the smallest
shift of 3 nm after immersion in the streptavidin solution. The broadening of the resonance peak in
this case is less pronounced than for the measure with angular response. However, the same behavior

141
is observed since the presence of the nanoparticles on the surface give an enhancement of the
response after exposition to the protein.

Figure 0-14. SPR measurements at 74° (internal angle) performed before (red curve) and after (green curve) exposition to
streptavidin for different surface coverages of the gold substrate by nanoparticles

Both angular and spectral interrogations allow us concluding that the spectral shift is enhanced
with the presence of nanoparticles on gold thin film, and that the largest density of nanoparticles is
the most effective.

b) Understanding the role of nanoparticle assembly on the detection

The response of the sensor is dependent of the intrinsic optical properties of the gold thin film
and of the chemical functionalization of the nanoparticle surface. To understand the influence of the
nanoparticles on the sensitivity of the system and its ability to detect streptavidin, we studied all
parameters given by equation (IV.3) on which the response R (angular or spectral shift) is dependent
(cf. chapter I):[29]
ªc’
¤ = ¥. ç&. [1 − G —’
] Equation IV.3

The enhancement of the SPR response can result from:

· The decrease of the decay length ld;

· The increase of the sensitivity factor m;
· The increase of surface area;
· The increase of the accessibility of functional groups;

142
The presence of nanoparticles impacts these factors which explained the increase of the shift. These
different parameters are studied in order to understand their influence on the sensors sensitivity.

(a) Decay length

The evanescent electromagnetic field decay length (which is half of the penetration depth) is
dependent on two parameters: the operating wavelength and the refractive index of the dielectric at
the surface of the gold thin film (equation IV.4).

¬: = 2 = n¡ ¡
è ,U r,’
‡Ã ,’ ²
Equation IV.4

The angular interrogation being performed at a fixed wavelength, the decay length is expected
to change only because of the increase of refractive index. Nevertheless, it is weakly affected (variation
of 7 %) as shown by measurements performed with or without nanoparticles (300 nm and 322 nm,
respectively).

In contrast, in the case of the spectral interrogation, the decay length is expected to be
significantly increased at longer wavelengths. The decay length increased almost two times from
without nanoparticles (160 nm) to with nanoparticles (302 nm).

Therefore, the decay length is more influenced by the operating wavelength than by the
refractive index. Indeed, a larger penetration depth increases the sensing volume which hampers the
contribution of the refractive index. Therefore, we would have expected that the presence of
nanoparticles would disfavor the SPR shift with a strong increase of ld. However, our experimental
results show that the effect of the decay length on the SPR shift is minor and the significant increase
of ld is compensated by others parameters.

(b) Sensitivity factor

(i) Angular interrogation

The sensitivity factor m is calculated from calibration curves which were recorded by
measuring the shift position of the SPR peak as function of the refractive index (RI) of different
surrounding media.[30, 31] m can then be extracted from the slope of these curves. Calibration curves
have been recorded with the Bionavis system for each sample with and without nanoparticles (figure
IV-15). The shift of the resonance peak was measured for several aqueous solutions of sucrose at
different concentrations which were correlated to different refractive index.

143
Figure 0-15. Calibration curves recorded at 785 nm for a gold thin film uncovered (red curve) and covered (blue curve) by
a full monolayer of nanoparticles

The calibration curves present different slopes without (105.4 ± 2.1°/RIU) and with
nanoparticles (115.7 ± 3.3°/RIU). This important result shows that the sensitivity of the surface
plasmon is slightly increased by the presence of nanoparticles (15%). It may result from the weak
interaction of nanoparticles with the light at 785 nm.

The sensitivity is measured for different densities of nanoparticles at the gold surface in order
to determine the sensitivity as function of the resonance angle. The shift of the resonance angle is
measured after the injection of aqueous solution of sucrose at 0.1 mol/L (meaning a variation of
refractive index of 5.10-5 RIU). Therefore, the sensitivity can be deducted with m = Δθ/Δn and
represented on figure IV-16.

144
 Figure 0-16. Sensitivity factor as function of the incident angle measured on a gold film with a 50 nm thickness
for different nanoparticles densities

The slight variation of the sensitivity is shown with the linear fit of the data for different
nanoparticle densities. The low increase of the sensitivity shows that the influence of the peak
resonance is weakly important on the global sensitivity of the sensor. Nevertheless, it shows that the
nanoparticles have a slight influence on this factor. This behavior is similar to the one of a multilayer
system where the over layer on the gold surface increases the sensitivity as shown in the literature.[23,
32]

(ii) Spectral interrogation

In the case of the spectral interrogation, m was calculated on film of 30 nm and 50 nm of gold
thickness for different wavelengths by changing the refractive index with aqueous solution of sucrose.
A polynomial adjustment was done to calculate the sensitivity for every wavelengths (figure IV-17).9

Figure 0-17. Sensitivity factor as function of the wavelength measured on a gold film with a 30 nm (red curve) and 50 nm
(blue curve) thicknesses

The resonance angle has slight influence on m, but here, only the wavelength can have an
impact on the sensitivity. Therefore, the sensitivity can be calculated for different wavelength which
will correspond to different densities of nanoparticles. In our study, m dramatically increased with
nanoparticle density and was directly correlated to the red shift of the SPR peak from 611 to 760 nm
which was reported for similar systems.[16] m is the largest for the gold film fully covered by iron oxide
nanoparticles (5600 nm/RIU) in comparison with the sample with nanoparticle density representing
15 % of the theoretical maximal coverage and without nanoparticles (2020 nm/RIU and 1580 nm
respectively).

9
Collaboration with Julien Moreau, IOGS

145
 These experiments have shown that the sensitivity factor m of the sensors is slightly dependent
of the position of the resonance angle (correlated to the deposition of nanoparticles) and strongly
dependent of the incident wavelength. Indeed the nanoparticles do not influence this factor directly
but indirectly in the case of the spectral interrogation by displacing the resonance peak. For the
spectral interrogation system we can easily see the difference of the sensitivity, but in the case of the
angular interrogation where the wavelength has no influence, others parameters should influence the
larger shift with nanoparticles.

(c) Available surface area

Besides the intrinsic parameters of the gold thin film, the sensitivity can be increased by the
surface functionalization.[10, 33] Indeed, the modulation of surface area and topography resulting from
nanoparticles may markedly increase the amount and the accessibility of biotin groups. Investigation
of the theoretical quantity of adsorbed streptavidin will be compared with the experimental
determination of thickness and mass of streptavidin.

(i) Number of streptavidin adsorbed

The surface area is increased with the presence of the nanoparticles due to their high
surface/volume ratio. The calculation of the number of available biotin groups can be estimated
knowing the occupied surface by alkane thiol molecules on a gold thin film (24 Ų) and by the alkane
phosphonate molecules on the iron oxide surface (26 Ų). The number of biotin groups was calculated
on a gold thin film fully covered by 10 nm-sized nanoparticles and was compared with a gold thin film
without nanoparticles. The number of biotin groups was calculated on the assumption that all azide
groups reacted with alkynated biotin derivatives as demonstrated at the beginning of this chapter.
Table 0-3. Biotin groups per surface unit onto a thin gold film and on a thin gold film fully covered by a 10 nm-sized
nanoparticles

Flat surface 100% NPs

Surface available per μm² 1,0.10-12 μm² 1,86.10-12 μm²
Number of biotin groups
4,17.106 3,10.106
available

Table IV-3 showed that the number of biotin groups per surface unit decreases by 25 % which is
unexpected. Furthermore, nanoparticles being grafted onto the gold thin film, we estimated that biotin
groups were grafted on only half of their surface.

146
 Nanoparticle

Streptavidin 4,5 nm
Organic layer
Biotin area available
for recognition
Conical volume of the binding sites
available for streptavidin

r
α

X nm

Figure 0-18. Schematic representation of the available area of biotin for the adsorption of streptavidin

Moreover, the number of “available” biotin groups should be considered with the size of the
streptavidin which will induce steric hindrance at the surface. Indeed, when the nanoparticles are
grafted on the surface, a part of the azide groups already reacted with the SAM and others are not
accessible because of the tight packing of the nanoparticles. If we consider the surface available for
the streptavidin in the case of a dense monolayer of nanoparticles, only few streptavidin can be
adsorbed at the nanoparticle surface. Indeed, the steric hindrance of the streptavidin on the surface
of the nanoparticles does not allow its adsorption on the total surface of nanoparticles when tightly
packed. The figure VI-18 shows the geometrically area where the available biotin groups are localized
for the adsorption of the streptavidin.

The red area corresponds to the surface where are localized the available biotin groups. With
this representation, it is obvious that the spacing between nanoparticles can increase the red area and
therefore allow the adsorption of a larger amount of streptavidin. The figure IV-19 presents the
theoretically maximum number of streptavidin per nanoparticle which can be adsorbed onto the
nanoparticle assembly as function of the inter-particle distance considering a streptavidin surface area
of 20.25 nm²[34] and 10 nm-sized nanoparticles.

147
Figure 0-19. Number of streptavidin which can be adsorbed onto a nanoparticle as function of the inter-particle distance
in the case of 10 nm-sized nanoparticles

As we can see the number of streptavidin per nanoparticle for a close packed nanoparticle
assemblies (and therefore considering an inter-particle distance null) is about 4
streptavidin/nanoparticle. This value agree with the surface available of each nanoparticle for
streptavidin adsorption (appendix A) and the surface occupied by each streptavidin.[15] In our case, a
surface covered with 5 800 NPs/μm² corresponds to a full coverage of 10 nm-sized nanoparticles with
an organic shell of 2 nm. Therefore, the maximal theoretical number of streptavidin grafted should be
23 200 streptavidin/μm². In the case of a coverage of 15 % by nanoparticles, the average inter-particle
distance is about 15 nm which is more than the size of the streptavidin. Therefore, we can consider
that neighboring nanoparticles will do hamper the grafting of the streptavidin and reach the value of
15 streptavidin/nanoparticle. However, the surface is only covered with 15 % in density which
corresponds to 870 nanoparticles/μm² and leads to a theoretical detection of 13050 streptavidin/μm².

As we can see, the number of streptavidin detected does not increase linearly with the quantity
of nanoparticles at the surface. Therefore, it should exists an optimum density of nanoparticle which
can favor the adsorption of a maximum of streptavidin (see part IV.A.3).

(ii) Determination of thickness and mass of

streptavidin adsorbed

Experimentally, the effective thickness of the adsorbate layer can be determined with the
equation of the sensor response (equation IV.3). Indeed, knowing the sensitivity factor and the decay
length, we can determine the effective thickness of the streptavidin which was adsorbed as function
of the measured response:

d= ln(1 − )
œ]’ * £
_ s ∆y
Equation IV.5

with Δn = (nadsorbate - nsolvent) = 0.24 for the streptavidin.[35, 36]

The spectral shift observed without nanoparticles gives a streptavidin thickness of 0.8 nm.
Considering the size of a streptavidin (4.5x4.5 nm²)[34] the surface covered by the protein is less than

148
20%. This low value can be explained by low accessibility of biotin groups which are tightly packed at
the surface of gold thin film. Indeed, steric hindrance hampers the recognition of the streptavidin.[37]
However, this value is very low and may have to be reconsidered. In contrast, the angular shift
measured for the same sample gives a thickness of 1.9 nm which is higher but still low in comparison
of the size streptavidin. In the case of the sample fully covered with nanoparticles, the streptavidin
thicknesses are 4.6 nm and 3.2 nm for the system with spectral and angular interrogation, respectively.
The thickness in the case of the spectral shift is surprisingly high and may represent a full layer of
streptavidin at the surface of the nanoparticles.

The effective thickness can be correlated to the mass of protein per unit area which can be
determined by using the Freiter equation[38] which gives the relation between the response and the
mass of adsorbed analyte:

ë = î†
ì†∗p
ïî»
Equation IV.6

ð&ï ð&
with Δn the variation of the refractive index at the surface, d the thickness of the adsorbate layer and
ð) the variation of the refractive index as function of the protein concentration ( ïð) =
0.212 g. cmœb for the streptavidin).[39]

The surface coverage is determined for the samples with and without nanoparticles (table IV-4).

Table 0-4. Summarize of surface coverage and number of streptavidin per surface area determined by the angular shift

Δθ Δλ Γ Streptavidin/μm²
0.304° 215.1 ng/cm² 24 500
Without NPs
3 90.6 ng/cm² 10 400
0.512° 362.3 ng/cm² 41 300
With NPs
41 520.8 ng/cm² 59 300

The values of the mass led to the same conclusion than for the thickness. The system with
angular interrogation shows an increase of almost twice with the presence of nanoparticles and the
system with spectral interrogation exhibit six times more streptavidin with the nanoparticles.

Moreover, the mass coverage allows determining the number of streptavidin per micrometer
square. Knowing the number of streptavidin per nanoparticle (4.3 streptavidin/nanoparticles)
determined by theoretically and corresponding to a density of nanoparticles (5800 NPs/μm²), we can
also determine the number of streptavidin per surface area. Nevertheless, the theoretical maximum
coverage is approximated to 25 000 streptavidin/μm². This value represents only half of the
streptavidin determined experimentally and may put questions on the reliability of the measurements.
Complementary characterization techniques such as quartz microbalance would help concluding on
the quantity of streptavidin at the surface.

The table IV-5 summarizes the different parameters for the both systems.

149
 Table 0-5. Summarize of the characteristics measured from angular and spectral interrogations

785 nm 785 nm 74° 74° 74°

Sans NPs Avec NPs Sans NPs 15 % NPs 100 % NPs
Angular response of
0.304° 0.582° / / /
the sensor Δθ
Spectral response of
/ / 3 nm 7 nm 41 nm
the sensor Δλ
Decay length
322 nm 300 nm 160 nm 186 nm 302 nm
ld
Sensitivity factor
105.4°/RIU 115.6°/RIU 1 300nm/RIU 2 020 nm/RIU 5 600 nm/RIU
m
Effective thickness d 1.9 nm 3.2 nm 0.8 nm 1.4 nm 4.6 nm

Mass coverage Γ 215.1 ng/cm² 362.3 ng/cm² 90.6 ng/cm² 158.5 ng/cm² 520.8 ng/cm²

In conclusion, the presence of nanoparticles significantly influences on the intrinsic optical

properties of the gold thin film and contributes actively to enhance the sensitivity to change in
refractive index resulting from the adsorption of streptavidin. The use of two different configurations
allowed us to distinguish the contribution of the parameters on the sensitivity. The system with
spectral interrogation showed that the presence of nanoparticles increases the penetration depth
which should decrease the response of the sensor. However, this effect is minor in comparison with
the large enhancement of the sensitivity factor. Nevertheless, these effects are mainly due to the
operating wavelength. The use of angular interrogation with a fixed wavelength allows showing the
effect of the surface functionalization without modifying significantly the penetration depth and
sensitivity factor. The presence of nanoparticles does not increase the number of biotin groups
available for the recognition of the streptavidin. However, the accessibility is increased since almost
twice streptavidin molecules are detected according to the increase of the thickness and mass of
protein.

3. Detection platform optimization

The optimization of our detection platform is related to the structure of the nanoparticle
assembly. Moreover, it has been shown that the immobilization of the streptavidin onto biotin surface
is affected by the concentration of the biotin groups and their availability at the surface.[37, 40-42] The
accessibility of biotin groups which is directly dependent on their packing may be enhanced by dilution
at the nanoparticle surface.

a) Influence of the spatial distribution of biotin groups at the nanoparticle

surface

150
 (1) Preliminary study: influence of the accessibility of the biotin
groups on a gold surface

To assess the accessibility of biotin groups on a gold thin film, they were diluted with alkane
chains. This control experiment is first performed on a gold thin film without nanoparticles. First, a
mixed SAM was prepared by immersing a gold substrate in an equimolar ethanolic solution of (11-
Undec-1-ynyl)thiol and dodecanethiol in a molar ratio 50/50 with a total concentration of 10 mM.
Second, the alkynated biotin derivative was grafted in the standard conditions.

SPR measurements were performed by injecting an aqueous solution of streptavidin (100

μg/mL) in standard conditions (figure IV-20).

Figure 0-20. SPR measurements of the gold thin film functionalized with 100% and 50% of biotin groups without
nanoparticles

The initial position of the resonance angle is similar for both samples (66.128° and 66.175° for
the 50% and 100% density of biotin groups, respectively). The slightly higher value for the sample with
100% biotin may result from the thicker organic layer implying a higher refractive index. More
interestingly, after exposition to streptavidin solution, the shift is significantly larger for the sample
with 50% biotin (Δθ50% = 0.477°) than with 100% biotin (Δθ100% = 0.349°). Although the density of biotin
groups is half, its dilution significantly enhances their accessibility to streptavidin. The close packed
biotin groups do not allow the good recognition of the streptavidin in contrast with separated biotin
groups favor a better recognition.

In order to study the influence of the density of biotin groups on nanoparticles, several samples
with different amounts of biotin at the nanoparticle surface have been performed.

The method consists in grafting a mixture of alkynated biotin and hex-1-yne molecules at the
surface of nanoparticle assemblies in order to prepare several samples with the same density of
nanoparticles. Hex-1-yne was chosen because it is inactive and as a similar length to alkynated biotin
derivative. Therefore, biotin groups were diluted with methylene groups at the surface of nanoparticle.
The biotin/hex-1-yne molar ratios studied were 5%, 20%, 50%, 80% and 100%. The nanoparticles

151
 assembled were characterized by SEM and PM-IRRAS in order to control the nanoparticle density and
the CH3/biotin ratio at their surface.

(a) SEM analysis

The SEM images show that the spatial arrangement of nanoparticle assemblies is very similar
whatever biotin/hex-1-yne ratio (figure IV-21). The densities of nanoparticles are the same for each
sample (with an average of 44.8 ± 3.8 % of the theoretical full density). The nanoparticles on the surface
are not fully packed, nevertheless the same density over the sample ensure the same position of the
resonance peak. Moreover, this result shows the high reproducibility of “click” chemistry approach for
nanoparticle assembly and their functionalization by biomolecules.

a) 45,3 ± 2,8 % b) 43,3 ± 3,4 % c) 45,5 ± 3,8 %

d) 43,0 ± 2,8 % e) 47,1 ± 1,7 %

Figure 0-21. SEM images of the sample with different ratio biotin/hex-1-yne a) 5%, b) 20%, c) 50%, d) 80% and e) 100% (inset: density of
nanoparticles)

(b) PM-IRRAS measurements

PM-IRRAS measurements were performed in order to control the functionalization at the

surface. The figure IV-22 showed the results on the surface functionalized with the different
biotin/hex-1-yne ratios.

152
 Figure 0-22. PM-IRRAS measurements of nanoparticle assemblies functionalized with different biotin/hex-1-yne ratios

Unfortunately, interferences during the measurement did not allow recording vibrational
modes below 1300 cm-1 which corresponds to the phosphonate area. The numbers of CH2 groups being
similar in the biotin and the hex-1-yne, the spectra were then normalized on the area of the ν(C-H)
band. Bands around 2300 - 2400 cm-1 were ascribed to ν(CO2) from the environment during the
measurement. The large bands in the 1500 - 1700 cm-1 correspond to the amide bands of the biotin
groups as mentioned above (cf. IV.A.1.b.4)). Moreover, the presence of the phosphonic acid band at
1600 cm-1 contributes to the enlargement of this band. Therefore, the quantitative determination of
amide groups is complicated. However, the relative evolution of these bands with respect to the
biotin/hex-1-yne ratios shows the increase of biotin groups. The figure IV-23 shows the evolution of
the surface area of the ν(C=O) band expected at 1680 cm-1 with respect to the ν(C-H) band at 2950 cm-
1
.

