Local direct and indirect reduction of electrografted ...

Page 1

Ls

FL

a

ARRAA

KAAPSP

1

smoptbao

tacumL

cbrnsfi

0d

Electrochimica Acta 54 (2009) 5127–5136

Contents lists available at ScienceDirect

Electrochimica Acta

journa l homepage: www.e lsev ier .com/ locate /e lec tac ta

ocal direct and indirect reduction of electrografted aryldiazonium/goldurfaces for polymer brushes patterning

anny Hauquier, Tarik Matrab, Frédéric Kanoufi, Catherine Combellas ∗

aboratoire Environnement et Chimie Analytique, CNRS UMR7121, ESPCI, 10 rue Vauquelin, 75231 Paris Cedex 05, France

r t i c l e i n f o

rticle history:eceived 7 November 2008eceived in revised form 16 January 2009ccepted 22 January 2009

a b s t r a c t

The patterning of conductive substrates by polymer brushes may be achieved by using successively scan-ning electrochemical microscopy (SECM) and atom transfer radical polymerization (ATRP). After thesurface functionalization by a brominated aryldiazonium initiator, SECM allows the local reduction atthe micrometer scale of the initiator grafted layer. Different channels sizes involved in charge transport

vailable online 31 January 2009

eywords:ryldiazoniumTRPolymer brush

within the initiator layers are evidenced by combining SECM, CV and observation of the aryl-grafted layertransformation. ATRP is performed on the SECM patterned substrate. Inside the pattern, the lower densityof initiator decreases the polymer thickness. The pattern resolution is enhanced when the direct modeof the SECM is used instead of the mediated indirect mode.

© 2009 Elsevier Ltd. All rights reserved.

ECMatterning

. Introduction

Much interest is currently directed toward new ways to modifyurfaces of solid substrates. In this respect, surface grafting of poly-er chains is a much effective method. Grafting from surfaces well

rganized polymeric systems with controlled length is especiallyromising. Such polymer brushes are particularly interesting sincehey can exhibit large and reversible deformations from stretchedrush to collapsed mushroom when submitted to a stimulus suchs a change of pH or ionic strength. These ordered macromolecularbjects can be obtained from radical polymerization.

The development of micropatterned polymer structures consti-utes an attractive approach for designing novel smart surfaces,ctuators, microdevices or chemical/biological sensors. The prin-ipal methods for producing such micropatterned surfacesse electron-beam lithography, [1–5] photolithography, [6–13]icrocontact printing, [14] nanoimprinting, inkjet printing, [15]

angmuir–Blodgett or dip-pen lithography.In a previous work, we have used the scanning electrochemi-

al microscopy (SECM) to pattern insulating surfaces with polymerrushes of controlled dimension obtained from atom transfer

adical polymerization (ATRP). [16] The combination of both tech-iques, SECM and ATRP, is an innovative lithography to design smarturfaces. This process was used to grow from Si or glass substrateslms of various grafting densities and therefore various thicknesses

∗ Corresponding author.E-mail address: [email protected] (C. Combellas).

013-4686/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.oi:10.1016/j.electacta.2009.01.059

or elongations in the locally etched regions. This lithography proce-dure can be successfully used to design either empty microdomainssurrounded by polymer brushes or isolated microdomains of poly-mer brushes.

Here, the same microelectrochemical strategy is generalized toconducting substrates such as gold surfaces. The objective is topattern insulating or conductive surfaces with polymer structuresof controlled morphologies and to provide an interesting alterna-tive to lithography. Styrene (S) and glycidyl methacrylate (GMA)were chosen for this study. PGMA is a potential surface linker forbiomolecules and has promising applications in advanced biotech-nology, such as DNA separation, targeted drug delivery, enzymeimmobilization, and immunological assay [17–19] because of theease in the conversion of epoxide groups into a variety of functionalgroups, such as –OH, –NH2, and –COOH. On the other hand, PS waschosen because of its hydrophobic character.

2. Experimental

2.1. Surface characterization

The thickness of the organic monolayer was estimated using aSentech SE 400 ellipsometer with a He–Ne laser and an angle ofincidence of 70◦. The refractive index of the initiator layer was set

at n = 1.46, and the refractive and optical indexes n and k for the goldsubstrate were obtained from the clean bare surfaces. Uncertainty±1 Å.