Figure 0-23. Evolution of the area surface ratio between ν(c=o) and ν(C-H) bands as function of the biotin/hex-1-yne ratio

153
 The increase of ν(C=O)/ν(C-H) ratio with the biotin/hex-1-yne ratio in the reaction medium shows
the linear increase of biotin at the surface of the nanoparticle assembly (excepted for 80%).

(c) SPR measurements

The SPR measurements were performed as function of the incident angle on the five samples
with different ratios of biotin. The streptavidin was adsorbed by dipping the samples in an aqueous
solution of streptavidin at a concentration of 100 μg/mL. The results are presented on figure IV-24.

Figure 0-24. SPR measurements before (red) and after (green) exposition of the samples containing
different biotin/hex-1-yne ratios in a streptavidin solution

Before adsorption of streptavidin the SPR peak is centered to 670 nm which agree with a
density about 45%. Some slight differences on the sample at 50% and 80% can be observed where the
peak position is slightly higher. These differences can be caused by the presence of aggregates on the
nanoparticle layer.

The shift induced by the adsorption of the streptavidin is similar for most of the samples
(around 25 nm). Nevertheless, the shift is larger for the sample with 50% of biotin and reaches 36 nm
and a little bit lower for the sample with 80% of biotin (21 nm). The higher shift could be expected due
to a lower steric hindrance thanks to dilution. However, the sample with 80% of biotin groups does
not seem to follow this trend.

The effective thickness of these samples with an average resonance peak at 670 nm and a shift
of 25 nm is about 4.2 nm. It represents almost a full coverage of streptavidin with less density that the
previous sample fully covered. Moreover, we expected the sample with only 5% biotin to be less
sensitive, however, the shift is similar to other samples. Furthermore, this sample and the one with
100% of biotin resulted in similar shifts which agree with similar amounts of adsorbed streptavidin.

154
 In conclusion, even low quantity of biotin at the nanoparticle surface affords the similar
amount adsorption of the streptavidin. This suggests only few biotin groups are required for the
recognition. Moreover, the effect of the diluted groups at the nanoparticle surface can be observed
since the recognition is performed no matter the quantity of biotin groups. The sample with 50%
presents a larger shift which may suggest a better accessibility for the streptavidin.

(2) Variation of the size of the nanoparticles

The amount of biotin being directly dependent on the surface of nanoparticles, we have used
assemblies of nanoparticles of 6 and 22 nm. Furthermore, the nanoparticle sizes control the packing
of biotin groups because of curvature and facets which become predominant in the case of 20 nm sized
nanoparticles. These features are expected to influence steric hindrance of biotin groups. For each
nanoparticle size, two assemblies with different nanoparticle densities were prepared. An assembly
with 10 nm-sized nanoparticles with high density was also prepared. The diameters and densities are
summarized in table IV-6 and the SEM images are shown in figure IV-25.

a) b) c)

d) e)

Figure 0-25. SEM images of nanoparticle assembly of a,c) 6 nm and b,d) 22 nm with a,b) high and c,d) low coverage and e)
10 nm. Magnification X50000

Table 0-6. Diameters and densities of the corresponding sample with SEM images

Diameter (nm) Density (%)

a) 5.7 ± 0.9 nm 18%
b) 21.9 ± 1.6 nm 100%
c) 5.7 ± 0.9 nm 36%

155
 d) 21.9 ± 1.6 nm 26%
e) 10.1 ± 1.1 nm 47%

These assemblies were then functionalized with the alkynated biotin derivative in order to
perform the SPR measurements before and after the immersion for 30 minutes in an aqueous solution
of streptavidin at 100 μg/mL (figure IV-26). The sample e) is shown on the figure IV-25.

Figure 0-26. SPR measurements before (green) and after (red) dipping in a solution of streptavidin of samples with
different size and density of nanoparticles

The values of the resonance plasmon shifts and other parameters are summarized in table IV-
7. Before dipping in the streptavidin solution, the SPR peak is shifted to longer wavelengths with larger
nanoparticle sizes and densities as expected.
Nevertheless, the shifts of resonance plasmon observed after streptavidin adsorption are
unexpected. Indeed, the shifts are not linearly increasing with the incident wavelength. The sample of
6 nm-sized nanoparticles with high coverage and the sample of 22 nm-sized nanoparticles with low
coverage exhibit larger shifts than expected (27 nm and 42 nm respectively).
Table 0-7. Summarize of the different parameters of the sample with different volume of iron oxide at the surface

NP6 18% NP6 36% NP22 26% NP22 100% NP10 60%

Peak position 611 nm 626 nm 653 nm 893 nm 668 nm

Spectral response
11 nm 27 nm 42 nm 53 nm 27 nm
of the sensor Δλ

156
 Decay length
158 nm 172 nm 198 nm 440 nm 210 nm
ld
Sensitivity factor
2115nm/RIU 2220 nm/RIU 2460 nm/RIU 8070 nm/RIU 2630 nm/RIU
m
Effective thickness
1.7 nm 4.5 nm 7.3 nm 6.1 nm 4.6 nm
d

For the 6 nm-sized nanoparticles, the SPR shift significantly increases from 11 to 27 when the
density doubles (from 18% to 36% of the maximal theoretical value). This important variation cannot
be caused by the incident wavelength which slightly increases (+15 nm). To understand this
phenomenon, the determination of the number of streptavidin per nanoparticle as function of the
inter-particle distance and the size of the nanoparticles is necessary (figure IV-27).

Figure 0-27. Number of streptavidin per nanoparticle as function of the size of nanoparticles and the inter-particle
distance.

The low coverage of the gold thin film by 6 nm-sized nanoparticles implies large inter-particle
distances as observed on the SEM images. An average inter-particle distance of 10 nm and 6 nm were
calculated for samples with 36 % and 18 % in densities, respectively. These inter-particle distances
correspond to the maximum of streptavidin adsorbed on the nanoparticles (7.5
streptavidin/nanoparticle for nanoparticles with a diameter of 5.7 nm). Indeed, the maximum of
streptavidin per nanoparticles was reached with inter-particles distances up to 4.5 nm corresponding
to the size of the streptavidin. Moreover, as shown in the chapter II, the assembly of small
nanoparticles presents isolated nanoparticles which promote the high adsorption of streptavidin. The

157
largest shift observed in the case of the sample with the highest density is then caused by the increase
of nanoparticle at the surface and therefore number of streptavidin:

18 % (2210 NPs/μm²) à 7.5 streptavidin/NPs à 16 600 streptavidin/μm²

36 % (4420 NPs/μm²) à 7.5 streptavidin/NPs à 33 200 streptavidin/μm²

In the case of the 22 nm-sized nanoparticles, the large shift of 53 nm observed for the sample
with high density coverage is due to the high sensitivity factor at this wavelength up to 8070 nm/RIU.
In contrast, the sample with low coverage of 22 nm nanoparticles exhibits also a large shift of 42 nm,
whereas the sensitivity factor is far lower (2460 nm/RIU). This large shift may result from the
accessibility of the streptavidin at the nanoparticle surface. Indeed, the quantity of streptavidin per
nanoparticles increases strongly (up to 52 streptavidin per nanoparticle for distance inter-particles
larger than 4.5 nm):

26 % (450 NPs/μm²) à 52 streptavidin/NPs à 23 300 streptavidin/μm²

100 % (1710 NPs/μm²) à 20 streptavidin/NPs à 34 900 streptavidin/μm²

Even though, the assembly of 22 nm-sized nanoparticles is driven by dipolar interactions (See
chapter II) and form alignment as shown on the SEM images. However, the accessibility of biotin
groups seems better for this sample.

Nevertheless, the effective thickness calculated (7.3 nm) for the sample with 22 nm-sized
nanoparticle low coverage shows an overestimation of this value.

In conclusion, the sensitivity of the sensor is improved by the intrinsic optical properties such
as the incident wavelength which can be tuned by the nanoparticle size and density in the case of a
spectral interrogation system.[2] Therefore, the presence of nanoparticles improved the sensitivity by
tuning the position of the plasmon resonance, however, the nanoparticles allow to increase the surface
topography and the filling factor shows a strong influence on the sensitivity of the platform. A good
compromise is a large active surface with large inter-particles distance to promote the highest
sensitivity of the sensor.

158
 4. Kinetics measurements

biomolecular interactions. In this case, the interaction A + B → AB can be followed in real time, where
One of the main advantages of SPR biosensing is the determination of the kinetics of

A is the analyte and B the receptor molecule immobilized on the sensor surface. The rate constants
provide information on the strength of association and the tendency to dissociation. These rates are
dependent of several parameters such as the temperature or the flow rate.[43] Table IV-8 summarizes
the kinetic parameters and the association/dissociation constants.
Table 0-8. Definition of the rate and affinity constants

Association Dissociation
Association rate Association Dissociation rate Dissociation

[õ÷] Xûy
constant, kon constant, KA constant, koff constant, KD
[õ][÷] Xû§§
õ + ÷ → õ÷ = õ÷ → õ + ÷ =
[õ][÷] Xû§§ [õ÷] Xûy
Definition

number of AB
complexes number of AB
formed per unit Affinity to complexes
Description Stability of AB
time and unit association dissociated per
concentration A unit time
and B
Units l. molœ* . s œ* l. molœ* s œ* mol. lœ*

Typical range 10b − 10ý 10þ − 10*_ 10œ* − 10œÍ 10œþ − 10œ*_

The determination of these kinetic parameters will ensure the understanding of the interaction
mechanisms between the receptor molecules and the specific proteins.

The kinetic of the recognition process were studied by measuring the angular shift as function
of time. The microfluidic system allows performing injection of proteins in real time and monitoring
the recognition process. The experimental conditions were set at a constant flow rate (50 μL/min) and
a constant temperature (21°C). These measurements allow the determination of different parameters
of the biosensors such as the limit of detection (LoD), the association and dissociation constants and
the concentration of the analyte.

To determine these parameters, it is important to define the factors of the adsorption event.
The analyte in solution (A0) has to be transported to the surface (As) and then to interact with the
recognition element (B) to form the adsorbate (AB). The equation IV.7 gives the two regimes: (i)
transport (driven by the flow rate coefficients km and k-m) and (ii) reaction (driven by the rate
coefficients kon and koff).

õF ⇄QU õ¨ + ÷ ⇄Qøú õ÷
Q Q
ªU øùù
Equation IV.7

159
 The determination of the binding affinity may lead to error if the mass transport limitation
(MTL) is unaccounted. The MTL occurs when the reaction flux is higher than the transport flux and
may influence the adsorption mechanisms. In this part, the flow rate will be fixed for the set of
experiments in order to avoid the mass transport limitation.[43]

The fluidic system is composed of an injection loop with two positions; position 1 “load” where
the solution is injected inside the loop and position 2 “injection” where the solution contained in the
loop is pushed on the surface of the sample (figure IV-28).

a) b)

Figure 0-28. Schematic representation of the injection loop with the a) loading and b) injecting position

The degasser is pumping the solution onto the surface of the sample. The flow rate can be
tuned to control the speed of the analyte on the surface. The flow rate is an important factor here
since it determines the kinetic of the recognition process. It is important to limit the diffusion of the
analyte at the surface and in the same time that the analyte have time to bind at the surface. These
kinetics of diffusion have been studied and the mechanisms of recognition are dependent of them.[43]

(1) Kinetic measurement and determination of the limit of

detection of the Streptavidin

The kinetic measurements are described by the sensorgram which measure the angular
position of the resonance peak as function of time. This sensorgram allows observing separately the
association, equilibrium and dissociation phases (figure IV-29).

160
 Figure 0-29. Schematic representation of a sensorgram correlated to the different phases. Extracted from [30]

The reaction kinetics was studied onto a thin gold film fully covered by 10 nm iron oxide
nanoparticles decorated with biotin groups. The sensorgram was recorded by using a wavelength of
785 nm (figure IV-30).

Figure 0-30. Sensorgram of the nanoparticle surface functionalized with biotin recorded after the injection of streptavidin

The injection of streptavidin with a concentration of 100 μg/mL in pure water shows a strong
change of the refractive index. A large shift about 0.6° is reached in less than two minutes meaning a
fast adsorption onto the surface. Pure water was injected to remove non-specific adsorbed proteins.
The rinsing step affects only slightly the position of resonance peak meaning that the streptavidin is
strongly bound to the biotin groups through specific interactions.

The experiment was conducted several times with different concentrations of aqueous solution of
streptavidin to investigate the limit of detection of the platform. In parallel, the same experiment was
performed with a gold surface directly functionalized with biotin groups (no nanoparticles) in order to
determine the influence of the nanoparticle assembly on the kinetics. The figure IV-31 shows the
corresponding sensorgrams.

161
a) b)

Figure 0-31. Sensorgram measured for different concentrations onto a) nanoparticle assemblies functionalized with biotin
and b) a gold thin film directly functionalized with biotin.

All the curves show an increase of the resonance angle followed by a plateau in few minutes
after streptavidin injection. The sensorgrams indicate that the shifts of SPR angle decrease with lower
streptavidin concentration. The shifts observed for a concentration of 100 μg/mL are coherent with
the ones reported on the SPR curves presented in section A.2.a.1 (0.589 ± 0.028° and 0.453 ± 0.028°
for surface with and without nanoparticles, respectively). With low concentrations, the shifts decrease
until the signal to noise ratio becomes very low which agree with the limit of detection. The sample
with the nanoparticle assembly allowed reaching a concentration of 10 ng/mL (189.4 pM) for
streptavidin detection, whereas the surface without nanoparticles allowed reaching a concentration
of 100 ng/mL (1.89 nM). The nanoparticle assembly seems to improve the LoD which decreases of one
magnitude order. Thereafter the limit of detection will be determined more precisely.

Moreover, these typical sensorgrams allow determining the association and dissociation
constants and evaluating precisely the limit of detection of the system. For this, two methods are used
by fitting calibration curves with the rate and the shifts after injection of the protein.

(2) Determination of the association constant by the coefficient

rate method

An approach to determine the association and dissociation constants is to calculate the rate
constants (kon and koff) and therefore determine the affinity constant KA defined by equation IV.8:

<ÿ = =
* Qøú
OÂ Qøùù
Equation IV.8

These coefficient rate constants can be extracted from a calibration based on the initial
reaction rate as function of the concentration. The slope coefficient extracted after each injection of
the protein corresponds to the rate constant of the system noted kobs. The rate being dependent of the
concentration of streptavidin in solution, the rate constants were fitted with a linear curve as function
of the concentration with a good regression coefficient (0.992). The kinetics analysis allows

162
determining kobs which is the rate of the reaction observed and can be defined as function of the
concentration by:

XûY¨ = Xûy [•] + Xû§§ Equation IV.9

Therefore, a calibration curve was obtained for a fixed flowing rate at 50 μL/min (figure IV-32).

Figure 0-32. Rate coefficient plotted with the analyte concentration onto a surface with (blue curve) and without (black
curve) nanoparticles

The slope of the curve corresponds to the rate association coefficient and the intercept
corresponds to the rate dissociation constant. They can be determined as function of the
concentration of streptavidin. Here, the values of the rate coefficients are summarized in table IV-9.
Table 0-9. Summarize of the association and dissociation rate constants for the biotin/streptavidin couple

kon koff KA KD
-1 -1
(°.s .M ) (°.s-1) (M-1) (M)
With
(2,663 ± 0,109) x103 (8,299 ± 0,157) x10-5 (3,209 ± 0,131) x107 (3,116 ± 0,127) x10-8
nanoparticles
Without
(2,326 ± 0,131) x103 (6,165 ± 0,138) x10-5 (3,773 ± 0,297) x107 (2,651 ± 0,209) x10-8
nanoparticles

The values of association and dissociation rate constants are similar for the gold films covered
with and without nanoparticles. The nanoparticles do not seem to impact significantly the adsorption
of streptavidin. The high value of kon indicates a fast binding reaction whereas the low value of koff is
reliable to a strong stability of the complex. Moreover, the association and dissociation constants are
comparable with values found in literature for biotin/streptavidin interaction onto a substrate (KA =
106-107 M-1)[44-46]

163
 (3) Determination of the association constant by the Langmuir
binding isotherm

Another method can be used to determine the affinity constants, by using the angular shifts measured
at different concentrations. The affinity constant of the couple can be determined by the Langmuir
binding isotherm which is a non-linear regression defined by equation IV.10:

¤7 = [!]rO ¤s98
[!]
Equation IV.10
Ê

with [C] the analyte concentration, KD the dissociation constant, Rmax the maximum binding capacity of
the sensor when all the binding sites are occupied and Req the response of the system at the
equilibrium. The different concentrations used to determine the calibration curve are used to perform
the binding isotherm showed on figure IV-33.

Figure 0-33. Langmuir binding isotherms calculated for a surface with (blue curve) and without (black curve)
nanoparticles

The binding isotherm allows determining directly the dissociation constant and the maximum
binding capacity of our system. The values of Rmax extracted from these curves are different for the
samples with and without nanoparticles which was expected from the previous results. The maximum
binding capacities for the samples with and without nanoparticles are 0.685 ± 0.049° and 0,506 ±
0,021°, respectively. We can observe that the plateau where all the binding sites are occupied is
reached faster when there are no nanoparticles at the surface. It can probably be explained by the fact
that the streptavidin molecules have more difficulties to reach all the binding sites on a rough surface.
The values of KD and KA determined by the Langmuir binding isotherm model are summarized in table
IV-10.

164
Table 0-10. Summarize of the value of the rate coefficients and affinity constants for the biotin/streptavidin

KA KD
(M-1) (M)
With nanoparticles (2,311 ± 0,608) x106 (4,327 ± 1,136) x10-7

Without nanoparticles (7,716 ± 2,417) x106 (1,296 ± 0,406) x10-7

The values of the affinity constants are in the range of what is found in the literature.[45, 46]
However, the values calculated with this method differ from the one previously calculated. The
determination of KA and KD by the rate constant method gives affinity constants with an order higher
than the one obtained with Langmuir binding isotherm. This behavior has already been observed[43]
and the reason is linked to the mass transport limitation. Indeed, Schuck and Minton have reported
that the MTL influence the kobs vs [C] plot. Even with the low flow rate used (50 μL/min), the transport
flux of the analyte at the surface can influence the binding. Therefore, the Langmuir binding isotherm
seems to be the more convincing method.

Moreover, the limit of detection of the system can be extracted from these curves. A linear
regression can be performed for low concentrations and give the sensitivity k of the system (k = 1.89
x106 and k = 2.83 x106 for the system with and without nanoparticles respectively). The LoD is defined
by the equation IV.11:

"#Â =
Á$
l
Equation IV.11

with σ the error on the SPR system (0.001°). Thank to this, the LoD for both systems can be
determined and are 1.6 nM and 1.1 nM for surface with and without nanoparticles respectively. Finally,
the system without nanoparticles presents the lowest LoD.