The morphologies of the pattern were determined using a FogaleNanotech Microsurf 3D optical profilometer or were imaged using

Page 2

5 imica

aismimIcetcrdoifc

Mto

2

C1cn

f3(br

2

e

l2o2enmis

faestdfitfAsiwtr

128 F. Hauquier et al. / Electroch

n INCAx-sight scanning electron microscope (SEM) from Oxfordnstruments. The amount of polymer grafted on the patternedurface was estimated by interferometric microscopy. The measure-ent is based on the non-contact acquisition of an interferometric

mage of a surface topography from the modification of interfero-etric fringes pattern upon the topography of a reflecting surface.

nterference fringes are obtained from the recombination on a CCDamera of two parent half-beams, a first beam is reflected on a ref-rence mirror while the second analyzing beam is reflected on theested surface. On a reflecting surface covered with thin polymeroatings, the analyzing beam crosses the polymer layer before itseflection on the gold surface. Therefore the path difference intro-uced by the polymer crossing is 2npt, where np is the polymerptical index (np = 1.46) and t is the polymer film thickness. Thenterferometric image of a patterned surface can then be trans-ormed into an image of the differential thickness of the polymeroating over the Au surface, as presented in Figure S1.

AFM images were recorded with a PicoSPM II microscope fromolecular Imaging in the tapping mode under ambient condi-

ions. We used silicon nitride cantilevers with a cantilever lengthf 125 �m.

.2. Electrochemical treatment of gold wafers

The synthesis of the starting diazonium salt, +N2–C6H4–H(CH3)–Br, was previously described [20]. Gold plaques were–2 cm2 pieces of a 4 in. diameter × 500 �m thickness silicon waferoated by a 1000 Å gold layer (Aldrich). Gold orientation is nomi-ally highly polycrystalline with 〈1 1 1〉 orientation.

Electrochemical grafting of this diazonium salt onto the sur-ace was achieved by chronoamperometry for 300 s, at a potential00 mV more negative than the diazonium salt peak potentialmeasured on carbon). At this potential, the gold surface should note covered by an oxide layer. The gold wafers were then thoroughlyinsed under sonication in deaerated acetone.

.3. Patterning

Patterning on the layer was done with a home-made scanninglectrochemical microscope using a CHI potentiostat.

In the feedback mode, typically, reduction of the initiatorayer was achieved in a solution of DMF containing NBu4BF4 and,2′-dipyridyl as the mediator, by applying a reductive potentialf −2.1 V/SCE to a platinum working microelectrode (diameter:5 �m). The counter electrode was a platinum wire, and the ref-rence electrode was Ag/AgCl. The whole device was kept underitrogen in a polyethylene glovebag during the experiment. Theicroelectrode-substrate positioning was obtained from the exper-

mental approach curve of ferrocyanide onto the aryl-grafted goldurface.

In the direct mode in a 2-electrode configuration, the gold sur-ace modified by the initiator layer is used as the working electrodend the microelectrode is used as the counter and the referencelectrodes. The reduction of the initiator layer was achieved in aolution of DMF containing NBu4BF4 (5 mM), by applying a reduc-ive potential of −2.6 V to the gold surface. This potential wasetermined by cyclic voltammetry in the same 2-electrode con-guration as the onset of the gold covered substrate reduction. Inhis mode, the microelectrode was positioned at a close distancerom the substrate using the “negative feedback” obtained fromC-current-SECM [21]. It consists of applying a 10 mV amplitude

inusoidal oscillation (E = 0 V) at the microelectrode and record-ng the AC tip current or the AC tip impedance. The tip is then

ithdrawn from a known distance from the substrate. The exactip–substrate distance is then obtained after the experiment byecording the SECM approach curve in the feedback mode with an

Acta 54 (2009) 5127–5136

acetonitrile solution of ferrocene. The whole device was kept undernitrogen in a polyethylene glovebag during the experiment.

2.4. Polymerization

Glycidyl methacrylate, GMA, and styrene, S, (Aldrich) were dis-tilled prior to polymerization and stored at 4 ◦C. CuBr, CuBr2 and2,2′-dipyridyl (Aldrich) were used as received. Surface-initiatedATRP was undertaken on the bromide derivated surface followingthe procedure described by Yu et al. [22] for GMA and Hikita et al.[23] for styrene.

2.4.1. Surface-initiated ATRP of GMAThe detailed procedure for preparing the polymer brushes

is the following. First, a 100 mL Schlenk flask equipped witha magnetic stirring bar and sealed with a rubber septum wasdeoxygenated under vacuum followed by back-filling with nitro-gen for three times. The CuBr (50 mg, 0.35 mmol) and CuBr2(19.5 mg, 0.087 mmol) powders, and the initiator grafted gold waferwere introduced into the flask under a nitrogen flow. A mixturecontaining GMA (9.5 mL, 70 mmol), and 2,2′-dipyridyl (137.5 mg,0.87 mmol), previously degassed, was added to the polymerizationflask using a double-tipped needle under a nitrogen flow. The flaskwas placed at room temperature for several hours. The polymer-ization was stopped by cooling and opening the flask in order toexpose the catalyst to air. The gold wafer-PGMA hybrids were thor-oughly rinsed in dichloromethane under sonication for five periodsof 5 min.

2.4.2. Surface-initiated ATRP of SCuBr (390 mg, 2.7 mmol) and a piece of the initiator grafted gold

wafer were placed into a 100 mL Schlenk flask equipped with amagnetic stirring bar and sealed with a rubber septum and deoxy-genated by a nitrogen flow. 2,2′-dipyridyl (1.17 g, 7.5 mmol) wasplaced into the two-neck round-bottom flask, and the flask wasevacuated and backfilled with nitrogen. To this flask, 10 mL oftoluene was added and this solution was stirred for 20 min undernitrogen. The resulting solution was transferred through a cannulato the Schlenk flask. In a second two-neck round-bottom flask,styrene (24 mL, 210 mmol) was deoxygenated by a nitrogen flow.The monomer solution was transferred through a cannular, andthe flask was held at 110 ◦C. After several hours, the substrate wasremoved from the flask, washed with dichloromethane, and soni-cated in toluene and dichloromethane.

3. Results and discussion

3.1. ATRP polymer growth from Au surfaces

First, the electrochemical reduction of a bromoaryl diazoniumsalt was used to prepare a brominated initiator layer on the goldsurface. Immobilization of the initiator by such electrograftingwas preferred to the use of alkanethiol self-assembled monolay-ers because it leads to more stable layers when submitted to highlyreductive or oxidative potentials; it also resists to sonication andtime [24,25]. Indeed, the electrografting of electrodes by diazoniumsalts provides the covalent attachment of an organic layer onto theelectrode surface. The covalent bond between the substrate and theorganic layer has been demonstrated both experimentally [26–28]and theoretically [29]. This is a particularly important feature inthe preparation of covalently bonded polymer brushes on the sur-

face, as a uniform and dense layer of initiator covalently tetheredto the surface is compulsory. Especially, the polymer brush is morestable over a spin-coated polymer layer with regard to solvent treat-ment and harsh conditions such as high temperature, because ofthe covalent bond between the polymer and the substrate. In this

Page 3

F. Hauquier et al. / Electrochimica

Fe

rpc

wcfrAcTcwpmkfo

mpbabpi

in Fig. 2. The curves were fitted by the general analytical expressions

Sm

ig. 1. Evolution of the film thickness t with polymerization time deduced fromllipsometric measurements (♦: PGMA, �: PS).

espect, aromatic moieties derived from diazonium ions have beenroposed to initiate polymer growths from iron [20,30] or carbona-eous surfaces (diamond, nanotubes. . .) [31–33].

Radical polymerization shows many advantages in comparisonith other types of polymerization; typical examples are its appli-

ability to various kinds of monomers and the absence of needor strict purification processes. However, radical species are soeactive that care should be taken for the control of the reaction.tom transfer radical polymerization minimizes the radical con-entration, which results in a precise chain growth control [34–40].herefore, the method allows to prepare various kinds of blockopolymers such as diblock, triblock, and gradient copolymers withell-controlled block lengths. There have been many reports for thereparation of “grafted from” type polymer brushes with the ATRPethod, that exploit its living nature and compatibility with various

inds of monomers. They show that it is possible to control the sur-ace properties systematically by controlling the initiator densityn the substrates or by selecting the monomer.