The SPR technique is a powerful tool to study precisely the effect of nanoparticle assemblies
on the kinetics of the recognition process between streptavidine and biotin groups. Several parameters
such asassociation and dissociation constants and limit of detection were calculated. The
determination of the affinity constant KA and KD has been done by the rate coefficient and the Langmuir
binding isotherm methods. The rate coefficient method is limited by the transportation of the analyte
at the surface. Therefore, the Langmuir binding isotherm is the most accurate technique to determine
precisely the association and dissociation constants. The determination of the LoD of the system
showed that the nanoparticles do not improve the system.

5. Biofouling

The biofouling constitutes a critical point in biosensing. Indeed, the interaction between the
molecular receptor and the analyte has to be highly specific in order to avoid non-specific recognition.
Therefore, control experiments were performed to study the specificity of our sensing platform. The
streptavidin has a strong affinity with most of organic groups. Moreover, hydrophilic groups may favor
the physisorption of the streptavidin onto the nanoparticle surface. It can be demonstrated by the
injection of streptavidin in the presence of sample which correspond to each step of the sensor
construction.

165
 The control experiments were performed by injection of an aqueous solution of streptavidin
(100 µg/mL) on the alkyno-terminated gold thin film (SAM-CC), the azido-terminated nanoparticle
assembly (NP@N3). Finally, a non-recognition protein to the biotin was injected onto a surface of
nanoparticles functionalized with biotin groups: the bovin serum albumin (BSA).

The figure IV-34 presents the sensorgrams of the injection of aqueous solution of streptavidin
onto the SAM alkyne, the nanoparticle assembly which is not functionalized with biotin groups and
injection of aqueous solution BSA onto biotin-terminated nanoparticle assembly.

a) b) c)

Figure 0-34. Sensorgrams of the adsorption of streptavidin onto a) SAM-CC and b) nanoparticle assembly and c)
adsorption of BSA onto nanoparticle assembly functionalized with biotin

In the case of the SAM-CC, the adsorption of the streptavidin is shown by the change of the
angular peak position demonstrating the adsorption of the protein on the surface. However, the
rinsing step showed a partial desorption of the protein which is relevant of non-specific interactions.
The streptavidin may probably be physisorbed on the alkyne chains at the surface.

In the case of the azido-terminated nanoparticle assembly, the SPR signal showed an effect of
streptavidin injection characterized by the change of the angular peak position. Nevertheless, the
increase of the position is followed by a slight diminution which could represent a non-specific
recognition at the surface. This recognition is different to the SAM alkyne.

Finally, the adsorption of BSA onto biotin-terminated nanoparticle assembly showed another
behavior. A fast increase is observed immediately followed by a rapid decrease.

These control experiments present different values of the rate and the shift after streptavidin
or BSA injection. These values are summarized in table IV-11.
Table 0-11. Summarize of the rate and shift of the control experiments

SAM-CC NP@N3 Biotin BSA

Rate (°/s) 1.09 ± 0.14 0.26 ± 0.01 0.303 ± 0.006 1.24 ± 0.04
Shift (°) 0.247 ± 0.03 0.347 ± 0.03 0.595 ± 0.03 0.166 ± 0.03

The binding rate and shift of the resonance peak values are not corresponding to the
adsorption between biotin and streptavidin. The angular shift after injection of the streptavidin is
lower on the SAM-CC and on the nanoparticle assembly without grafting of the biotin. Moreover, the
rates of the binding are much higher in the case of the SAM-CC and the BSA onto the biotinylated
surface. The fast desorption of the streptavidin and the BSA after the rinsing step confirms the non-
specific interactions.

166
 6. Conclusion

The SPR systems constitute a strong tool to study the mechanisms of molecular recognition.
Here, the assembly of nanoparticles has been used as a platform to tune the signal of the surface
plasmon. The change of the optical properties allows increasing significantly the sensitivity of the
signal. Moreover, the presence of the nanoparticles influenced the surface chemistry and promoted a
higher accessibility of the receptor molecules by the analytes which increase more the sensitivity. It
has been confirmed by the change of the surface chemistry by varying the number of receptor
molecules at the surface and the size and density of nanoparticle assembly. The use of a microfluidic
device allowed understanding the mechanisms of detection between the biotin and the streptavidin
and identifying the rates of the recognition process. Finally, control experiments have been performed
to ensure the specificity of our platform. This specificity was highlighted with the shifts and rates of
different binding events.

167
 E. Iminosugar – Glycosidase

After using the biotin/streptavidin couple as a model system to demonstrate and to optimize
our nanostructured SPR platform, we decided to focus on the molecular recognition process between
an iminosugar and the α-Mannosidase (Jack Bean).10

1. Context and motivation

The interest in glycosidase expands since these enzymes play a key role in many biological
processes.[47] Indeed, these enzymes are able to cleave the glycosidic bond between two saccharide
molecules. The lack or dysfunction of glycosidase is involved in several diseases such as diabetes,
Gaucher disease, cancer, cystic fibrosis among others. Therefore, glycosidases inhibitors are used in
the treatments of these metabolic disorder and viral infections. Among different classes of
glycosidases inhibitors such as disaccharides, carbasugars or thiosugars, iminosugar are very attractive
because of their strong inhibition and their better oral bioavailability.[47]

Therefore, we can take advantage of such efficient inhibitors for detection of enzymes. On the
basis of previous works reported by P. Compain’s group we have focused on the interaction between
an iminosugar and the α-Mannosidase which has two recognition sites.[7-9] The grafting of the
iminosugar onto the surface of the nanoparticle assembly allowed us demonstrating the versatility of
our SPR platform as well as studying precisely the recognition mechanisms between the iminosugar
and the α-mannosidase.

2. Substrate post functionalization by iminosugar

One more time, we are taking advantage of our SPR bio-platform which consists in a
nanoparticle assembly grafted onto a gold substrate and which exhibit azide groups. In a similar way
we reported for biotin groups, the azide-terminated nanoparticle assembly was functionalized with an
alkyne modified iminosugar (DNJ-Alcyne)11 by performing the CuAAC reaction. In the light of the
previous results obtained with the biotin and streptavidin, 10 nm-sized nanoparticles were used in
order to compare the obtained results.

a) Experimental protocol

A dense monolayer of nanoparticles with a diameter centered at 10.4 ± 0.7 nm (5570 + 260
NPs/μm²) was prepared prior to graft the alkynated iminosugar at the surface (figure IV-35). The
iminosugar was grafted on the surface of the nanoparticle assembly by performing the CuAAC reaction.

10
Work conducted with P. Compain’s group, SYBIO, Strasbourg
11
The (2R,3R,4R,5S)-2-(hydroxymethyl)-1-(6-(prop-2-yn-1-yloxy)hexyl)piperidine-3,4,5-triol (DNJ Alcyne)
iminosugar is provided by the team of P. Compain.

168
 Figure 0-35. Schematic representation of the alkynated iminosugar

CuAAC “Click” reaction of the alkynated iminosugar on the surface of NP@N3 assembly was realized as
reported in chapter II. 6,5 mg of CuBr(PPh3)3 and 0,5 mL of Et3N were mixed with 5 mL of THF. 10 mg
of alkynated iminosugar were added. The nanoparticle assembly was immersed in the solution and
refluxed for 24 h under argon. After the reaction, the substrate was rinsed extensively with THF and
exposed for 1 minute to ultrasounds. The substrate was then dried by air stream and stored in a box
for further characterizations.

b) Characterizations

(1) Scanning electron microscopy

SEM images (figure IV-36) show the surface of the nanoparticle assembly before and after
grafting of the alkynated iminosugar.

a) b)

Figure 0-36. SEM images of the nanoparticle assembly (a) before and (b) after grafting of the iminosugar. Magnification
X50000

The density of nanoparticles is estimated by counting the nanoparticles on twelve different

areas for different samples and averaged at 5115 ± 315 NPs/μm² and 4985 ± 240 NPs/μm² before and
after grafting respectively. As observed for the biotin grafting, a slight decrease of the number of
nanoparticles can be noticed. However, both values are very close (below the standard deviation).

169
 (2) Water contact angle

Water contact angle measurements were performed before and after iminosugar grafting
(figure IV-37). The value for the azide terminated nanoparticles is 99.7 ± 0.4° which is in accordance
with the value obtained on previous samples. After iminosugar grafting, the contact angle value
decreased significantly (52.9 ± 0.6°) which agree with a more hydrophilic surface due to the polar
groups in the iminosugar.

a)
a) b)
99,7 ± 0.4° b) 52,9 ± 0.6°

Figure 0-37. Water contact angle measurements of a) NP@N3 surface and b) iminosugar surface

(3) XPS

XPS measurements were performed before and after the immobilization of the iminosugar in
order to study the grafting of the iminosugar at the surface of the nanoparticle assembly (figure IV-
38).

170
a) b)

c) d)

Figure 0-38. XPS measurements of the NP@N3 surface (a,b) and iminosugar surface (c,d) of the N1S (a,c) and C1S core
(b,d) level.

After the grafting of the iminosugar, the N1s core level region shows a single peak centered at
400.4 eV. The absence of peak at 405.5 eV indicates the azide groups are no longer present at the
surface meaning a total reaction with the alkyne groups. The peak located at 400.4 eV can be attributed
to the triazole bridge formed during the click reaction. The N-C bonds within the iminosugar can also
be convoluted in this peak. The C1s region show two components located at 284.9 eV and 288.3 eV and
can be attributed to C-H and C-C binding and O-C-O binding, respectively.[48]

(4) PM-IRRAS

PM-IRRAS measurements have also been performed on the surface before and after
iminosugar immobilization and showed on figure IV-39. The disappearance of the azide vibrational
band at 2100 cm-1 confirms the absence of azide groups at the nanoparticle surface after the “click”
reaction. The apparition of the strong vibrational band of the ether group ν(C-O-C) at 1100 cm-1 present
in the iminosugar also, indicates the correct grafting of the receptor at the surface. Moreover, no signal
of the alkyne band is observed after the immobilization of the iminosugar meaning a specific grafting
through the CuAAC “click” reaction.

171
a) b)

Figure 0-39. a) FT-IR spectrum of the alkynated iminosugar and b) PM-IRRAS spectra of the film of azide terminated
nanoparticles (red curve) and after the grafting of iminosugar (blue curve)

(5) Construction of the film followed by Surface Plasmon

Resonance Measurements

The different steps of the bio-platform have also been characterized with SPR measurements
to monitor the construction of the film. As for the biotin, the film construction includes several steps
which increase the refractive index at the surface of the gold thin film:

· Gold naked substrate (Au)

· Gold substrate functionalized with alkyne-SAM (SAM-CC)
· Assembly of nanoparticles functionalized with azide groups onto the SAM (NPs@N3)
· Assembly of nanoparticles functionalized with alkynated iminosugar (Iminosugar)

The SPR measurements were performed at a fixed wavelength of 785 nm. The reflectivity as
function of the incident angle is shown on figure IV-40 and the values of the resonance angles are
summarized in table IV-12.

172
 Figure 0-40. SPR reflectivity measurements with angular response at a fixed wavelength of 785 nm

The values measured for the naked gold thin film and the alkyne-terminated SAM are similar
from the values obtained previously. The small shift (< 0.05°) agrees with the SAM formation. The
position of the resonance angle for the assembly of nanoparticles is 69.764° and corresponds to a full
monolayer for 10 nm-sized nanoparticles. Surprisingly, the grafting of the iminosugar on the assembly
of nanoparticles does not induce a blue shift as observed for the grafting of biotin. The red-shift of
about 0.8° may correspond to a more stable assembly with less physisorbed nanoparticles.

Table 0-12. Summarize of the angular peak position for each step of the film construction

Wavelength Au SAM NPs@N3 Iminosugar

785 nm 65.698° 66.163° 69.764° 70.596°

To conclude, the different surface characterizations demonstrated the stability and the
reproducibility of the grafting of the iminosugar derivative on the surface of nanoparticle assembly.
Moreover, the total replacement of the azide groups at the surface by the iminosugar has been
highlighted thanks to XPS and PM-IRRAS measurements leading to an available surface for detection
of the α-mannosidase.

3. SPR measurements for the glycosidase

a) α-Mannosidase solution preparation

In the following measurements, α-Mannosidase was used from Canavalia ensiformis (Jack
Bean). The protein is used in a buffer solution for a better stabilization. The buffer solution is sodium
acetate at 0.2 M fixed at pH 5 which is optimum for the enzyme stabilization.

173
 The α-Mannosidase solution was then prepared with different concentrations starting with a
100 μg/mL solution and then diluted down to 10 ng/mL. The error on the measurement increases for
each dilution with really small amounts of protein because it can be adsorbed on the laboratory
equipment.

b) SPR measurements

The SPR measurements were performed on the iminosugar functionalized platform before and
after the injection of α-mannosidase at 100 μg/mL with a flow rate 50 μL/min. The figure IV-41 shows
the measurements performed at 785 nm.

Figure 0-41. SPR measurements performed at 785 nm before (red curve) and after (purple curve) adsorption of α-
mannosidase

A shift of the resonance peak is observed after the injection of the α-mannosidase which
confirms the adsorption of the enzyme and the accessibility of the iminosugar at the surface of the
nanoparticle assembly. The immobilization of the α-mannosidase resulted in an angular shift of 0.773
± 0.002°.

The increment refractive index ð&ïð) of the α-mannosidase has not been calculated. The used
of Freiter equation is therefore not possible. However, the mass of the adsorbed enzyme can be
estimated with good approximation by using the factor conversion 0.0001°≈ 0.038 ng/cm²
corresponding at the biotin/streptavidin couple. Considering the molecular weight of the α-
mannosidase (MW = 220,000 g/mol), 1.4 enzyme per nanoparticles were estimated. As previously seen
with the streptavidin, this value may be overestimate. SEM images reported by Lepage and al. showed
that the surface occupied by of α-mannosidase on cyclopeptides is about 10x10 nm².[7] Moreover,
granulometry measurements by DLS have been performed to estimate the radius of the
α_mannosidase in aqueous solution with a concentration of 50 μg/mL (figure IV-42).

174
 Figure 0-42. Granulometry measurements in volume of the α-mannosidase in aqueous solution

The size of the enzyme is centered to 10.1 nm which is in accordance with the size shown by
SEM analysis. Therefore, the α-mannosidase can be considered as an enzyme with a gyration radius
about 5 nm. Thus, the approximation of the surface occupied by an enzyme is about 78.5 nm².
According to our calculation (see part IV.A.4.), for a dense packed assembly of nanoparticles of 10.4
nm, the theoretical maximum coverage is 1.2 α-mannosidase per nanoparticle. Furthermore, the
conformation of the enzyme at the surface of the nanoparticle may allow more adsorption, for
example the formation of dimers.[7] The figure IV-43 gives an example of different configurations of
the α-mannosidase adsorption on the nanoparticle surface.

Figure 0-43. Schematic representation of two configuration examples of the α-mannosidase adsorption at the
nanoparticle surface

175
 c) Kinetics measurements

(1) Sensorgrams

A sensorgram allows monitoring the adsorption of the analyte to the surface. The sensorgram
corresponding to the injection of α-mannosidase on the iminosugar modified platform is studied and
presented on figure IV-44.

Figure 0-44. Sensorgram of α-mannosidase adsorbed on the iminosugar modified platform

A significant shift of the resonance angle was immediately observed after the injection of the
α-mannosidase. The shift reached 0.864° and leveled off after 2 minutes meaning a fast adsorption.
After rinsing with the buffer solution, a small shift to the lower angle was observed. This behavior can
be explained by the non-specific adsorption at the surface, the buffer has the effect to remove these
non-specific grafted proteins. This behavior is slightly different from the recognition between biotin
and streptavidin and can be relevant of a physisorption of the α-mannosidase at the surface of the
nanoparticle assembly.

In the same way that for the streptavidin, the limit of detection and the affinity constants of
the system can be determined by injection of α-mannosidase solutions with different concentrations
(figure IV-45).

176
a) b)

Figure 0-45. Sensorgrams after injection of the α-mannosidase with different concentrations onto a) nanoparticle
assembly functionalized with iminosugars and b) iminosugars directly grafted onto a SAM-CC

The system is responding to the range of concentration from 100 μg/mL down to 100 ng/mL
at which the signal to noise ratio becomes too low. When the nanoparticle assembly functionalized by
alkynated iminosugar is subjected to a concentration of α-mannosidase of 100 μg/mL, a large shift is
observed (0.925 ± 0.014°). This shift is in accordance with the previous shift observed and the slight
variations observed are caused by the difference in the quantity of nanoparticles at the surface. In the
case of the surface without nanoparticles, for the same concentration the angular shift is twice shorter
0.449 ± 0.014°. Moreover, for each concentration, the angular shifts are lower for the surface without
nanoparticles than the film. This behavior is similar to the biotin/streptavidin couple where the
sensitivity is higher on the nanoparticle assembly. The angular shifts can be detected down to a
concentration of 100 ng/mL which corresponds to 454 pM whatever the presence of nanoparticles.

(2) Determination of the rate constant affinity of the iminosugar

and α-mannosidase couple

As seen previously, the binding affinity constant can be calculated by two different methods.
The rate of the binding reaction allowed determining the rate constant of the recognition process
between the iminosugar and the α-mannosidase. The slope and angular shift were extracted from
sensorgram recorded for each concentration (table IV-13).
Table 0-13. Angular shift and the corresponding measured slope for different concentrations of α-mannosidase

With Nanoparticles Without Nanoparticles

Angular Slope Correlation Angular Slope Correlation
shift (°) (°/min) coefficient shift (°) (°/min) coefficient
100 μg/mL 0,9245 3,18 0,9998 0.449 0.74 0.956
10 μg/mL 0,7926 0,37 0,9973 0.265 0.079 0.995
1 μg/mL 0,0948 0,0117 0,9832 0.0212 0.0048 0.988
0.1 μg/mL 0,0122 0,0075 0,9137 0.0032 0.0033 0.874

177
 The figure IV-46 presents the curve of the rate of binding events versus the concentration of
the analyte.

Figure 0-46. Rate coefficient calculated for different concentrations of α-mannosidase with (blue curve) and without (red
curve) nanoparticles

The first observation is a different behavior between the samples with and without
nanoparticles. The association rate constant is much higher in the case of the nanoparticle film
assembly implying a faster binding reaction than for flat surface. The rate association and dissociation
coefficient kon and koff were determined by the slope and intercept of the curve and are summarized
in table IV-14.
Table 0-14. Summarize of the value of association and dissociation rate constant for the biotin/streptavidin couple

kon koff KA KD
-1 -1
(°.s .M ) (°.s-1) (M-1) (M)
With
(1,162 ± 0,081) x105 (1,788 ± 0,198) x10-5 (6,499 ± 1,173) x109 (1,549 ± 0,278) x10-10
nanoparticles
Without
(2,586 ± 0,042) x104 (8,681 ± 3,962) x10-5 (2,979 ± 1,407) x108 (3,357 ± 1,586) x10-9
nanoparticles

The affinity constants determined here are different for the sample with and without
nanoparticles. The highest association constant and lowest dissociation constant for the sample with
nanoparticles show a higher affinity and stability of the iminosugar with the α-mannosidase. In
contrary to the biotin/streptavidin couple, the nanoparticle assembly here seems having an effect of
the association of the enzyme. The important size and steric hindrance generated by the enzyme may
be responsible of the difference between a flat surface and a rough surface.