ATRP is suitable for the polymerization of acrylic and vinyliconomers such as styrene (S) or glycidyl methacrylate (GMA). The

resence of a grafted polymer layer on the surface was ascertainedy ellipsometry after the substrate has been washed exhaustively

nd sonicated with several solvents. The thickness of the polymerrushes, grown on the surface, was followed as a function of theolymerization time in the presence of the two monomers as shown

n Fig. 1.

cheme 1. Preparation of micropatterned brushes on gold substrate by scanning electroodes.

Acta 54 (2009) 5127–5136 5129

A linear increase in the grafted layer thickness with the polymer-ization time is observed for both polymers. These results indicatethat the process of surface-initiated ATRP is controlled in the twocases. Furthermore, the polymerization rate varies with the natureof the monomer. Surface-initiated polymerization of styrene givesonly a 16 nm thick PS film after 6 h, compared to the 28 nm thickfilm obtained when the graft polymerization was carried out inthe presence of GMA. This marked difference can be attributed tothe already reported difference between the two polymerizationkinetics [22,23].

3.2. SECM patterning of Au surfaces with polymers

Following the procedure described previously, we have pat-terned surfaces with PS and PGMA polymer brushes. Differentstrategies have been proposed to pattern surfaces with polymerbrushes; they are mainly based on the use of light or beam irradi-ation of masked surfaces [1–5]. Following the procedure describedpreviously [16] but using a bromoaryl moiety as the initiator forATRP [20,30–33], we propose an electrochemical alternative basedon the use of scanning electrochemical microscopy in the feedback(indirect) mode or in the direct mode. Scheme 1 outlines the syn-thesis pathway for the preparation of micropatterned brushes ofPS and PGMA polymers on a gold surface. As previously reported,the microelectrode of the SECM will be used as an electrochem-ical eraser of the initiator layer, in order to impede ATRP in thereduced regions. In this way, it will be possible to create structuresof polymer brushes.

3.2.1. SECM approach curves of electrografted Au surfacesAs the pattern formation depends on the local electrochemical

etching of the initiator layer, we first dealt with the SECM character-ization of an electrografted 1-bromoethylaryl layer obtained fromthe reduction of the corresponding diazonium ion. To do so, SECMexperiments were achieved above the initiator layer covered goldsurface. The approach curves obtained for ferrocyanide, tereph-thalonitrile, 2,2′-dipyridyl, p-tolunitrile and ferrocene are displayed

given for an irreversible charge transfer at the substrate [41,42].In an aqueous medium, in the presence of potassium ferro-

cyanide, a decrease of the faradic current with the distance Lis observed. Nevertheless, the approach curve deviates from the

chemical microscopy (SECM) in the indirect (top left) and the direct (bottom left)

Page 4

5130 F. Hauquier et al. / Electrochimica

Fig. 2. SECM approach curves obtained at a Pt disk UME tip (radiusa = 12.5 �m) on an initiator layer covered Au substrate in the presence of: ( )potassium ferrocyanide (5 × 10−3 mol L−1 in H2O + 0.1 mol L−1 KCl); ( ) tereph-thalonitrile (5 × 10−2 mol L−1 in DMF + 0.1 mol L−1 NBu4BF4); ( ) 2,2′-dipyridyl(5 × 10−2 mol L−1 in DMF + 0.1 mol L−1 NBu4BF4); ( ) ferrocene (5 × 10−3 mol L−1 inDNia

edmtdtfibbdgeoTinafTtfp

ateah

itssooTdorlcl

Fig. 4 (×). For a similar distance between the microelectrode and thesubstrate, the pattern dimensions are smaller for the conductingsubstrate.

In previous works, we have attributed the patterns widthevolution to the diffusive-convective transport of the tip-

MF + 0.1 mol L−1 NBu4BF4); ( ) p-tolunitrile (5 × 10−2 mol L−1 in DMF + 0.1 mol L−1

Bu4BF4). The solid lines are the theoretical curves for finite electron-transfer kinet-cs. The dashed and dotted lines are the theoretical curves for a conducting substratend for a totally insulating substrate respectively.

xpected curve for a totally insulating substrate. The electrore-uction of the diazonium salt on gold surfaces gives a polyarylultilayer (2–3 nm from ellipsometry) that is likely more complex

han the classical self-assembled monolayers [43,44]. It is also lessensely packed than classical SAMs, as it does not block the electronransfer (ET) to potassium ferricyanide as kapp, obtained from thet of SECM approach curves, reaches a value of 5 × 10−4 cm/s. Thelocking of the electrode by the electrografted layer was confirmedy a cyclic voltammetry (CV) study. Indeed, the CVs for the oxi-ation of ferrocyanide at naked and electrografted Au surfaces areiven in Figure S2. The potential dependency of the apparent het-rogeneous ET, khetCV, at the naked and grafted electrodes can bebtained from mathematical transformation of the CV curves intoafel plots (given in Figure S3). We followed the procedure detailedn Refs. [45–47]. The values of the apparent standard heteroge-eous ET, k0

hetCV, are obtained by extrapolation of such Tafel plotst E = E0. Once grafted by the polyaryl multilayer, the k0

hetCV valueor ferrocyanide oxidation has decreased by a factor of 3–5 × 103.his decrease is not high enough to account for tunnelling electronransfer along a distance as long as 2 nm. The low electron trans-er rate would rather indicate ET at defects sites, as was alreadyroposed for other similarly electrografted systems [45–49].

As already observed too, for the organic redox probes, thepproach curve was always higher, indicating a higher regenera-ion of the redox probe at the electrografted layer. The values of thelectron transfer rates at the substrate depend on the redox probend also on the substrate surface, but they are always 5–10 timesigher than the value obtained with the aqueous ferrocyanide.

The difference between aqueous and organic probes can benterpreted by either (i) permeation of the organic probes withinhe hydrophobic layer followed by an ET at the surface or (ii) ET atmaller pinholes. We make here a distinction between large defectites and pinholes, and define the former ones as pores or defectsf large, �m or sub-�m, dimension (macropores) than the latternes that would rather characterize smaller meso- or nano-pores.his distinction is important as different contributions from thoseefects could be expected according to the solvation properties

f the electrochemical solvent used. Indeed, transport of aqueousedox probes into macropores or �m scale defects in the organicayer could be expected, while such species would not, owing toapillarity restrictions (related to the hydrophobicity of the organicayer), explore within nanopores or pinholes in a 2 nm thick organic

Acta 54 (2009) 5127–5136

film. The organic redox probe could then explore both larger andsmaller pores, and the rate of ET at the larger pores would cor-respond to the value observed for the aqueous redox probe (herea value of 5 × 10−4 cm/s was obtained from the SECM approachcurve).

The electron transfer rates depend on the layer preparation andon the position at the surface, however we detected from SECMapproach curves that they generally fall within, in cm/s, 0.002 and0.004 for terephthalonitrile, 0.006 and 0.01 for 2,2′-dipyridyl and0.013 and 0.033 for p-tolunitrile. The trend is that the more reducingthe anion radical, the faster the ET rate at the layer. Finally, theoxidizing approach curve performed with ferrocene as the redoxprobe provides, with a rate of 0.012 cm/s, an intermediate situationbetween the 2,2′-dipyridyl and the p-tolunitrile.

3.2.2. Patterning in the feedback mode of the SECMIn the feedback mode of the SECM, the initiator layer is scanned

with a microelectrode electrogenerating the reduced form of aredox mediator M. This step leads to the local reduction of the C–Brbond and its conversion into a C–H bond. In that way, the initiationstep is deactivated in the reduced area. Following that, the wholesurface is submitted to the polymerization; this process provides alocal etching of the surface.

In previous works, water condensation was used to estimatepattern dimensions on silicon or glass substrates [16]. Here, thehydrophobicity contrast is not so high and several techniques, suchas SEM or interferometric microscopy, were used to estimate thepattern width, w.