In another hand, the measure of the shift at the equilibrium versus the concentration of the
analyte (Langmuir binding isotherm) allows the determination of the dissociation constant and the
limit of detection of the system (figure IV-47).

178
 Figure 0-47. Langmuir binding isotherm calculated for a surface with nanoparticles (blue curve) and
without (red curve)

The binding isotherm allows determining directly the dissociation constant and the maximum
binding capacity of our system. The values of Rmax extracted from these curves are 1.006 ± 0.108° and
0.493 ± 0.024° for samples with and without nanoparticles, respectively. Here, the nanoparticle
assembly results in an increase about twice the maximum binding capacity. This means a more
favorable surface for adsorption of the enzyme. The values of KD and KA determined by the Langmuir
binding isotherm model and from the kon and koff values are summarized in table IV-15.
Table 0-15. Summarize of the affinity and rate constants

KA KD
(M-1) (M)
With nanoparticles (5.26 ± 2.60).107 (1.90 ± 0.94) .10-8
Without nanoparticles (2,37 ± 0.46) x107 (4,21 ± 0,79) x10-8

The values of the KD and Ka determined by the rate curves differ from the ones determined
with the binding isotherm. As discussed previously, the rate constant method is very sensitive to MTL
limitation and therefore can induce a difference of about one order of magnitude for the affinity
constants.[47] The Langmuir binding isotherm gives us KD values of 1.90 x10-8 mol/L and 4.21 x10-8 mol/L.
These values can be compared to the values of systems using the same couple in solution KD = 1.0 -
33.0 10-7 mol/L.[8] The major difference is the immobilization on the surface of the iminosugar in our
system which can modify the stability of the complex. Therefore, the inhibition of the iminosugar is
stronger when it is located onto a substrate than in solution.

The LoD for both systems can also be determined and are 1.2 nM and 4.6 nM for surfaces with
and without nanoparticles, respectively. Comparable values are obtained whatever the presence of
nanoparticles. In contrary, to the biotin/streptavidin couple, the system with nanoparticles has a better
LoD.

d) Competitive association

179
 The injection of a competitive substrate in solution will allow understanding the interaction
between α-mannosidase and the inhibiting agent. Therefore, the adsorption of α-mannosidase is
followed by the injection of an aqueous solution of methyl α-D-mannopyranoside at 10 mM.[49, 50] The
injection of the competitive substrate to iminosugar is followed by a rinsing step with pure water. The
figure IV-48 presents the sensorgram of the experiment.

Figure 0-48. Sensorgram after the injection α-mannosidase and methyl α-D-mannopyranoside

The injection of methyl α-D-mannopyranoside at 10 mM with a flow rate of 50 μL/min on the

surface covered with α-mannosidase showed a change of the angular position. However, the rinsing
step decreases the angular position of the resonance peak. Few conclusions can be deduced from this
observation: first, the methyl α-D-mannopyranoside does not remove the enzyme from the surface by
replacing the iminosugar. Indeed, the constant association between the iminosugar and the enzyme
seems larger than with this substrate. Second, the rinsing step removed most of the α-
mannopyranoside from the surface. However, a slight difference is observed after the injection of the
substrate (about 0.023°) meaning an immobilization of the substrate. Indeed, the molecular
recognition of α-mannosidase may happen from two specific sites. This information indicates that both
sites are not necessarily involved in the interaction with iminosugar. Therefore, α-mannosidase may
be adsorbed at the surface of nanoparticle assembly through only one binding site (figure IV-49).

180
 Figure 0-49. Schematic representation of the most favorable configuration of the α-mannosidase adsorption at the
nanoparticle surface

e) Regeneration of the surface

One of the current challenges on biosensing is the possibility to regenerate sensing surfaces in
order to develop reusable devices. The regeneration of the sensors requires the desorption of the
enzyme. Some studies showed the use of acidic medium to break bonds between enzyme and sugar.[33]
Here, we used a solution of phosphoric acid at 0.02 M and pH 2.5 which are subsequently injected
after the α-mannosidase injection onto a SAM functionalized with iminosugar. A second injection of α-
mannosidase was performed thereafter to see the correct binding. These experiments were conducted
first in the absence of nanoparticles, the iminosugar was directly grafted on the alkyne-terminated
gold thin film (figure IV-50).

181
Figure 0-50. Sensorgram after α-mannosidase injection and regeneration with phosphoric acid onto SAM functionalized
with iminosugar

The first injection of α-mannosidase at 10 μg/mL with a 50 μL/min flow led to a shift of 0.271°
as expected for this concentration. After the rinsing step with the buffer, the phosphoric acid solution
was injected. The injection was immediately followed by a strong and fast angular shift. This shift is
observed because of the change of refractive index of the solution of acid phosphoric (n = 1.341 ±
0.001). After rinsing with pure water, the resonance angle dropped fast to the initial value before
enzyme injection meaning a successful desorption of the glycosidase. A second injection was
performed in order to control the non-destruction of the iminosugar surface. The second angular shift
observed was about 0.265° which is quite close from the first one. Moreover, the binding rates were
calculated and were 0.091°/min and 0.081°/min for the first and second injections, respectively. These
results show the possibility to regenerate the surface. Nevertheless, some complementary
characterizations are required to control the presence of iminosugar at the surface after the acid
injection.

The same experiment was performed onto a gold thin film covered with nanoparticles and the
sensorgram is presented in figure IV-51.

182
 Figure 0-51. Sensorgram after α-mannosidase injection and regeneration with phosphoric acid onto nanoparticle
assembly functionalized with iminosugar

Here the sensorgram shows that after the injection with phosphonic acid the value of the
resonance angle drops below its initial value. Although SEM images show that nanoparticles are still
present at the surface, such a decrease of the SPR angle may be ascribed to the partial dissolution of
iron oxide nanoparticle in solution with low pH (< 5).

Nevertheless, the second injection of α-mannosidase resulted in the similar angular shift to
the first injection of enzyme. It may correspond to non-specific interactions at the surface of iron oxide
nanoparticles. Some additional experiments (use of pH > 5) and characterizations (XPS, PM-IRRAS,
AFM) are required to study the way the phosphoric acid affects the iron oxide nanoparticles and the
organic layer at their surface. Nevertheless, the study in the absence of nanoparticles showed that
phosphoric acid is very promising in the development of reusable sensors.

4. Biofouling

The specific recognition between the receptor molecule and the analyte is a key parameter as
we saw previously. Different control experiments have been performed in order to testify the
specificity of the recognition.

a) Control experiment: α-mannosidase on NPs@N3 surface

183
 The injection of α-mannosidase in a buffer solution of sodium acetate at 100 μg/mL was first
performed onto a nanoparticle assembly without functionalization with iminosugar. The azide surface
should not be specific for the enzyme recognition. The sensorgram was performed and presented on
figure IV-52.

Figure 0-52. Sensorgram after the injection of α-mannosidase onto nanoparticle assembly surface

The shift of the resonance angle (ΔθNP@N3 = 0.487°) shows the adsorption of α-mannosidase.
After the rinsing step, the desorption of species can be observed by the change of the resonance angle.
As observed for the streptavidin, the α-mannosidase was immobilized on the surface of azide
terminated nanoparticles. Nevertheless, the kinetics of the reaction is different from that in the
presence of iminosugars. Indeed, the shift of the resonance angle is twice lower (ΔθNP@N3 = 0.487° and
Δθiminosugar = 0.924°) for injections at the same concentration. The same difference is observed for the
rate of the reaction with kon values of 1.13°/min and 3.03°/min for the surface without and with
iminosugar respectively. Therefore, the interaction between α-mannosidase and an azido terminated
surface is not specific.

b) Control experiment: BSA on iminosugar surface

The injection of BSA onto a nanoparticle assembly functionalized by iminosugars has been
performed in order to control the specificity of the recognition. An aqueous solution of BSA at 1 mg/mL
was injected with a flow rate of 50 μL/min and the sensorgram is presented on figure IV-53.

184
 Figure 0-53. Sensorgram of the injection of BSA onto NP@N3 surface functionalized with iminosugar

The injection of BSA was noticed by an angular shift of 0.624° and a rate of 1.59°/min which
are also different from the α-mannosidase onto the same surface.

The biofouling is highlighted by different rates and shifts which allow the control of non-
specific recognition (table IV-16).
Table 0-16. Summarize of the rate and shift of the control experiment

NP@N3 Iminosugar BSA

Rate (°/s) 1.13 ± 0.02 3.03 ± 0.01 1.59 ± 0.04
Shift (°) 0.487 ± 0.002 0.864 ± 0.002 0.6264 ± 0.002

As observed for the biotin/streptavidin, the angular shift and rate after α-mannosidase
injection are lower in the case of the azide-terminated nanoparticle assembly. The same behavior is
observed for the adsorption of BSA onto iminosugar-terminated nanoparticle assembly even if, in this
case, the shift and rate are slightly higher.

5. Conclusion

The SPR allowed understanding the mechanism of recognition of the couple between α-
mannosidase and one of its natural inhibitor iminosugar. The use of our bio-platform showed that in
contrary to the biotin/streptavidin couple, the nanoparticles have an influence on the binding of the
enzyme. Moreover, the interaction between the iminosugar and the enzyme seems to be stronger
when the inhibitor is immobilized onto a substrate than in solution. This result has been highlighted
with the comparison with similar system in solution.[7-9]

Experiment with a competitive substrate gave us indications on the configuration of the

binding of the α-mannosidase. Moreover, preliminary experiment on the regeneration of the surface
showed that the system is promising and require more optimization and supplementary
characterizations.

185
 F. General conclusion

The assembly of nanoparticles allowed designing a bio-platform with post-functionalization of

receptor molecules. The different surface characterizations permitted to confirm the grafting of biotin
and iminosugar.

The SPR measurements allowed monitoring the detection of streptavidin. Moreover, the use
of this couple allowed investigating the sensitivity of our bio-platform and studying the recognition
mechanisms. The comparison between gold films covered with and without nanoparticles has shown
the enhancement of the sensitivity of the sensors. The use of different configurations permitted to
understand the origin of this enhancement. The control of the structuration of assemblies of iron oxide
nanoparticles allowed tuning the plasmonic resonance of gold substrates to enhance the sensitivity
factor. Moreover, the accessibility of functional groups at their surface increased in comparison with
flat surface to permit adsorbing a larger quantity of analyte.

The kinetic measurements in real-time allowed determining the affinity constants between
biotin and streptavidin. This kinetic study was also studied for the coupling between an enzyme, the
α-mannosidase, and iminosugar, one of its natural inhibitors. The recognition process monitored by
SPR showed that the affinity constants when the inhibitor is immobilized on a substrate are larger than
in solution.

186
 H. Bibliography

[1] Teramura, Y.; Arima, Y.; Iwata, H., Analytical biochemistry, (2006) 357, 208.
[2] Pichon, B. P.; Barbillon, G.; Marie, P.; Pauly, M.; Begin-Colin, S., Nanoscale, (2011) 3, 4696.
[3] L. Haussling and H. Ringsdorf, F.-J. S. a. W. K., Langmuir, (1991),
[4] J. Spinke, M. L., H.-J. Guder, L. Angermaier and W. Knoll, Langmuir, (1993),
[5] Rempel, B. P.; Withers, S. G., Glycobiology, (2008) 18, 570.
[6] Borges de Melo, E.; da Silveira Gomes, A.; Carvalho, I., Tetrahedron, (2006) 62, 10277.
[7] Lepage, M. L.; Schneider, J. P.; Bodlenner, A.; Meli, A.; De Riccardis, F.; Schmitt, M.; Tarnus,
C.; Nguyen-Huynh, N. T.; Francois, Y. N.; Leize-Wagner, E.; Birck, C.; Cousido-Siah, A.; Podjarny, A.;
Izzo, I.; Compain, P., Chemistry, (2016) 22, 5151.
[8] Decroocq, C.; Joosten, A.; Sergent, R.; Mena Barragan, T.; Ortiz Mellet, C.; Compain, P.,
Chembiochem : a European journal of chemical biology, (2013) 14, 2038.
[9] Bonduelle, C.; Huang, J.; Mena-Barragan, T.; Ortiz Mellet, C.; Decroocq, C.; Etame, E.; Heise,
A.; Compain, P.; Lecommandoux, S., Chem Commun (Camb), (2014) 50, 3350.
[10] Gouget-Laemmel, A. C.; Yang, J.; Lodhi, M. A.; Siriwardena, A.; Aureau, D.; Boukherroub, R.;
Chazalviel, J. N.; Ozanam, F.; Szunerits, S., The Journal of Physical Chemistry C, (2013) 117, 368.
[11] Spinke, J.; Liley, M.; Schmitt, F. J.; Guder, H. J.; Angermaier, L.; Knoll, W., The Journal of
Chemical Physics, (1993) 99, 7012.
[12] Vıctor H. Perez-Luna, M. J. O. B., Kimberly A. Opperman, Philip D. Hampton, Gabriel P. Lopez,
Lisa A. Klumb, and Patrick S. Stayton, JACS, (1999),
[13] Marinakos, S. M.; Chen, S.; Chilkoti, A., Anal Chem, (2007) 79, 5278.
[14] Nath, N.; Chilkoti, A., Anal Chem, (2002) 74, 504.
[15] D'Agata, R.; Palladino, P.; Spoto, G., Beilstein journal of nanotechnology, (2017) 8, 1.
[16] Rios, F.; Smirnov, S., ACS applied materials & interfaces, (2009) 1, 768.
[17] Williams, E. H.; Davydov, A. V.; Motayed, A.; Sundaresan, S. G.; Bocchini, P.; Richter, L. J.;
Stan, G.; Steffens, K.; Zangmeister, R.; Schreifels, J. A.; Rao, M. V., Applied Surface Science, (2012)
258, 6056.
[18] James P. Collman, N. K. D., Todd P. A. Eberspacher, and Christopher E. D. Chidsey, Langmuir,
(2006) 22, 2457.
[19] Claire-Marie Pradier, M. S., Liu Zheng, Gerard Jaouen, Surface Science (2002) 502, 193.
[20] Liu, Y.; RamaRao, N.; Miller, T.; Hadjipanayis, G.; Teplyakov, A. V., The Journal of Physical
Chemistry C, (2013) 117, 19974.
[21] Gao, S.; Koshizaki, N.; Tokuhisa, H.; Koyama, E.; Sasaki, T.; Kim, J.-K.; Ryu, J.; Kim, D.-S.;
Shimizu, Y., Advanced Functional Materials, (2010) 20, 78.
[22] Lapin, N. A.; Chabal, Y. J., The journal of physical chemistry. B, (2009) 113, 8776.
[23] A. Lahav, M. A., I. Abdulhalim, Optics Letters, (2008) 33, 2539.
[24] Christy, P. B. J. a. R. W., Phys. Rev. B, (1972) 6, 4370.
[25] B. Liedberg, I. L., Sensors and Actuators B: Chemical, (1993) 11, 63.
[26] B. P. Pichon, A. D., M. Pauly, K. Mougin, A. Derory, S. Bégin-Colin, J Phys Chem C, (2010) 114,
9041.
[27] S. Szunerits, R. B., Langmuir, (2006) 22, 1660.
[28] Wijaya, E.; Lenaerts, C.; Maricot, S.; Hastanin, J.; Habraken, S.; Vilcot, J.-P.; Boukherroub, R.;
Szunerits, S., Current Opinion in Solid State and Materials Science, (2011) 15, 208.
[29] Linda S. Jung, C. T. C., Timothy M. Chinowsky, Mimi N. Mar, and Sinclair S. Yee, Langmuir,
(1998) 14, 5636.
[30] Shalabney, A.; Abdulhalim, I., Laser & Photonics Reviews, (2011) 5, 571.
[31] Szunerits, S.; Boukherroub, R., Chem Commun (Camb), (2012) 48, 8999.
[32] Szunerits, S.; Shalabney, A.; Boukherroub, R.; Abdulhalim, I., Reviews in Analytical Chemistry,
(2012) 31,

187
[33] Maalouli, N.; Barras, A.; Siriwardena, A.; Bouazaoui, M.; Boukherroub, R.; Szunerits, S., The
Analyst, (2013) 138, 805.
[34] R. Reiter, H. M., W. Knoll, Langmuir, (1993) 9, 2430.
[35] Barbillon, G.; Faure, A. C.; El Kork, N.; Moretti, P.; Roux, S.; Tillement, O.; Ou, M. G.;
Descamps, A.; Perriat, P.; Vial, A.; Bijeon, J. L.; Marquette, C. A.; Jacquier, B., Nanotechnology, (2008)
19, 035705.
[36] Barbillon, G.; Bijeon, J. L.; Plain, J.; Royer, P., Thin Solid Films, (2009) 517, 2997.
[37] J. Spinke, M. L., H-J. Guder, L. Angermaier, W. Knoll, Langmuir, (1993) 9, 1821.
[38] W. M. Albers, I. V.-L., Nano-Bio-Sensors, Springer Ed. New York, NY, (2011).
[39] W. M. Albers, I. V.-L. Ed. Nano-Bio-Sensing: Surface Plasmon Resonance on Nanoscale
Organic Films, (2011); Vol. p 83.
[40] V. H. Pérez-Luna, M. J. O. B., K A. Opperman, P. D. Hampton, Gabriel P. Lopez, L A. Klumb, P.
S. Stayton, JACS, (1999) 121, 6469.
[41] Shulei Zhao, D. S. W., W. M. Reichert, Langmuir, (1993) 9, 3166.
[42] L. S. Jung, K. E. N., C. T. Campbell, P. S. Stayton, S. S. Yee, V. Pérez-Luna, G. P. Lopez, Sensors
and Actuators B, (1999) 54, 137.
[43] R. Schasfoort, A. J. T., Handbook of Surface Plasmon Resonance, The Royal Society of
Chemistry: (2008).
[44] Y.Tang, R. M., X. Zeng, Anal Chem, (2006) 78, 1841.
[45] Tang, Y.; Mernaugh, R.; Zeng, X., Anal Chem, (2006) 78, 1841.
[46] N. M. Green, E. J. T., Biochem. J., (1970) 118, 67.
[47] N. F. Bras, N. M. C., M. J. Ramos, P. A. Fernandes, Expert Opinion, (2014) 24, 1.
[48] Dhayal, M.; Ratner, D. M., Langmuir, (2009) 25, 2181.
[49] B. Hultberg, P. K. M., S. Sjoblad, Biochimica et Biophysica Acta, (1976) 445, 398.
[50] Eiji ICHISHIMA, N. T., Masamichi IKEGUCHI, Yasunori CHIBA, Motoyoshi NAKAMURA, Choko
KAWABATA, Takashi INOUE, Koji TAKAHASHI, Toshiki MINETOKI, Kenji OZEKI, Chieko KUMAGAI,
Katsuya GOMI, Takashi YOSHIDA, Tasuku NAKAJIMA, Biochem. J., (1999) 339, 589.

188
 General Conclusion

The purpose of this work was to control the structuration of nanoparticle assemblies in order
to control the magnetic collective properties of these objects and their influence on plasmonic
properties carried by a substrate. Therefore, assemblies of nanoparticles onto gold substrates were
prepared through specific intermolecular interactions.

The thermal decomposition, by controlling the operating conditions, allowed synthesizing

spherical iron oxide nanoparticles with different sizes. The structural and magnetic properties were
investigated as function of their diameter. The XRD measurements highlighted the partial oxidation of
the nanoparticles, especially in the case of 5 nm-sized nanoparticles. The magnetic measurements
showed the superparamagnetic behavior of these nanoparticles at room temperature. The blocking
temperature was studied as function of the size of the nanoparticles. The increase of the nanoparticle
diameter led to an increase of the magnetic anisotropy which displaces the blocking temperature to
higher values.

Thereafter, these nanoparticles were functionalized with a ligand carrying a functional group
active for a CuAAC “click” reaction. The infrared spectroscopy and granulometry measurements
highlighted the correct grafting of the ligand at the surface of the nanoparticle and the colloidal
stability of the suspension after functionalization. Moreover, gold substrates were functionalized with
thiol molecules with a complementary head group to form a SAM. The infrared spectroscopy and
contact angle measurements of the gold surface showed the presence of the thiol molecules.

The assemblies of the azido-terminated nanoparticles onto the alkyne-terminated SAM were
performed by CuAAC “click” reaction and the conditions of reaction were studied. The variation of
reaction time allowed controlling the spatial distribution of the nanoparticles on the gold surface from
few isolated nanoparticles to dense monolayers. The kinetic of the assembly was studied for two
different concentrations and showed a faster kinetic for larger nanoparticles than small nanoparticles
which do not reach the full monolayer. The investigation on the colloidal stability of the nanoparticle
suspension after different reaction times showed that the aggregation of nanoparticles was favored
for larger nanoparticles. The assembly kinetics seems to be driven by the aggregation in suspension
and the dipolar interactions especially in the case of 20 nm-sized nanoparticles. Finally, the magnetic
properties of assemblies prepared with different sizes and different densities were performed. The
measure of the magnetization as function of an applied field and the temperature highlighted that the
NP20 present stronger dipolar interactions than the NP5 and NP10 whatever the density of
nanoparticles. This behavior confirmed that the assembly is driven by these dipolar interactions and
explains the faster kinetic and stronger tight-packing of the NP20.

A new assembly strategy was reported based on specific intermolecular interactions between
nucleobases. By taking advantage of the “click” chemistry, the nanoparticles and substrates were post-
functionalized with complementary nucleobases. The use of micro-wave irradiations allowed
decreasing strongly the reaction time down to few minutes. Alkyne-modified thymine was grafted onto
azido-terminated nanoparticles and the control of the spatial arrangement of the functional groups at
their surface was performed by diluting the functional groups during the ligand exchange. The control
of the quantity of thymine groups at the nanoparticle surface was ensured by infrared spectroscopy
and the stability of the suspension was investigated. The nanoparticles functionalized with thymine

189
groups presented a different stability as function of the solvent used. It was shown that the use of
polar solvent favored the stability of thymine-terminated nanoparticles due to the interactions
between the thymine molecules and the solvent. The functionalization of the substrate was also
performed by grafting azido-modified adenine on the alkyne-terminated SAM. The control of the
spatial arrangement was also performed by dilution of the functional groups at the surface. Infrared
and photo-electron spectroscopies and contact angle measurements highlighted the presence of the
adenine at the surface. The assembly process through multiple hydrogen binding was then performed
by dipping the adenine-terminated substrate in the thymine-terminated nanoparticle suspension. The
control of the density was investigated as function of the solvent, the quantity of nucleobases and the
temperature. The control of these parameter allowed finding the best conditions to favor high
coverage of the surface.

In the last part of this work, the assembly of nanoparticles was used to control the plasmonic
properties of the gold substrate. The control of the presence of functional groups after the assembly
through “click” reaction allowed obtaining a “bio-platform” able to be functionalized with a large
variety of receptor molecules. The grafting of biotin, confirmed by infrared and photo-electron
spectroscopies, allowed detecting the streptavidin by surface plasmon resonance measurements. The
sensitivity and the recognition process were investigated with and without the presence of
nanoparticles at the gold thin film surface. The enhancement of the sensitivity with the presence of
nanoparticles was highlighted. Investigations on the nature of the enhancement showed the
nanoparticles modified the intrinsic optical properties of the gold substrate. Moreover, the
nanoparticles gave a better accessibility of the biotin groups at the surface which allowed the detection
of a higher number of streptavidin.

This strategy developed for protein was extended to enzyme detection. The recognition
process was studied with the biotin/streptavidin as well as iminosugar/glycosidase couple. The alkyne-
modified iminosugar was grafted onto the azido-terminated nanoparticle assembly in order to detect
the α-mannosidase. The measure of the surface plasmon resonance as function of the time allowed
monitoring in real time the adsorption of the analyte. From these data, affinity constants and limit of
detection of the analyte were determined. The results, in comparison with similar study in solution,
showed a higher inhibitor power of the iminosugar grafted onto a solid surface than in solution.

190
 Outlook

This work led to a great number of perspectives. The nanoparticle assembly through
nucleobases recognition was studied with nanoparticles functionalized with thymine. However, it is
possible to functionalized nanoparticles with adenine. Therefore, the self-assembly of multilayer
induce by the external conditions (solvent, temperature, π-stacking) can be imagined.

A preliminary study was done by functionalizing 10 nm-sized nanoparticles with adenine

diluted at the surface with a terminal methylene group in a ratio 50/50 (NP@Ade 50%). The suspension
of a mixture with nanoparticles functionalized with both type of nucleobases allowed forming
aggregation in solution through the base pairing (figure I).

Figure LIV. DLS measurements on a suspension of NP@Thym 50% and NP@Ade 50% in DMF at 25°C (red curve) and 55°C
(blue curve)

The control of the solvent and the temperature could control the aggregation since the hydrogen
bonds are sufficiently weak. Indeed, granulometry measurements showed that aggregate of NP@Ade
50% and NP@Thym 50% at room temperature do not present aggregation when the temperature
reaches 55°C.

Moreover, the assembly of nanoparticles through self-recognition of nucleobases can be

extended to larger structures. The assembly of nanoparticle via DNA strands has already been
performed in literature. However, most of the studies were done with gold nanoparticles which are
easily functionalized in water. Preliminary study showed that the iron oxide nanoparticles can be
functionalized with DNA strand mixed with a PEG on their surface to have pseudo stability in a mix

191
water/ethanol. Moreover, the functionalization of a gold substrate with the complementary DNA
strand resulted in the self-assembly of nanoparticles (figure II).12

Figure LV. SEM image, AFM image and the corresponding topographic profile of the assembly of NP10 via DNA strands

This experiment showed the feasibility of this system even if the optimization of the operating
conditions has to be done.

In the case of the SPR measurement, the bio-platform allowed grafting a large panel of bio-
receptors. Therefore, the study of the recognition process of different couples of molecules can be
performed. Moreover, the use of different sizes and shapes of nanoparticles could allow finding a
perfect configuration to optimize the sensitivity.

Finally, the CuAAC “click” chemistry, thanks to its versatility, allows performing assembly of
nanoparticles from different compositions. Therefore, gold nanoparticles which carry plasmonic
properties could be grafted onto substrates in order to induce magneto-plasmonic properties.

12
Collaboration X. Cattoën, Institut Néel, Grenoble
D. Gasparutto, CEA, Grenoble

192
 Appendix A

Anisotropy energy of a nanoparticle

The anisotropy energy results of the contribution of different energies:

69 = 6!! + 6[ + 6%

with EMC the magnetocrystalline anisotropy energy, ES the surface anisotropy energy and EF
the shape energy.

Ø The magnetocrystalline energy EMC comes from the interaction between the
magnetization and the crystal lattice which try to align the magnetic moments in a direction
specific of the material. These crystallographic directions are called magnetic easy axis. For a
single nanoparticle, the magnetocrystalline energy at first order is:

6!! = <> @A&²(C)

Ø When the size of a nanoparticle decrease, the ratio surface/volume increase strongly
this implies a lead of the surface effect. The diminution of neighbor atom at the surface leads
to break the symmetry and create disorder of the spin at the surface of the nanoparticle. The
direct consequence of this size reduction is the disordered layer at the surface which has a spin
glass magnetic behavior where the magnetization of surface atom is different of the one from
volume atom. This effect is called spin canting and is responsible of the reduction of saturation
magnetization in nanoparticle. The spin canting is strongly dependent of the size of the
nanoparticle but also of the synthesis method and the ligand bound on the surface atoms.[104,
105]

Ø When a magnetic field is applied on a nanoparticle, charge appears around the

nanostructure which induces a magnetic field opposite of the field applied. This field called
demagnetizing field can be written:

&: = −'Z
where M is the magnetization of the system and N the demagnetizing factor. This field
is dependent of the magnetization direction relative to the shape of the sample. The associate
energy is given by:

69̈ = _ 3F >('8 Z8_ + '• Z•_ + '• Z•_ )

(9˜7 *

193
 Where Nx, Ny, Nz, Mx², My² and Mz² are the component of the demagnetizing factor and
the magnetization in the directions x,y and z and with Nx + Ny + Nz = 1, μ0 is the vacuum
permeability and V the nanoparticle volume. In the case of a spherical nanoparticle, Nx = Ny =
Nz = 1/3 and does not present shape anisotropy.

The magnetic behavior of a nanoparticle will dependent of contribution of these energies. All
energies are influenced of by the field applied, the temperature, the shape and size of the nanoparticle.
Therefore, the magnetic properties, especially the reversal magnetization dynamic will depend of
these parameters.

Atomic Absorption Spectroscopy (AAS)

The determination of concentration of the nanoparticles suspension was performed by atomic

absorption spectroscopy.

A calibration curve of absorbance was performed by measuring the absorption of solution with
an iron concentration known. The sample of iron oxide nanoparticle suspension was dissolved in
hydrochloric acid (12 M). The measure of adsorption is then performed to know the iron concentration
[Fe]. The iron oxide concentration [Fe3O4] is then calculated by multiplying the iron concentration with
a factor 1.38.

Inter-particle distance

The inter-particle distance d is calculated with the approximation that the nanoparticles are
homogenous distributed onto the surface and is expressed as:

+ ¿ _ 1
® = )* − À Ç Ì ,
' 2 À

with S the area considered for the density calculus (in our case, S will always be 1 μm² = 106
nm²), N the number of nanoparticles in the surface S (determined by SEM) and D, the nanoparticles
diameter.

Alkynated biotin synthesis

194
Alkynated Biotin was synthetized by a method from Cao and al (2013)[14]. N-(3-Dimethylaminopropyl)-
Nʹ-ethylcarbodiimide hydrochloride (EDC.HCl) (92 mg, 0.48 mmol) was added to a solution of biotin
(100 mg, 0.4 mmol) and N-hydroxysuccinimide (51 mg, 0.448 mmol) in DMF (15 mL) in a round bottom
flask. The solution was stirred for 24 h at room temperature under nitrogen and evaporated in vacuum.
The crude solid was triturated with methanol several times to form N-hydroxysuccinimidobiotin as a
white solid product (73 mg, 54%). Propargylamine (21 μL, 0.32 mmol) was added to a solution of this
compound (54 mg, 0.16 mmol) and triethylamine (43 μL, 0.32 mmol) in DMF (8 mL). The solution was
stirred for 24 h at room temperature under nitrogen, concentrated, and purified by column
chromatography (silica gel, with a 6:1 chloroform/methanol eluent) to give alkynated biotin as a white
solid (57 mg, 94% yield).

Angular position model

The theoretical position of the resonance angle is determined by:

1: 1s 1
C = arcsin -) .
1: + 1s &˜

with np, the prism refractive index, εd and εm the dielectric and metal permittivities, respectively.
Moreover, the dielectric permittivity εd is determined with the model of Paletto et al.

9 121 12 + (12 + 211 )2 (1 − /)11 9(1 − /)122 11 + (11 + 212 )2 /12

1G// = (1 − /) +/
[(12 + 211 )(1 − /) + 311 /]2 [(11 + 212 )/ + 312 (1 − /)]2
with f and 1 - f the volume percentage of the material 1 and 2, in the layer respectively. In our case,
we will consider the first material as iron oxide ε1 = 5.57 and the second material as water ε2 = 1.77.
The values of f and 1 - f are calculated by the volume occupied by the nanoparticles in a cube of 1 x 1
x δ. With δ the penetration depth of the surface plasmon (figure A-1).

1 μm

Figure.A 1. Schematic representation of the volume occupied by the nanoparticle

195
Determination of the available surface by biotin

The available surface of biotin groups correspond to the spherical cap at the surface of the
nanoparticle where the streptavidin can reach. This surface is expressed as:
¿ − 4.5 + 0
+ = À¿(1 − sin(arccos - 2 .)
¿
2
with D the nanoparticle diameter with the organic layer at the surface (2 nm) and x the inter-
particule distance. The number of streptavidin is then determined by division of the area occupied by
a streptavidin (20.25 nm²).

196
 Appendice B

In this appendice, we will describe the surface plasmons behavior by using these equations at
the metal/dielectric interface.

In free space, the Maxwell equations are defined as:

®A/ 6‰⃗ =
1
,-
3.1

‰⃗ = 0
®A/ ÷ 3.2

‰‰‰‰‰‰⃗46‰⃗ 5 = œ6N
‰⃗
2Ž3 6ˆ
3.3

‰⃗5 = 3F (1F 6Î + 7⃗)

‰‰‰‰‰‰⃗4÷
‰⃗
2Ž3 6ˆ
3.4

‰⃗ represent the electric field, ‰B⃗ the induction of the magnetic field, εF the dielectric permittivity
where E
in vacuum (εF = 8.85 ∗ 10œ*_ A_ . s d . kg œ* . mœb ), µF the magnetic permeability in vacuum (µF =
4π. 10œý kg. m. Aœ_ . s œ_ ), ρ the charge density and ⃗ȷ the current density.

We introduce ‰D⃗ and H

‰‰⃗ which are the electric displacement field and magnetic field, for non-magnetic
and with an external current density null (μr = 1 and ⃗ȷ = 0), given by:

‰⃗ = 1F 6‰⃗ + <‰⃗
¿ 3.5

‰⃗ = ‰‰⃗
‰⃗
−Z
N
&
.-
3.6

where P‰⃗ and ‰M

‰‰⃗ represents the electric and magnetic dipolar moments respectively. Here we will
consider non-magnetic materials implying M ‰‰‰⃗ = 0 and linear implying that the following expression of
‰⃗ is:
P

<‰⃗ = 1F 76
‰⃗ 3.7

with χe the electric susceptibility

The displacement of the electric field (3.5) and magnetic field (3.6) leads to:

‰⃗ = 1F 12 6‰⃗
¿ 3.8

‰⃗ = N‰⃗
& .-
3.9

with εØ = 1 + χ7 the relative permittivity also called the dielectric constant of the material.

If we consider now two semi-infinite medium, non-magnetic and with a zero external current density
(μr = 1 and j = 0) as represented on figure B-1. The Maxwell equations for this system are:

197
 y
Dielectric
x
z Metal

Figure.B 1. Schematic representation of the metal/dielectric interface

‰⃗ = 0
®A/ ¿ 3.10

‰⃗ = 0
®A/ ÷ 3.11

2Ž346‰⃗ 5 = 6ˆ
‰‰‰‰‰‰⃗ œ6N ‰⃗
3.12

‰‰‰‰‰‰⃗ ‰⃗ 5 = 1F 12 6Î
‰⃗
2Ž34&
6ˆ
3.13

The wave equations can be expressed with the following equations:

‰‰‰‰‰‰⃗ ‰‰‰‰‰‰⃗46‰⃗ 5Ò = ‰‰‰‰‰‰⃗ ‰⃗5)

‰‰‰‰‰‰⃗4÷
‰⃗
2Ž3 Ñ2Ž3 2Ž3 Ñ 6ˆ Ò = − 6ˆ (2Ž3
œ6N 6
3.14

®A/4®A/ 6‰⃗ 5 − ∆6‰⃗ = −3F 1F 12

6²Î‰⃗
6ˆ c
3.15

knowing ∆6‰⃗ = ∇²6‰⃗ , we obtain:

∇²6‰⃗ − =0
,? 6²Î‰⃗
'² 6ˆ c
3.16

with ) = n.
*
- ,-

The same reasoning can be used for the magnetic field:

‰⃗ − =0
‰⃗
,? 6²"
∇²& '² 6ˆ c
3.17

Considering the fields to have a harmonic time dependency, the electric field can be expressed as

6‰⃗ = 6
‰‰‰‰⃗F G •Q- 2œ•Œˆr@ 3.18

with ω = 2πf the angular frequency, r the position vector, k F the wave number and ϕ the phase (we
4−A“6‰⃗ 5
6²Î‰⃗
= =
6
6ˆ c 6ˆ
will consider ϕ = 0). The derivative from time of the electric field becomes

A _ “_ 6‰⃗ = −“²6‰⃗ and ∇²6‰⃗ =

6²Î‰⃗ 6²Î‰⃗ 6²Î‰⃗
+ +
68 c 6• c 6• c
. If we assume that the electric field is propagating along x

= 0 and the field can be expressed as 6‰⃗ = 6‰⃗ (•)G

6²Î‰⃗ •F8
6• c
axis and invariant on z axis, we have: with β
the propagation constant of the electric field.

= 68 4A•6‰⃗ 5 = A²•²6‰⃗ = −•²6‰⃗

6²Î‰⃗ 6
68 c
So

and (3.16) becomes:

198
 ‰‰‰‰⃗ ‰‰‰‰⃗
‰‰‰‰‰⃗
− •_6 •+ “²6 =0
6²Î ž ,?
6• c '² • 3.19

‰‰‰‰⃗
‰‰‰‰‰⃗
+6 • ( XF 12 − • ) = 0
6²Î ž _ _
6• c
3.20

with XF =
Œ
'

In the same way, we find for the magnetic field:

‰‰‰‰‰⃗
‰‰‰‰‰⃗
+& • ( XF 12 − • ) = 0
6²" ž _ _
6• c
3.21

To find the component of the fields 6‰⃗ and &

‰⃗ we use the equations (3.12) and (3.13) knowing that = 6
6ˆ
−A“ we can assume the following equations:
ð6• ð6• ð&• ð&•
⎧ − = A“3F &8 ⎧ − = −A“1F 12 68
⎪ ðL ð• ⎪ ðL ð•
⎪ ⎪
ð6• ð68 ð&• ð&8
− = A“3F &• − = −A“1F 12 6•
⎨ ð0 ðL ⎨ ð0 ðL
⎪ð68 − ð6• = A“3 &
⎪ ⎪ð&8 − ð&• = −A“1 1 6
⎪
⎩ ð• ð0 F •
⎩ ð• ð0 F 2 •

= A• and the system is invariant in z, therefore =

6 6
68 6•
0, conducting to simplified equations:
The wave propagation is following the x axis,

ð6• ð&•
⎧ − = A“3F &8 ⎧ = A“1F 12 68
⎪ ð• ⎪ ð•
A•6• = A“3F &• A•&• = −A“1F 12 6•
⎨ð6 ⎨ð&
⎪ 8 − A•68 = A“3F &• ⎪ 8 − A•&• = −A“1F 12 6•
⎩ ð• ⎩ ð•
Two sets of solutions can be obtained, the first represents the transverse magnetic modes (TM) where

into the incidence plane) where HN = 0 , EO = 0 and therefore HQ = 0, bringing the equation system
the magnetic field is polarized perpendicularly to the incidence plane (and the electric field is contained

to:
ð&•
⎧ = A“1F 12 68
⎪ ð•
A•&• = −A“1F 12 6•
⎨ð6
⎪ 8 − A•68 = A“3F &•
⎩ ð•
Two solutions can be deduced from this system:

6• = Œ,
œF
&•
- ,?
3.22

68 =
ϥ 6"R
Œ,- ,? 6•
3.23

By substituting (3.22) in the equation (3.20) we obtain the wave equation for the radial modes TM :

‰‰‰‰⃗• ( XF _ 12 − • _ ) = 0
‰‰‰‰‰⃗R
+&
6²"
6• c
3.24

199
The second set of solution represent the transverse electric modes (TE) where the electric field is

plane) where EN = 0 and therefore EQ = 0 and HO = 0, bringing the equation system to:
polarized perpendicularly to the incidence plane (and the magnetic field is contained into the incidence

ð6•
⎧ − = A“3F &8
⎪ ð•
A•6• = A“3F &•
⎨ð&
⎪ 8 − A•&• = −A“1F 12 6•
⎩ ð•

&• = 6
F
Œ.- •
3.25

&8 = Œ.
• 6ÎR
6•
3.26
-

By substituting (3.25) in the equation (3.21) we obtain the wave equation for the azimuthal modes TE:

‰‰‰‰⃗• ( XF _ 12 − • _ ) = 0
‰‰‰‰⃗
+ 6
6²Î R
6• c
3.27

Equations (3.24) and (3.27) allow studying plasmons confined at the interface.