The SEM image (Fig. 3) displays the gold surface covered by aPGMA film that exhibits several clear lines, which are consistentwith the patterns. The larger white protrusions observed at one endof each etched line are due to a longer reduction time at this place;indeed, the SECM tip positioning was deduced from the approachcurve performed there. The morphology of the patterns is governedby the tip velocities from 200 to 50 �m/s. For both polymers, theevolution of the pattern width as a function of the tip velocity v andthe microelectrode/substrate separation was investigated (Fig. 4).For all distances between the substrate and the microelectrode, thepattern width decreases as the tip velocity increases. As a compar-ison, data obtained for an insulating substrate are also reported in

Fig. 3. SEM image of the patterned PGMA brush.

Page 5

F. Hauquier et al. / Electrochimica Acta 54 (2009) 5127–5136 5131

Fig. 4. Evolution of the pattern width w as a function of the tip velocity v andthe microelectrode/substrate distance d (with L = d/a). Local reduction in the indi-rNs(

essctdlgce

dstcvbapotop

ftsdvtbtpiw

bmT1sev

Fig. 5. Evolution of the ratio �e/t as a function of the tip velocity. Local reduction inindirect mode in the presence of 2,2′-dipyridyl (5 × 10−2 mol L−1 in DMF + 0.1 mol L−1

ect mode in the presence of 2,2′-dipyridyl (5 × 10−2 mol L−1 in DMF + 0.1 mol L−1

Bu4BF4): L = 1; ( ), ( ) and ( ) L = 0.7; ( ) L = 0.5; ( ) L = 0.45; ( ) insulatingubstrate at L = 0.5; Local reduction in the direct mode (DMF + 5 mmol L−1 NBu4BF4)

) L = 1.6. L values are given with a ±0.08 (±1 �m) confidence.

lectrogenerated reagent (here the reducer M•−) at the substrateurface. The difference of the pattern width on Au and insulatingurfaces may be due to the evolution of the diffusion cone with theonductive nature of the substrate. Indeed, partial regeneration ofhe redox mediator, M, is expected at the surface of a partially con-ucting substrate, such as the gold surface covered by an organic

ayer. This feedback impedes the lateral expansion of the electro-enerated reagent over the conductive substrate and the higher theonductivity of the substrate, the lower the lateral expansion of thelectrogenerated reagent.

For a given tip scan rate, when the microelectrode/substrateistance decreases, the pattern width decreases. Owing to the pos-ible regeneration of M at the substrate, it seems surprising thathe expansion of the tip-electrogenerated reagent varies so signifi-antly within the explored tip/substrate separation distance, L. Theariations observed could indicate that at shorter L the substrateehaves as a partially conducting substrate (focusing the reagentround the microelectrode disk) while at higher L it presents a moreronounced insulating character (allowing the reagent expansionver a wider region). This situation could be due to the influence ofhe solution ohmic drop onto the (unbiased) substrate kinetics [50]r the intervention of kinetic limitations related to lateral chargeropagation within the layer [51–53].

The chemical content of the pattern was investigated by inter-erometric microscopy that allows to follow the pattern depth andhen to check whether it contains polymeric material or not. Fig. 5hows the evolution of the ratio �e/t, where �e and t are the patternepth and the film thickness, respectively, as a function of the tipelocity. We can see that there is not much effect of the velocity onhis ratio and that the mean �e/t value is around 0.4. If all the C–Bronds present at the surface were totally reduced into C–H bonds,he value would aim toward 1, as previously observed for the localatterning of insulating silanized glass or Si surfaces [16]. However,

n our case, the pattern depth is lower than the polymer thickness,hich suggests that there is some polymer inside the pattern.

In order to ascertain this suggestion, AFM analysis of the polymerrushes modified surfaces was achieved. Fig. 6A shows the initialorphology of a gold surface modified by bromoaryl diazonium.

he surface is smooth and its average roughness is of the order of5 Å. The enlarged image of this surface, at 500 nm × 500 nm scale,hows the presence of gold terraces as shown in Fig. 6A inset. How-ver, the image presents a slight difference by comparison with theirgin substrate at the same scale since small ∼20–30 nm diameter

NBu4BF4): ( ) L = 0.65, ( ) and ( ) L = 0.7, ( ) L = 0.5; terephthalonitrile(5 × 10−2 mol L−1 in DMF + 0.1 mol L−1 NBu4BF4) ( ) L = 0.5; Local reduction in thedirect mode (DMF + 5 mmol L−1 NBu4BF4) ( ) L = 1.6.