The solution for radial modes (TM) is given by the solution of equation (3.24), then combined with
equations (3.22) and (3.23) we obtain the solutions for Ey(y) and Ex(y) inside the dielectric (y > 0) and
inside the metal (y < 0).

In the dielectric, y > 0:

&• (•) = õ: G •F8 G œ–’ • 3.28

68 (•) = Aõ: Œ, S: G •F8 G œ–’•

*
- ,’
3.29

6• (•) = −õ: G •F8 G œ–’ •

*
Œ,- ,’
3.30

In the metal, y < 0:

&• (•) = õs G •F8 G –U• 3.31

68 (•) = −Aõs S G •F8 G –U•

*
Œ,- ,U s
3.32

6• (•) = −õs Œ, G •F8 G –U•

*
- ,U
3.33

with Ad et Am the amplitudes in the dielectric and in the metal respectively and γd,m = β² - k0²εd,m.

Considering these two media as continuous, the conditions at the limits have to be applied. This imply
than the tangential component of the electric field is continuous through the section while the
tangential component of the magnetic field is equal to the current density through the interface. The
tangential fields at the interface have to be continuous, to satisfy this condition. using equations from
(3.28) to (3.33) it implies that:

õs = õ: 3.34

=−
–U –’
,U ,’
3.35

200
If we consider a metal without absorption, equation (3.35) implies that the real part of the permittivity
has to be of opposite sign. It is effective for a system with a metal/dielectric interface. This relation
allows obtaining the dispersion relation for plasmon propagation at the interface:

• = XF n =
,’ ,U Œ ,’ ,U
,’ r,U
n
' ,’ r,U
3.36

In the same way, for the azimuthal modes, these equations allow finding the expression of Ez(y) and
bring to the following relations:

In the dielectric, y > 0:

6• (•) = õ: G •F8 G œ–’ • 3.37

&8 (•) = −Aõ: S G •F8 G œ–’ •

*
Œ.- :
3.38

&• (•) = õ: G •F8 G œ–’ •

*
Œ.-
3.39

In the metal, y < 0:

6• (•) = õs G •F8 G –U• 3.40

&8 (•) = −Aõs Œ. Ss G •F8 G –U•

*
3.41
-

&• (•) = õs Œ. G •F8 G –U•

*
3.42
-

Respecting the conditions at the limits conduct to:

Tq (Uq + Up ) = m 3.43

The solution for this condition is Am = 0 and therefore surface plasmons do not exist for the TE
polarization.

201
202
 Appendix C

Infrared and granulometry characterization of NP20 and NP5

Figure.C 1. Granulometry measurements of the nanoparticles in THF before (black curve) and after (blue curve) ligand
exchange for NP20 (left) and NP5 (right)

Figure.C 2. Infrared spectra of the nanoparticles with oleic acid (black curve) and the nanoparticles functionalized with
the azide phosphonic acid (blue curve) for NP20 (left) and NP5 (right)

1
HNMR of 11-mercapto-undecyn

203
 Figure.C 3. 1HNMR spectra of the 11-mercapto-undecyn

1
HNMR (CDCl3, 400 MHz): 2.50 (q, J = 7.0 Hz, 2H); 2.15 (td, 2H); 1.94 (t, 1H); 1.4-1.6 (m, 4H); 1.22-1.41
(m, 14H)

FT-IR and 1HNMR of alkynated biotin

Figure.C 4. FT-IR spectrum of the derivative alkynated biotin molecule

204
 Figure.C 5. 1HNMR spectra of the alkynated biotin

1H NMR (DMSO-d6) δH: 1.29−1.33 (m,2H), 1.44−1.51 (m, 2H), 1.59−1.62 (m, 2H), 2.07−2.10 (d, J = 7.5
Hz, 2H), 2.57−2.59 (d, J = 13.0 Hz, 1H), 2.81−2.84 (dd, J1 = 4.5 Hz, J2 = 5.0 Hz, 1H), 3.04−3.09 (m, 2H),
3.83−3.83 (m, 2H), 4.130 (m,1H), 4.29−4.31 (m, 1H), 6.31 (s, 1H), 6.36 (s, 1H), 8.17 (s, 1H).

Characterizations for 0% nanoparticle assemblies

Figure.C 6. PM-IRRAS spectra of SAM-CC (red line) and biotin functionalized film (blue line)

205
206
 Résumé de thèse en français

Chapitre I – Etat de l’art

Le travail effectué dans ce manuscrit porte sur l’intérêt d’assembler des nanoparticules sur des
surface dans le but d’étudier les propriétés des nanoparticules ainsi que de la surface en fonction de
la structuration de leur assemblage. Plus particulièrement, l’influence de l’assemblage des
nanoparticules sur des surface pour permettre d’améliorer les performances de détection de
biomolécules a été étudié.

Le control de l’assemblage de nanoparticules sur une surface est primordial puisque leurs propriétés
sont dépendantes de leur taille, leur forme et leur composition, mais aussi de leur distance à chacune
d’entre elles. De plus leur incorporation dans des dispositifs pour des applications nécessite
fréquemment leur immobilisation sur une surface. Il va donc de soi que pour moduler leurs propriétés
physiques, dépendantes des interactions individuelles et collectives des nanoparticules, le contrôle de
leur arrangement spatial est crucial. L’organisation à la surface d’un substrat va dépendre de la
technique de déposition ainsi que de la catégorie d’interactions. Il existe de nombreuses techniques
qui permettent d’assembler des nanoparticules sur des surfaces de façon simple et rapide comme les
méthodes de déposition dite « externes » tel que le dipping, le drop casting, le spin coating et le dip
coating. Ces méthodes permettent d’assembler des nanoparticules sur des grandes surfaces et de
façon compactes, néanmoins ces techniques souffrent du manque de spécificité lors des assemblages.
D’autres méthodes permettent une spécificité et donc une reconnaissance entre la surface du substrat
et des nanoparticules. Pour cela, les surfaces sont fonctionnalisées avec des groupements spécifiques
complémentaires qui vont permettre la reconnaissance entre les nanoparticules et le substrat. Cette
spécificité permet donc un bon contrôle de l’arrangement spatial des nanoparticules.

L’assemblage de nanoparticules permet d’accéder à un grand nombre d’application, notamment dans

le cas de nanoparticules d’oxyde de fer qui présentent des moments magnétiques. Le contrôle de leurs
distribution spatial à la surface d’un substrat permet le contrôle des propriétés magnétiques et permet
donc de réaliser des dispositifs pour le stockage d’information ainsi que des capteurs magnéto-résistifs.
Cependant, au-delà des propriétés intrinsèques des nanoparticules qui peuvent être modulées, les
propriétés physiques du substrat sur lequel elles reposent peuvent également être contrôlées. En
effet, en utilisant un substrat présentant des propriétés optiques particulières, le dépôt de
nanoparticules peut les modifier. En effet, le phénomène de résonnance plasmon, qui provient de
l’oscillation des électrons de conduction couplé avec une onde lumineuse produit à la surface d’un
substrat, une extinction lumineuse. Cette extinction est dépendante de l’angle et de la longueur d’onde
incidente de l’onde lumineuse et est également très sensible aux variations d’indice de réfraction à la
surface.

207
 diélectrique
βsp

y θ2 métal

x
z prisme
ky θ1

kx

Figure 56. Représentation schématique du phénomène de résonance plasmon en utilisant un prisme dans la
configuration de Kretschmann.

En effet, le phénomène de résonance plasmon de surface se produit lorsque le vecteur d’onde d’une
onde évanescente transmise à travers un prisme se couple avec l’oscillation des électrons à l’interface
entre le métal et un diélectrique (figure 1). Lorsque se phénomène se produit, il y a une brusque
diminution de l’onde réfléchie dans le prisme pour un angle précis définit par :

1: 1s 1
C[‘£ = sinœ* )
1: + 1s &˜2•¨s

Ainsi, la dépendance avec l’indice de réfraction du diélectrique nd² = εd est clairement démontrée. La
détection moléculaire va pouvoir avoir lieu entre des molécules réceptrices greffées à la surface du
substrat et une espèce à détecter en solution.

La détection plasmonique est aujourd’hui déjà utilisée dans de nombreux domaines (biologie,
agroalimentaire, médecine, environnement, sécurité) et présente certains avantages tel que la non
nécessité de marquer les éléments à détecter ainsi que le fait d’avoir une détection rapide et en temps
réel. Néanmoins, aujourd’hui il reste des limites à ce système de détection, notamment dans le cas de
solution extrêmement diluée ou pour des très petites molécules. C’est pourquoi l’augmentation des
performances est primordial pour améliorer ce type de dispositifs.

L’utilisation de nanoparticules pour augmenter les performances des capteurs SPR a déjà été étudié
principalement en utilisant les nanoparticules pour faire de la détection de type « sandwich » où
l’adsorption de l’analyte va induire l’assemblage ou l’agrégation des nanoparticules sur la surface et
donc amplifier le signal SPR. Cependant, dans ce manuscrit, on va s’intéresser à l’assemblage des
nanoparticules dans un premier lieu pour ensuite permettre l’augmentation des propriétés physique
du substrat. Cela passera par la synthèse et fonctionnalisation des nanoparticules dans un premier
temps, puis la fonctionnalisation du substrat et l’assemblage des nanoparticules aux travers de liaisons
spécifiques avant de pouvoir étudier la détection moléculaire en utilisant des couples de molécules
ayant un intérêt biologique.

208
Chapitre II – Structuration de l’assemblage de nanoparticules d’oxyde de fer

L’intérêt de cette partie est de synthétiser et d’assembler des nanoparticules d’oxyde de fer
sur des substrats d’or. La synthèse des nanoparticules par décomposition thermique permet un bon
contrôle de leurs tailles et de leurs morphologies permettant ainsi de contrôler leurs propriétés
intrinsèques. Les nanoparticules sont ensuite assemblées sur des surfaces au travers de liaison
covalentes réalisés par chimie « click ». Les assemblages sont ensuite étudiés en fonction de la taille
des nanoparticules et des conditions opératoires de la réaction de « click ».

Synthèse des nanoparticules

Les nanoparticules sont synthétisées par la méthode de décomposition thermique qui consiste
à décomposer un précurseur métallique dans un solvant à haute température d’ébullition en présence
d’un tensioactif. L’avantage de cette méthode est la séparation des phases de nucléation et de
croissance des germes formées pendant la réaction. Ainsi, le contrôle de la température permet ainsi
que la quantité des réactifs permet de modifier la taille et la morphologie des nanoparticules formées.
Le stéarate de fer II est ainsi chauffé dans le dioctyléther avec de l’acide oléique en suivant une rampe
en température contrôlée par un ordinateur. Plusieurs synthèses sont donc réalisées en modifiant les
conditions opératoires pour obtenir des nanoparticules avec un diamètre de 5 nm, 10 nm et 20 nm.

Les synthèses sont ensuite purifiées pour retirer l’acide oléique libre en solution ainsi que les
réactifs résiduels. La purification se fait par centrifugation en dispersant les nanoparticules dans un
mélange de solvant. Les solvants utilisés sont le chloroforme, qui présente une forte affinité avec les
nanoparticules, et l’acétone qui présente une meilleure affinité avec l’acide oléique. Ces étapes de
purifications sont suivies par spectroscopie infrarouge (FT-IR) afin de visualiser la disparition des
réactifs n’ayant pas réagi et de l’acide oléique libre en solution. Ainsi, après 22 lavages, les bandes
vibrationnelles caractéristiques de l’acide oléique libre en solution (située à 1710 cm-1) ont totalement
disparues. De plus la proportion des bandes vibrationnelles correspondant aux chaînes alkylènes (à
2920 cm-1, 2850 cm-1 et 720 cm-1) ont considérablement diminuées par rapport aux bandes
vibrationnelles correspondant au liaisons Fe-O de l’oxyde de fer (située entre 500 cm-1 et 800 cm-1).

La taille et la morphologie des nanoparticules est ensuite caractérisée par microscopie

électronique à transmission. Les images TEM permettent d’observer des nanoparticules sphériques et
homogènes en taille pour les trois différentes synthèses. Une mesure statistique est réalisée à l’aide
du logiciel imageJ en comptant plus de 200 nanoparticules. La distribution en taille est ensuite fitter à

échantillons et sont centrées à 5. 1 ± 0.8 nm, 10.1 ± 1.1 nm et 21.2 ± 1.2 nm pour les nanoparticules
l’aide d’une fonction de Gauss (figure 2). Les distributions en taille sont étroites pour les trois

qui seront appelées NP5, NP10 et NP20 respectivement.

209
 a)

b)

c)

Figure 57. Images TEM des nanoparticules fonctionnalisées par l'acide oléique et la distribution en taille
correspondantes. a) NP5, b) NP10 et c) NP20.

La stabilité colloïdale des nanoparticules est vérifiée par des mesures de granulométrie dans
le tétrahydrofuran (THF). Les mesure par DLS sont monomodales avec un facteur de polydispersité
faible ce qui indique une seule population de nanoparticules en suspension. Les diamètres
hydrodynamiques mesurées sont de 8.7 nm, 12.5 nm et 23.8 nm pour les NP5, NP10 et NP20
respectivement. Le léger décalage de 2-3 nm qui peut être observé entre la mesure par TEM et par DLS
est expliqué par la mesure du diamètre hydrodynamique dans le cas de la DLS qui inclut les molécules
organiques en surface des nanoparticules ainsi que la sphère de solvatation. Dans les trois cas, les
nanoparticules présentent une bonne stabilité colloïdale dans le THF.

Les nanoparticules sont également caractérisées par diffraction des rayons X (XRD) afin de
déterminer les phases crystallines et leurs compositions. Les diffractogrammes obtenus pour les trois
tailles de nanoparticules permettent d’identifier une structure spinelle. Un affinement de Rietveld
permet de déterminer avec précision le paramètre de maille des oxydes de fer qui se situe entre laes
phases de la magnétite (8.396 ± 0.001 Å, JCPDS file n°00-019-0629) et de la maghemite (8.338 ± 0.001
Å, JCPDS file n°00-013-0458). Cette proportion est plus importante pour les NP5 et NP10, cela
s’explique par la plus forte contribution de l’oxidation de surface pour les nanoparticules avec un faible
diamètre.

210
 Les propriétés magnétiques des trois tailles de nanoparticules sont ensuite étudiées en
utilisant un magnétomètre SQUID. Des cycles d’aimantations mesurant l’aimantation d’un échantillon
en fonction d’un champ magnétique appliqué sont réalisés à 300 K et à 5 K. Les cycles d’aimantation
fermés à 300 K indiquent un comportement superparamagnétique des nanoparticules. A faible
température, l’ouverture des cycles d’aimantation indiquent que les moments magnétiques sont
bloqués. L’aimantation à saturation MS, le champ coercitif HC et le champ rémanent HR peuvent être
extraits des courbes d’aimantations et sont résumés dans le tableau 1.

Tableau 1. Champs coercitifs, aimantations rémanentes et aimantations à saturation pour les différentes tailles de
nanoparticules à 5K.

Echantillon HC (Oe) MR/MS MS (emu/g)

NP5 240 ± 30 0.36 ± 0.05 55 ± 5
NP10 495 ± 30 0.35 ± 0.05 55 ± 5
NP20 330 ± 30 0.16 ± 0.05 75 ± 5

L’aimantation à saturation est inférieure à celle de la magnétite à l’état massif (92 emu/g) à
cause de la présence d’une couche oxydée en surface des nanoparticules. En effet la maghémite
présente une aimantation à saturation plus faible (74 emu/g), et la composition des nanoparticules
avec un plus faible diamètre (NP5 et NP10) possédant une plus grande proportion de maghémite que
les NP20, leurs aimantations à saturation est donc plus faible. De plus, des défauts à la surface pouvant
entraîner un désordre magnétique (canting de spin) peut également diminuer l’aimantation à
saturation.

Des mesures d’aimantations en fonction de la température sont ensuite réalisées. Les courbes
obtenues passent par un maximum d’aimantation qui est assimilé à la température de blocage TB des
nanoparticules qui correspond au passage de l’état superparamagnétique à l’état ferromagnétique.
Les valeurs de TB sont de 20 K, 75 K et185 K pour les NP5, NP10 et NP20 respectivement.
L’augmentation de la température de blocage avec la taille des nanoparticules est attendu. En effet,
l’augmentation du volume des nanoparticules augmente leurs énergies magnétocrystalline, et
l’énergie thermique nécessaire pour retourner les moments magnétiques va donc être plus
importante.

Fonctionnalisation des nanoparticules et du substrat

Pour pouvoir réaliser l’assemblage des nanoparticules sur des substrats au travers de la
réaction de chimie « click », il est nécessaire de fonctionnaliser les nanoparticules et les substrats avec
des molécules possédant les groupements complémentaires.

Les nanoparticules subissent donc un échange de ligand pour remplacer l’acide oléique en
surface par une molécule portant un groupement azoture. Le 12-azido-dodecyl-phosphonic acid
(AP12N3) est utilisé, car la fonction acide phosphonique possède une interaction avec l’oxyde de fer
plus forte que l’acide oléique. Il va donc spontanément le remplacer lors d’un échange direct de ligand
dans le THF. L’échange se fait sous agitation pendant 16 heures. A la fin de l’échange de ligand, la
suspension de nanoparticules est purifiée pour éliminer les molécules libres en solution. La purification

211
se fait par ultrafiltration qui consiste à filtrer la suspension colloïdale sur une membrane de cellulose
(30 kDa). Les molécules libres vont ainsi passer à travers la membrane, tandis que les nanoparticules
resteront en suspension. La purification est suivie par spectroscopie infrarouge et par granulométrie
pour observer la disparition des molécules libres en solution ainsi que la stabilité des suspensions.

La caractérisation par spectroscopie infrarouge montre l’apparition de bandes vibrationnelles

caractéristiques des acides phosphoniques entre 900 cm-1 et 1150 cm-1. De plus, la bande
caractéristique des azotures à 2100 cm-1 peut être observée ce qui confirme le greffage des molécules
d’AP12N3 à la surface des nanoparticules. La mesure DLS montre un diamètre hydrodynamique de 6.5
nm, 11.8 nm et 21.2 nm pour les NP5, NP10 et NP20 respectivement. La légère diminution du diamètre
hydrodynamique correspond à la différence de longueur de chaine entre la molécule AP12N3 et l’acide
oléique. Les nanoparticules sont donc correctement fonctionnalisées et présente une bonne stabilité
en suspension après l’échange de ligand.

La seconde étape consiste à fonctionnaliser des substrats d’or avec les groupes
complémentaires pour la réaction de chimie « click ». Des groupes azotures étant localisés à la surface
des nanoparticules, des groupes alcynes doivent être greffés à la surface des substrats. L’utilisation
d’une surface d’or permet la formation de monocouche auto-assemblés (SAM) en greffant des thiols.
L’arrangement de dérivés thiolés avec des chaines allylènes permet d’avoir les groupes terminaux
accessible à la surface de la SAM.

La SAM est formée en utilisant le 11-mercapto-undécyn préalablement synthétisé. Les

substrats d’or sont nettoyés à l’aide d’un plasma d’oxygène et d’hydrogène pendant 2 minutes pour
éliminer les éléments organiques à la surface. Les substrats sont ensuite immergés dans une solution
éthanolique à 10 mM de 11-mercapto-undécyn. Les substrats restent dans la solution pendant 24
heures avant d’être rincés à l’éthanol.

Les substrats ainsi fonctionnalisés sont caractérisés par microscopie à force atomique (AFM)
pour observer une augmentation de l’épaisseur à la surface correspondant à la présence d’une
monocouche de molécules. Les substrats sont ensuite caractérisés par angle de contact (WCA) en
déposant une goutte d’eau à la surface. L’angle de contact avant fonctionnalisation est de 21.7° ce qui
correspond à la surface très hydrophile de l’or. Il passe à 84.1° après le greffage des thiols, ce qui
correspond à une surface relativement hydrophobe causé par la présence des longues chaines
alkylènes des thiols. La spectroscopie infrarouge en polarisation de phase (PM-IRRAS) permet de
confirmer la présence des molécules à la surface du substrat avec la présence des bandes
caractéristiques des groupes alcynes terminaux situés à 3320 cm-1.

Assemblage des nanoparticules par chimie « click »

Les nanoparticules et les substrats d’or ayant été fonctionnalisés par les groupes
complémentaires azotures et alcynes respectivement, l’assemblage des nanoparticules va pouvoir être
réalisé. Les SAM sont rincés avec du THF avant d’être introduit dans une solution d’assemblage avec
5 mL de nanoparticules dans le THF, 0.5 mL de triéthylamine et 6.5 mg d’un catalyseur au cuivre
(CuBr(PPh3)3). La solution est chauffée à reflux à 70°C pendant 48 heures avec un réfrigérant sous
argon. Une fois la réaction terminée, les substrats sont retirés, puis rincés au THF 1 minute dans un
bain à ultrasons pour retirer les nanoparticules physisorbées à la surface.

212
 Les assemblages sont caractérisés par microscopie électronique à balayage (SEM) afin de
déterminer la couverture du substrat. Les images SEM permettent de déterminer la densité de
nanoparticules à la surface et de la comparer à la densité théorique maximale obtenue dans le cas d’un
empilement hexagonal compact (figure 3). Pour les trois tailles de nanoparticules on obtient un taux
de couverture de 13940 NPs/μm², 5810 NPs/μm² et 1820 NPs/μm² pour les NP5, NP10 et NP20
respectivement.

a) b)

Figure 58. Images SEM des nanoparticules de 10 nm assemblées à la surface de la SAM aux grandissement a) X100000 et
b) X50000.

Le contrôle de la structuration de l’assemblage peut se faire en jouant sur la cinétique de la

réaction. En effet, des assemblages réalisés avec différents temps de réaction permet de modifier la
densité de nanoparticules à la surface des substrats. Ainsi, pour des temps de réaction courts (inférieur
à 4 heures), des nanoparticules isolées sont observées sur la surface des substrats. Néanmoins, la
vitesse de recouvrement dépend de la taille des nanoparticules. Dans le cas des NP20, la surface arrive
à 80% de la densité théorique maximale après 4 heures de réaction, tandis que pour les NP10 il faut
attendre 48 heures. Dans le cas des NP5, après 48 heures de réaction, la densité atteint seulement
40% de la densité théorique maximale. Cela peut s’expliquer par l’influence des interactions dipolaires
magnétiques qui est plus importante pour les nanoparticules de 20 nm et qui peut diriger et accélérer
l’assemblage.

Pour conclure cette partie, la synthèse par décomposition thermique permet la formation de
nanoparticules d’oxyde de fer stable en solution avec un très bon contrôle de la taille et de la
morphologie. Trois différentes tailles ont donc été synthétisées : 5.1 ± 0.8 nm, 10.1 ± 1.1 nm et 21.2 ±
1.2 nm. Les nanoparticules ont été fonctionnalisées avec un groupement azoture tandis que les
substrats d’or ont été fonctionnalisés avec un groupement alcyne. L’assemblage des nanoparticules a
ainsi pu être réalisé par la réaction de chimie « click ». La variation du temps de réaction à permis de
contrôler la distribution spatiale des nanoparticules sur les surface d’or. L’étude de l’assemblage des
différentes tailles de nanoparticules a montré l’influence des interactions dipolaires magnétiques sur
leurs assemblages.

213
Chapitre III – Auto-assemblage des nanoparticules au travers de liaisons hydrogène multiple entre
des bases azotées

Dans cette seconde partie, l’assemblage des nanoparticules d’oxyde de fer sera étudié au
travers de liaisons hydrogènes multiples. L’auto-assemblage via des liaisons faibles permet une
géométrie plus flexible et dynamique que les liaisons covalentes. La stratégie repose sur la
fonctionnalisation des nanoparticules et du substrat par des bases azotées complémentaires greffées
par l’intermédiaire de la chimie « click ». Le processus de reconnaissance se base sur l’interaction
Watson-Crick entre les bases azotées de l’ADN : adénine-thymine (AT). Les bases azotées sont
modifiées avec des groupements azotures et alcynes pour pouvoir être greffées à la surface des
nanoparticules recouvertes par des azotures et des SAMs alcynes (figure 4).

a) b)

Figure 59. Représentation schématique de a) la thymine modifiée avec un alcyne et b) l'adénine modifiée avec un
azoture.

La première étape consiste donc à fonctionnaliser le substrat d’or par l’adénine modifiée avec
azoture. Pour cela, une SAM alcyne est immergé dans une solution d’assemblage contenant 10 mg
d’adénine-azoture. La réaction se déroule sous irradiation micro-onde, ce qui permet de diminuer
fortement le temps de réaction à 30 minutes. Les substrats sont ensuite rincés et séchés à l’air sec. La
quantité d’adénine à la surface peut être modifié en réalisant des SAM avec différentes quantités
d’alcyne à leurs surfaces. Ainsi l’alcyne est « dilué » avec une autre molécule : le dodécanethiol (DDT)
ne possédant pas de groupes actifs pour la réaction de « click ». Des SAM avec différents ratio
alcyne/DDT sont ainsi préparées : 0%, 20%, 50%, 80% et 100%.

Les SAMs modifiées avec l’adénine sont ensuite caractérisées. La mesure de l’angle de contact
permet d’observer une diminution de l’hydrophobicité à la surface avec l’augmentation du ratio
alcyne/DDT. Cela confirme une plus grande présence d’adénine, qui possède des groupements polaires
et donc un comportement hydrophile, à la surface du substrat. Les SAMs sont caractérisées par
spectroscopie des photo-électrons (XPS) et PM-IRRAS pour confirmer la présence de l’adénine à la
surface. L’XPS présente un pic intense à 400 eV correspondant aux liaisons N-C et N-H contenus dans
l’adénine. De plus, l’analyse par spectroscopie infrarouge présente deux bandes vibrationnelles à 1660
cm-1 et 1580 cm-1 qui correspondent aux bandes caractéristiques des liaisons N-C contenus dans
l’adénine et qui confirme sont greffage à la surface de la SAM.

La seconde étape consiste à fonctionnaliser les nanoparticules terminées par un azoture avec
la thymine modifiée avec un alcyne. De la même manière que pour les substrats, la thymine à la surface
des nanoparticules est diluée avec une autre molécule ne présentant pas de groupes actifs pour la

214
réaction de « click » : le dodecylphosphonique acid (DPA). Les nanoparticules sont placées dans une
solution d’assemblage avec la thymine et la réaction est réalisée sous irradiation micro-onde pendant
5 minutes. Les nanoparticules sont ensuite purifiées par ultrafiltration pour retirer les molécules libres
en solution. Ainsi des nanoparticules avec différents ratios thymine/DPA sont obtenues : 20%, 50%,
80%, 100%.

La caractérisation des nanoparticules se fait par spectroscopie infrarouge dans un premier

temps. L’apparition des bandes vibrationnelles caractéristiques de la thymine à 1710 cm-1 et 1660 cm-
1
(correspondant aux liaisons C=O), ainsi que la disparition de la bande vibrationnelle de l’azoture,
permet de confirmer le greffage de la thymine à la surface des nanoparticules. De plus l’analyse
infrarouge des nanoparticules recouvertes par l’azoture et le DPA à différents ratios montre une
évolution du ratio des bandes vibrationnelles N3/Fe-O qui augmente avec le ratio azoture/DPA utilisé
pour la fonctionnalisation. Cela confirme que la quantité d’azoture est contrôlé à la surface des
nanoparticules.

Une mesure DLS est ensuite effectué pour contrôler la stabilité des nanoparticules recouvertes
par la thymine à 100%. La mesure DLS effectué dans le THF indique un diamètre hydrodynamique
centré vers 100 nm indiquant une agrégation des nanoparticules en suspension. Cependant, lorsque
la mesure est réalisée dans le diméthylformamide (DMF), le diamètre hydrodynamique est centré à 12
nm indiquant que les nanoparticules ne présentent pas d’agrégation. Ainsi, le solvant joue un rôle
prépondérant dans la stabilité des nanoparticules en suspension. Les nanoparticules entièrement
recouvertes par la thymine présentent donc de nombreux groupements polaires à leurs surfaces qui
sont capable de créer des liaisons hydrogènes avec des atomes électronégatifs. Lorsque ces
nanoparticules sont en suspension dans un solvant apolaire ou à faible polarité, comme le THF, elles
ne réaliseront pas de liaison avec le solvant et les liaisons formées seront des liaisons thymine-thymine
qui permettra la formation d’un réseau de nanoparticules. En revanche, l’utilisation d’un solvant
permettant les interactions entre la thymine et le solvant ne permettra pas la formation de ces liaisons
thymine-thymine et les nanoparticules conserveront leurs stabilités, c’est le cas du DMF. Cette
agrégation contrôlée est parfaitement réversible, de plus, en fonction de la quantité de thymine à la
surface des nanoparticules, le solvant permettant une meilleure stabilité va varier. Ainsi, les
nanoparticules possédant des ratios thymine/DPA de 100%, 80% et 50% sont stable dans le DMF,
tandis que les nanoparticules recouvertes à 20% sont stables dans le chloroforme (CHCl3).

La surface des nanoparticules et des substrats étant fonctionnalisées par les bases azotées
complémentaires, l’assemblage par reconnaissance moléculaire peut être réalisé. Le processus
d’assemblage consiste à immerger les substrats fonctionnalisés dans une suspension de nanoparticules
à température ambiante. Les substrats sont ensuite retirés de la solution puis rincés au DMF et exposé
15 secondes au bain à ultrasons. Des tests préliminaires ont été effectués avec différents temps
d’immersions pour réaliser que le temps d’immersion idéal est de 30 minutes.

L’influence de la quantité de groupes à la surface des nanoparticules et des substrats est

étudiée en réalisant différents assemblages en variant la quantité d’adénine à la surface des SAMs et
la quantité de thymine à la surface des nanoparticules. Malgré une densité faible observé en SEM pour
l’ensemble des échantillons, le meilleur taux de couverture est celui réalisé avec une SAM-adénine à
80% immergée dans une suspension de nanoparticules recouvertes de thymine à 50%. Ainsi, ces
conditions sont utilisées pour le reste de l’étude pour limiter le π-stacking qui pourrait réduire la force
des interactions des liaisons hydrogènes.

Le second facteur qui est étudié est le solvant, puisqu’il intervient dans la stabilité des
nanoparticules. Les assemblages ont été réalisés avec des mélange de solvants à différentes polarités.

215
Ainsi les substrats 80% ont été immergés dans des suspension de nanoparticules recouvertes à 50%
dans un mélange DMF/CHCl3 avec différents ratios : [4 :1], [1 :1] et [1 :4]. Les images SEM (figure 5) de
ces assemblages montrent une densité de nanoparticules de 20% dans le cas des ratios [4 :1] et [1 :4],
alors qu’une densité de 53% est observée dans le cas du ratio [1 :1]. Ces résultats peuvent être reliés
à la stabilité en suspension des nanoparticules pour ces différents ratios. Dans le cas du ratio [4:1], les
nanoparticules sont parfaitement stable avec un diamètre hydrodynamique centré à 12.6 nm. Dans le
cas du ratio [1 :4] les nanoparticules sont complétement agrégées avec un diamètre hydrodynamique
supérieur à 1 μm. Tandis que pour le ratio [1 :1], les nanoparticules sont plutôt stables, mais présentent
un début d’agrégation avec un diamètre hydrodynamique de 21.3 nm. Cette étude permet de conclure
sur l’énergie d’interactions entre les bases azotés et le solvant. En effet, le solvant peut rentrer en
compétition avec la formation des liaisons entre l’adénine et la thymine. Cela explique le faible taux
de couverture dans le cas du ratio avec un grand volume de DMF [4 :1] qui empêche la reconnaissance
entre les bases complémentaires et favorise les liaisons avec le solvant. Au contraire, un ratio faible en
DMF [1 :4] va favoriser cette reconnaissance, néanmoins les nanoparticules se retrouvent agrégées
dans ce cas et ne seront pas disponible pour l’assemblage, ce qui explique également le faible taux de
couverture. Ainsi, la densité maximale est obtenue en trouvant un équilibre entre les énergies
d’interactions de la thymine avec le solvant et l’adénine pour un ratio DMF/CHCl3 [1 :1].

Figure 60. Images SEM de l'assemblage des nanoparticules fonctionnalisées par la thymine à 50% sur une SAM
fonctionnalisée par de l'adénine à 80% avec différents ratios DMF/CHCl3. a) [4:1], b) [1:1] et c) [1:4].

Le dernier facteur étudié qui influence la reconnaissance des bases azotées est la température.
Ainsi, un assemblage réalisé dans les conditions optimales (avec un taux de couverture de 57%) est
placé dans le DMF et chauffé à 80°C pendant 2 heures. Les images SEM montrent une diminution de
la densité qui passe à 18% de la densité théorique maximale. En effet l’énergie thermique permet de
briser les liaisons hydrogènes et donc de désorber les nanoparticules à la surface du substrat.

Pour conclure cette partie, en tirant avantage de la méthode de chimie « click », les
nanoparticules et les substrats ont pu être fonctionnalisés par des bases azotées complémentaires.
L’utilisation du micro-onde à grandement réduit le temps de réaction de plusieurs heures à quelques
minutes. L’assemblage des nanoparticules à pu être réalisé et être étudié en fonction de plusieurs
paramètres tel que le π-stacking, le solvant et la température. De plus, le contrôle de la distribution
spatiale des nanoparticules est effectué en contrôlant ces différents paramètres.

216
Chapitre IV – Reconnaissance biomoléculaires au travers de l’assemblage de nanoparticules d’oxyde
de fer déposées sur des films d’or

Dans cette dernière partie, l’assemblage des nanoparticules va être utilisé pour élaborer une
plateforme de détection basé sur la résonnance plasmon du substrat d’or (figure 6).

Figure 61. Représentation schématique du greffage d'un bio récepteur modifié avec un alcyne sur la surface de
l'assemblage de nanoparticules fonctionnalisées avec un azoture.

En utilisant la réaction de « click » pour moduler la structuration de l’assemblage, les

propriétés du substrat peuvent être modulés. De plus, les nanoparticules assemblées possèdent
toujours un groupement azoture en surface qui va pouvoir être utilisé pour greffer une molécule
réceptrice nécessaire pour la détection. Deux couples récepteur/analyte ont été étudié dans ce
manuscrit : biotine-streptavidine et iminosucre-α-mannosidase.

Biotine-Streptavidine

Pour élaborer une bio-plateforme la biotine doit être greffée à la surface de l’assemblage de
nanoparticules. Une biotine modifiée avec un groupement alcyne est donc synthétiser (biotin-CC). La
biotin-CC est greffer à la surface des nanoparticules de la même manière que les nanoparticules ont
été assemblées sur la surface d’or. Le substrat est immergé dans une solution d’assemblage avec 5 mg
de biotin-CC et est chauffé à reflux pendant 24 heures. L’échantillon est caractérisé par microscopie
pour s’assurer que le greffage de la biotine n’ait pas perturbé l’assemblage de nanoparticules. Les
images SEM et AFM montrent en effet que la densité reste inchangée et que les nanoparticules sont
toujours homogènement distribuées sur la surface du substrat. La mesure de l’angle de contact avant
et après greffage de la biotine montre un brusque changement d’hydrophobicité de la surface qui
passe d’une surface hydrophobe (100°) due à la présence des longues chaines alkylènes de l’acide
phosphonique, à une surface plus hydrophile (50°) expliqué par la présence de la biotine et des groupes
polaires contenus dans sa structure. L’analyse XPS permet de montrer la disparition des groupements
azotures à la surface des nanoparticules avec la disparition des contributions des liaisons N3 à 405.3

217
eV et 401.8 eV pour laisser place à deux contributions à 400.8 eV et 400.5 eV qui correspondent aux
liaisons N-C et N-H présentes dans la biotine. De plus une nouvelle contribution apparaît à 287.4 eV
qui correspond aux liaisons C=O présente dans le groupement amide de la biotine. L’analyse PM-IRRAS
permet de confirmer le remplacement total de l’azoture par la biotine avec la disparition de la bande
vibrationnelle caractéristique des azotures à 2100 cm-1 et l’apparition d’une bande large à 1660 cm-1
qui correspond aux liaisons C=O des groupements amides.

Le signal plasmon du substrat d’or étant très sensible aux variations d’indice de réfraction à sa
surface, la construction des différentes étapes de la bio-plateforme peut être suivi par mesure SPR. Le
tableau 2 résume la position du pic de résonance plasmon pour deux différentes longueurs d’onde
pour chaque étape de la construction de la bio-plateforme.

Tableau 2. Position angulaire des pics de résonnance plasmon pour chaque étape de l'élaboration de la plateforme de
détection pour deux longueurs d'onde de travail.