nodules are observed on the grafted gold surface. The presence ofsuch topography has been attributed to the grafting of aryl groupsonto the surface [48,54–56]. After local reduction and ATR polymer-ization, the AFM tip was placed outside the patterned area (Fig. 6B).The morphology of the initiator layer has completely disappearedand has been replaced by large spheroids with diameters around150 nm. Consequently, the average roughness increased to 14 nm.These dramatic changes in the morphology reflect the growth ofpolymer brushes onto the initiator layer. Next to that, the AFM tipwas removed and placed inside the patterned area (Fig. 6C). In thisarea, the initial morphology of the gold surface modified by bro-moaryl groups is not totally recovered and the same large spheroids,which are consistent with polymer brushes, are persistent. Theseresults confirm the presence of some polymer inside the pattern,which means that the reduction of the initiator is quite inefficient.However, areas without trace of polymer inside the pattern mayalso be observed. The comparison of the different cross-sections(Fig. 6D) reveals that the roughness inside this area is the roughnessof a surface only covered by aryl groups derived from diazoniumsalts. In this area whose width varies from 1 to 3 �m diameter, theC–Br bonds are totally reduced.

Even if there is still some polymer inside the reduced area, localmodification is still visible. This phenomenon can be explainedby the behavior of the polymer brushes. Indeed, this orderedmacromolecular structure exhibits large and reversible deforma-tions from a stretched brush to a collapsed mushroom. Outsidethe pattern, there is a uniform and dense layer of initiator cova-lently tethered to the surface. This dense initiator layer leads tothe formation of a stretched brush layer. On the contrary, insidethe local modification, the initiator layer is partially reduced andthen partially debrominated. The initiator density decreases butis still sufficient to initiate an ATR polymerization. In that way, alower density of polymer brushes facilitates its spreading over thesurface and therefore yields an apparent thinner film. The differ-ence between the more or less elongated states explains this localpatterning.

The presence of some polymer inside the patterns then indi-cates that the C–Br reduction of the initiator layer is not complete.Two hypotheses may explain this behavior: (i) the anion radical of

the redox mediator is not enough reducing to reduce completelythe initiator layer and to prevent totally the polymerization in thereduced area, or (ii) the reduction is incomplete because of trans-port limitations of the reducer within the initiator layer. Severalexperiments were achieved to test these assumptions.

Page 6

5132 F. Hauquier et al. / Electrochimica Acta 54 (2009) 5127–5136

F er grap rfaces

3i

iiloEtvrsfitgc1omoe[

M

ig. 6. Tapping-mode (5 �m × 5 �m) AFM images for gold modified surface (A) aftattern (C) after ATR polymerization inside pattern and (D) cross-sections of the su

.2.3. Electrochemical reduction of bromoethylbenzene—nfluence of the reductive strength?

As a preliminary test, we have investigated the reductive behav-or of a solution of 1-bromoethylbenzene, 1, in order to mimic,n solution, the reductive behavior of the ATRP initiator immobi-ized on the Au surface. The cyclic voltammograms of the reductionf a DMF solution of 1 present a large and irreversible peak,p − Ep/2 = 145 mV, located at −1.67 V vs SCE for an electrode poten-ial scan rate of 0.1 V/s. The shape, location and peak potentialariations with the potential scan rate clearly indicate that theeaction is under the kinetic control of a slow electron transfertep characterized by a transfer coefficient of ˛ ∼0.32. As observedor analogous alkylbromides and arylalkylbromides, the reductions kinetically controlled by the first dissociative electron transferhat corresponds to the debromination of 1 [57,58]. The homo-eneous reduction of 1 by 2,2′-dipyridyl could not be studied byyclic voltammetry as 2,2′-dipyridyl is more difficult to reduce than. We have then performed the redox catalysis of 1 by a morexidizing redox mediator such as phthalonitrile. The cyclic voltam-ograms of these redox catalysis experiments (for 2 concentrations

f redox mediator and 2 concentrations of 1) coincide with the gen-ral scheme of homogeneous reduction of alkylaryl halides (RX):59,60]

+ e = M•− (1)

fting of the initiator layer (local zoom inset), (B) after ATR polymerization outside.