Wavelength Au SAM NPs@N3 Biotin

670 nm 68.627° 69.260° 74.029° 72.845°
785 nm 65.598° 65.960° 68.410° 67.832°

Après chaque étape, un shift vers grand angle est observé ce qui confirme le dépôt d’une
couche à la surface du substrat. Cependant pour le greffage de la biotine, un shift vers les faibles angles
est observé. Ce shift inattendu est le résultat de la désorption des nanoparticules greffées de façon
non covalente lors de l’étape de greffage de la biotine.

Pour étudier l’efficacité de la bio-plateforme de détection, l’adsorption de la streptavidine est

étudié en fonction de différents paramètres comme la densité de nanoparticules, la taille des
nanoparticules et la quantité de biotine. Deux configurations ont été utilisés pour étudier les
propriétés SPR : une interrogation spectrale à angle constant (74°) et une interrogation angulaire à
longueur d’onde fixe (785 nm). Le protocole de détection se fait en injectant une solution aqueuse de
streptavidine à 100 μg/mL sur la surface du substrat.

L’influence des nanoparticules est étudiée en réalisant l’adsorption de la streptavidine sur un

substrat avec et sans nanoparticules recouvert de biotine (figure 7).

218
 Figure 62. Mesures SPR à 785 nm avant (rouge) et après (vert) adsorption de la streptavidine sur un film mince sans
nanoparticules et avec nanoparticules d'oxyde de fer.

Le pic de résonnance est décalé de 0.303° dans le cas d’une surface sans nanoparticules et
0.582° dans le cas d’une surface recouverte à 100% de nanoparticules. Le même phénomène se produit
avec une interrogation spectrale où l’augmentation du signal est exacerbée avec un décalage lors de
la détection de la streptavidine de 3 nm pour une surface sans nanoparticules et 41 nm pour une
surface recouverte de nanoparticules. Il convient donc de comprendre le rôle des nanoparticules sur
l’augmentation de la sensibilité de détection, pour cela il faut découpler l’origine de l’évolution du
signal qui provient des propriétés intrinsèques optiques du capteur et les propriétés de chimie de
surface qu’apporte les nanoparticules.

La réponse du signal SPR est donné par la formule suivante :

œ_:
¤ = ¥. ç&. µ1 − G ]’ ¸

La réponse du capteur dépend directement de la variation d’indice de réfraction ç&, mais

également des paramètre optique m et ld qui sont le facteur de sensibilité et la longueur de pénétration
respectivement. La valeur de ld ne varie quasiment pas avec la présence des nanoparticules, en
revanche, le facteur m lui augmente avec la présence des nanoparticules. La mesure de la sensibilité
en fonction de l’angle d’incidence permet de calculer m avec et sans nanoparticules. La sensibilité
augmente légèrement avec la présence des nanoparticules en passant de 105,4 °/RIU à 115.7 °/RIU.
Néanmoins la valeur de m augmente significativement lors de l’interrogation spectrale, car la
sensibilité est directement liée à la longueur d’onde incidente. La présence des nanoparticules va
décaler le pic de résonance vers les hautes longueurs d’ondes et passera de 1580 nm/RIU à 5600
nm/RIU.

La présence des nanoparticules augmente donc la sensibilité du capteur en modifiant les

propriétés optiques intrinsèques de la surface d’or. Cependant, la présence des nanoparticules modifie
également la chimie de surface et peut modifier le nombre de groupements et leurs accessibilités pour

219
la détection de la streptavidine. Le nombre de streptavidine que la surface peut détecter peut être
estimé de façon géométrique en fonction de la taille de la protéine de streptavidine et de la surface
des nanoparticules accessible. Le nombre de streptavidine détecté dépend ainsi du diamètre des
nanoparticules et la distance inter-particules. Une étude a ainsi été menée pour optimiser la
plateforme de détection en faisant varier la taille et la densité des nanoparticules à la surface du
substrat.

Des échantillons ont donc été réalisés, le diamètre et la densité des nanoparticules à la surface
est résumé dans le tableau 3. Trois tailles différentes sont considérées 5.7 nm, 10.1 nm et 21.9 nm qui
seront appelées NP6, NP10 et NP22 respectivement.

Tableau 3. Diamètres et densités des nanoparticules assemblées sur la surface de la SAM.

Diameter (nm) Density (%)

a) 5.7 ± 0.9 nm 18%
b) 21.9 ± 1.6 nm 100%
c) 5.7 ± 0.9 nm 36%
d) 21.9 ± 1.6 nm 26%
e) 10.1 ± 1.1 nm 47%

Ces échantillons sont soumis à une injection de streptavidine et le décalage spectral est mesuré après
l’injection. Dans le cas des NP6, le décalage est de 11 nm pour une densité de 18% et 27 nm dans le
cas d’une densité de 36%. Pour ces deux densités, la distance inter-particules est très importante et
permet l’adsorption d’un maximum de streptavidine (7.5 streptavidine par nanoparticules). De plus le
facteur de sensibilité m est quasiment le même (2115 nm/RIU et 2220 nm/RIU). La seule différence
réside dans le nombre de nanoparticules qui est doublé à la surface et donc permet d’adsorber deux
fois plus de streptavidine. Dans le cas des NP22, les décalages sont de 42 nm pour une densité de 26%
et 53 nm pour une densité de 100%. Ici le facteur de sensibilité est nettement plus important pour
l’échantillon recouvert à 100% (8070 nm/RIU par rapport à 2460 nm/RIU). Cependant la détection de
streptavidine reste importante malgré le faible facteur de sensibilité. La distance inter-particule est
plus élevée dans le cas d’une densité plus faible et permet la détection de 52 streptavidine par
nanoparticule pour la densité de 26% et seulement 20 streptavidine par nanoparticule pour la densité
de 100%. Ainsi le nombre de streptavidine adsorbé est plus important dans le cas d’une faible densité
présentant une distance inter-particules élevée.

Une étude a également été menée pour déterminer l’influence de la quantité de biotine à la
surface des nanoparticules. Le greffage de la biotine à donc été diluée avec une autre molécule : l’hex-
1-yne. Des ratios biotin/hex-1-yne de 5%, 20%, 50%, 80% et 100% ont été réalisés. Après adsorption
de la streptavidine, le décalage spectral est similaire pour chacun des échantillons (25-27 nm). Ainsi,
même avec uniquement 5% de biotine à la surface des nanoparticules, la streptavidine est détectée.
Cela s’explique par l’augmentation de l’accessibilité à la surface des nanoparticules due à leurs rayons
de courbure.

Pour conclure, la sensibilité du facteur est augmentée par les propriétés optiques du substrat
qui sont modifiés par l’assemblage des nanoparticules, mais aussi par la quantité et l’accessibilité de
la streptavidine sur la surface du capteur.

Les mesures SPR permettent également, en étudiant les cinétiques d’adsorption, de

déterminer les processus de reconnaissances tel que les constantes d’association des couples

220
molécules réceptrices/analytes. Le sensorgramme décrit la mesure de la réponse du capteur en
fonction du temps. Il est utilisé pour déterminer les phases d’association, d’équilibre et de dissociation
des couples étudiés. Une étude a été réalisée en injectant une solution aqueuse de streptavidine à
différentes concentrations à la surface d’un substrat avec et sans nanoparticules couvert de biotine.
La valeur des pentes lors de la phase d’association ainsi que la valeur des décalages angulaires en
fonction de la concentration sont relevées. La constante d’association KA est déterminée de deux
manières distinctes en se basant sur la cinétique de réaction et en déterminer la constante effective
d’association et de dissociation, respectivement kon et koff (avec KA = kon/koff) et en se basant sur la
réponse angulaire en utilisant l’isotherme de Langmuir.

XŽV@ = XŽ& [•] + XŽ//

[•]
¤7 = ¤
[•] + <a s98
Ces deux méthodes nous permettent de trouver des valeurs de KA similaires pour la surface
recouvertes de nanoparticules et celles sans nanoparticules (KA = 10-6-10-7).

Pour compléter l’étude des contre-tests visant à prouver l’interaction spécifique de la

streptavidine sur la surface du substrat ont été réalisés. Ainsi la streptavidine a été injecté sur une
surface de nanoparticules ne possédant pas de groupements biotine ainsi que le test complémentaire
avec l’injection d’une protéine non spécifique, l’albumine sérique bovine (BSA), sur une surface de
nanoparticules recouverte de biotine. Les deux expériences ont montré un décalage angulaire
indiquant l’adsorption de la streptavidine. Cependant, la valeur de ces décalages ainsi que la cinétique
d’adsorption est plus faible que dans le cas de la reconnaissance entre biotine et streptavidine.

Iminosucre-α-mannosidase

La même étude a été réalisé en utilisant un couple iminosucre et une glycosidase, l’α-
mannosidase, qui est une enzyme impliquée dans le traitement de certaines maladies (maladie de
Gaucher, mucoviscidose). La surface de nanoparticules a donc été fonctionnalisées avec un iminosucre
modifié avec un groupement alcyne, puis caractérisées avec les différentes techniques d’analyses pour
confirmer la présence de l’iminosucre à la surface du substrat.

L’α-mannosidase est une enzyme qui possède deux sites actifs et par conséquent peut former
différents types d’accroche avec les iminosucres localisés à la surface des nanoparticules (figure 8).
Une configuration existe pour laquelle les deux sites actifs sont inhibés par les sucres et une autre pour
laquelle un seul site actif est inhibé laissant le second libre à la surface.

221
 Figure 63. Représentation schématique de deux configurations possibles pour l’adsorption de l’α-mannosidase à la
surface des nanoparticules.

La détection de la glycosidase est effectuée en injectant une solution d’α-mannosidase sur la

surface du substrat. L’adsorption de l’enzyme est suivie par un sensorgramme qui mesure en temps
réel l’angle de résonnance. De la même façon que pour la streptavidine, l’enzyme est injectée à
différentes concentrations afin de déterminer les constantes d’association sur une surface avec et sans
nanoparticules.

Les résultats dans le cas de l’α-mannosidase sont différents de ceux observés avec la
streptavidine. En effet, la constante d’association dans le cas d’une surface avec nanoparticules est
deux fois plus importante que dans le cas d’une surface plane (5,26.107 M-1 et 2,37.107 M-1
respectivement). Cela signifie que la présence des nanoparticules favorise l’adsorption de l’enzyme sur
la surface. Le processus impliqué dans cette adsorption n’est pas connu, mais la géométrie d’une
surface recouverte de sphères semble accélérer l’accroche de l’enzyme. De plus, les valeurs des
constantes d’associations ont été comparées aux études réalisées en solutions sur des couplages
multivalents entre l’enzyme et l’iminosucre et l’interaction semble plus forte dans le cas où l’enzyme
est immobilisée sur la surface.

D’autre part, une étude a été réalisée en injectant un substrat compétitif pour l’inhibition de
l’enzyme, le methyl α-D-mannopyranoside à 10 mM, après l’injection de l’α-mannosidase. Un décalage
de l’angle de résonance plasmon est observé après rinçage, ce qui implique que le substrat s’est
déposé sur la surface. L’absence de diminution de l’angle de résonance après injection du sucre
compétitif indique qu’il n’a pas désorbé l’enzyme de la surface et donc que l’association entre l’α-
mannosidase et l’iminosucre est trop forte. Cependant, le faible décalage de l’angle de résonance
indiquant l’adsorption du substrat permet de confirmer qu’il reste des sites actifs de l’enzyme
disponible à la surface. Ceci permet de conclure sur les configurations possibles de l’α-mannosidase à
la surface des nanoparticules en admettant que celle-ci se greffe en laissant des sites actifs disponibles.

222
 Une étude a également été réalisé pour tenter de régénérer la surface du capteur en désorbant
l’enzyme. Pour cela une solution d’acide phosphorique est injectée après l’adsorption de l’enzyme
dans le but de pourvoir casser les liaisons avec le sucre. Une solution aqueuse d’acide phosphorique à
0,02 M et à pH 2,5 est donc injecté sur un substrat non recouvert de nanoparticules. Après injection
de la solution acide, la position de l’angle de résonance plasmon retrouve sa valeur initiale avant
injection de l’’α-mannosidase. De plus, une injection de l’enzyme avec la même concentration effectué
par la suite donne une variation de l’angle de résonance identique à la première, ce qui indique que la
reconnaissance à la surface peut se faire. Cela indiquerait que le capteur est régénérable en pouvant
désorber l’enzyme et laisser la surface active pour une reconnaissance par la suite. L’expérience a
également été réalisée avec un substrat recouvert de nanoparticules. Cependant, la valeur de l’angle
de résonance est descendue en dessous de sa valeur initiale impliquant la disparition de matériau à la
surface. En effet, les nanoparticules d’oxyde de fer sont sensibles aux pH trop acides, et la solution
d’acide phosphorique à détériorer la qualité de la surface.

Enfin, une étude sur la reconnaissance spécifique a été réalisée de la même manière que pour
la streptavidine en injectant sur une surface de nanoparticules sans iminosucres l’α-mannosidase et
en injectant sur une surface de nanoparticules fonctionnalisées par l’iminosucre, une enzyme non
spécifique à la reconnaissance, la BSA. Les résultats sont similaires à l’étude précédente avec une
réponse du capteur avec des décalage angulaire et des vitesses d’association différentes que pour le
couple iminosucre/glycosidase.

Pour conclure, l’assemblage des nanoparticules constitue une plateforme de détection de

biomolécules qui peut être facilement fonctionnalisée avec différentes molécules réceptrices pour
permettre la reconnaissance d’une analyte. La sensibilité de ce capteur a été étudié avec et sans
nanoparticules pour s’apercevoir que ces dernières ont un rôle prépondérant dans la sensibilité de la
détection. Tout d’abord d’un point de vue optique, puisqu’elles modulent le plasmon de surface du
substrat sur lequel elles reposent, mais aussi d’un point de vue géométrique où la rugosité qu’elles
induisent permet de détecter un plus grand nombre de biomolécules. De plus, les mesures SPR
permettent de suivre l’adsorption de protéine ou d’enzyme en temps réel, ce qui a permis de définir
les constantes d’association pour les couples biotin/streptavidine et iminosucre/α-mannosidase.
Malgré la reconnaissance spécifique qui nécessite des expériences complémentaires, la plateforme a
montré qu’elle pouvait être régénérable avec une solution acide. Le système est donc prometteur mais
nécessite une optimisation et des caractérisations supplémentaires.

223
Conclusion générale

Le but de ce travail était de contrôler la structuration de l’assemblage de nanoparticules pour

contrôler les propriétés magnétiques collectives et individuelles de ces objets ainsi que leurs influences
sur les propriétés plasmoniques portées par le substrat sur lequel elles reposent. L’assemblage a été
réalisé au travers d’interactions moléculaires spécifiques.
La décomposition thermique a permis, en contrôlant les conditions opératoires, de synthétiser
des nanoparticules d’oxide de fer de différentes tailles. La caractérisation structurale a permis
d’identifier un cœur de magnétite avec une coquille de maghémite. Les propriétés magnétiques ont
montré que la température de blocage augmentait avec le diamètre des nanoparticules. Par la suite,
les nanoparticules ont été fonctionnalisées avec des groupements spécifique pour la réaction de
chimie « click ». Des substrats d’or ont également été fonctionnalisés avec les groupements
spécifiques complémentaires. L’assemblage a été réalisé par la réaction de chimie « click » et les
conditions d’assemblage ont été étudiées. La variation du temps de réaction a permis de contrôler la
distribution spatiale des nanoparticules sur le substrat d’or. Les interactions dipolaires magnétiques
entre les nanoparticules a ainsi pu être contrôlées et les propriétés magnétiques des films ont été
étudiées. Il a été montré que l’assemblage est dirigé en partie par ces interactions dipolaires
notamment dans le cas de nanoparticules avec un diamètre supérieur à 10 nm.
Une nouvelle stratégie d’assemblage a été reportée qui est basée sur la reconnaissance
spécifique entre des bases azotées. En prenant avantage de la réaction de chimie « click », les
nanoparticules et les substrats ont été fonctionnalisés avec des molécules de thymine et d’adénine.
De plus, l’utilisation de l’énergie micro-onde a permis de diminuer drastiquement le temps de réaction.
Le contrôle de la distribution spatiale des nanoparticules a été démontré en utilisant les différents
paramètres lors de la reconnaissance. Ainsi, le solvant, la température et la quantité de groupements
fonctionnels à la surface des nanoparticules et du substrats a été étudiés.
La dernière partie de ce travail a été d’utilisé ces assemblages de nanoparticules pour pouvoir
contrôler les propriétés plasmoniques des substrats d’or sur lesquelles elles ont été déposées. Une
« bio-plateforme » est ainsi réalisé en greffant des molécules réceptrices à la surface des films de
nanoparticules. La détection de la streptavidine ainsi que de l’α-mannosidase a été réalisée et
l’augmentation de la sensibilité avec la présence des nanoparticules a été démontrée. Les
nanoparticules modulent les propriétés optiques intrinsèques du substrat et donne une meilleure
accessibilité aux molécules cibles. De plus, l’utilisation de mesure SPR a permis d’étudier les processus
de reconnaissance entre les couples étudiés pour déterminer leurs constantes d’association.

224
 Mathias DOLCI

Design of magnetic iron oxide nanoparticle

assemblies supported onto gold thin films
for SPR biosensor applications

Résumé
La bio-détection de molécules reposant sur le phénomène de résonnance
plasmon permet de détecter des espèces en utilisant les propriétés optiques de films
métalliques. L’utilisation de ce type de capteurs nécessite néanmoins l’augmentation
de leurs performances afin de détecter des concentrations faibles d’analyte dans des
milieux complexes. L’assemblage de nanoparticules d’oxyde de fer sur des substrats
d’or, en utilisant des groupements complémentaires spécifiques via la méthode de
chimie « click », permet de contrôler leur distribution spatiale à la surface du substrat.
Les propriétés magnétiques portées par les nanoparticules sont ainsi étudiées en
fonction de leurs distances inter-particules ainsi que de leurs tailles. Par ailleurs, le
plasmon de surface du substrat étant directement influencé par l’assemblage des
nanoparticules, il sera possible de contrôler la sensibilité du capteur pour étudier la
détection de différentes biomolécules impliquées dans des processus biologiques. La
présence des nanoparticules augmente les propriétés optiques intrinsèques de la
surface du substrat et la géométrie de l’assemblage permet d’augmenter la quantité
de biomolécules détectées.

Mots-clés : Nanoparticules, Biocapteurs, Plasmon de Surface, Magnétisme,

Fonctionnalisation de surface, Détection Moléculaire, Biomolécules.

Résumé en anglais
Biomolecular detection based on the surface plasmon resonance phenomenon
allow detecting species by using the optics properties of metallic thin films. This kind
of biosensors require the increase of their performances in order to detect low
concentration analyte in complex medium. The assembly of iron oxide nanoparticles
on gold substrates by using specific complementary groups via the “click” chemistry
technique allows controlling their spatial distribution on the substrate surface. The
magnetic properties carried by the nanoparticles are studied as function of their inter-
particle distances and their sizes. Moreover, the surface plasmon of the substrate is
directly influenced by the nanoparticle assembly and the control of the sensor
sensitivity will be possible in order to study the detection of different biomolecules
implies in biological processes. The presence of nanoparticles increases the intrinsic
optical properties at the substrate surface and the geometry of the assembly allow
increasing the number of biomolecules detected.

Key-words : Nanoparticles, Biosensors, Surface Plasmon, Magnetism, Surface

Functionalization, Molecular Recognition, Biomolecules.

225