M•− + RXk2−→M + R• + X− (2)

M•− + R• k3−→MR− (3)

M•− + R• k4−→M + R− (4)

where the radical R• can either couple with M•− via a radical–radicalcoupling reaction (3) or get reduced by ET with M•− (4). In a typ-ical redox catalysis experiment, the reduction peak currents ofthe redox mediator in the absence and in the presence of the RXmoieties are compared (see the example in Figure S4). A currentincrease is expected upon RX addition as reaction (2) regeneratesM at the electrode surface. The current increase depends on theelectrode potential scan rate, the excess factor [RX]/[M] and onthe kinetic parameters k2 and k3/k4. The simulation of the redoxcatalysis experiments allows the estimation of the first homoge-neous dissociative electron transfer rate, k2, and of the competitionbetween the two ensuing reactions for the fate of the radical.This is obtained from the comparison of experimental data withthe appropriate theoretical working curves obtained for exam-

ple from Digisim®, an example is given in Figure S5. Assumingthe radical–radical coupling reaction at a rate of, k3 = 107 M−1 s−1,k2 = 8 × 105 M−1 s−1 and k4 = 6 × 108 M−1 s−1 (average of 10 values)are obtained. These values fall in the range of what could beexpected from ET theories. Indeed, the value of k2 for the redox

Page 7

imica

cwdpb�E

tSEv0aeftrad

rit

rptdlimggrecdortnttwc

3

dlrcisi

nefaDetosa

mp np red

F. Hauquier et al. / Electroch

atalysis of 1 by the anion radical of phthalonitrile is in agreementith the heterogeneous dissociative reduction of 1 and Savéantescription of homogeneous dissociative ET [61]. Briefly, from theeak reduction potential of 1 at 0.1 V/s, one can compare the C–Brond dissociation energy of 1 to that of C–Br in PhCH2Br (�D = −2/3Ep = −20 mV) and obtain the E0 of the 1st dissociative ET of 1,

0 = EPhCH2Br − �D = 0.61 V vs SCE [62]. From the knowledge ofhe E0, the value of the homogeneous rate constant k2 would fit theavéant homogeneous dissociative ET theory, if the homogeneousT has an intrinsic activation barrier of 0.73 eV in agreement withalues obtained for alkyl halides and from the predicted value of.74 eV. The value of the 2nd homogeneous ET rate constant, k3, islso in good agreement with data from the reduction dynamic of thethylphenyl radical [63]. From this homogeneous investigation androm Savéant dissociative homogeneous ET, one can then predicthat the homogeneous rate constants for 1 reduction by the anionadicals of 2,2′-dipyridyl and p-tolunitrile are respectively 2 × 108

nd 3 × 109 M−1 s−1, meaning that the reduction of 1 is close toiffusion control.

It is then expected that the surface bound ATRP initiator will geteduced by these redox catalysts at rates close to diffusion limit. Thencomplete reduction of the initiator layer is then likely not due tohe reducing strength of the redox catalyst.

Furthermore, local reduction was also achieved using two otheredox couples, namely p-tolunitrile (E◦ = −2.36 V vs SCE) that dis-lays a more negative standard potential than 2,2′-dipyridyl, anderephthalonitrile (E◦ = −1.50 V vs SCE) with a more positive stan-ard potential than 2,2′-dipyridyl. In both cases, the local reduction

eads to the formation of patterns with similar dimensions withncomplete reduction. There is, however, a little influence of the

ediator reducing power on the pattern width. It seems that, for aiven L value, the less reducing anion radical of terephthalonitrileenerates slightly wider patterns while the more reducing anionadical of p-tolunitrile generates patterns slightly thinner. Thesexperiments were performed only once and should be taken withare but they seem to correlate the trends in the approach curvesepicted in Fig. 2. As the pattern depth is concerned, even the usef a mediator with a higher reducing power does not allow the totaleduction of the C–Br bonds since some polymer still grows insidehe pattern. It then appears that the mediator reducing power hasot much influence on the pattern depth, meaning that with thewo most reducing anion radicals used, the reduction of the ini-iator layer is achieved at its maximum yield. This is in agreementith what was observed on Si surfaces where the initiator layer was

ompletely reduced with 2,2′-dipyridyl.

.2.4. Patterning in the direct mode of the SECMIn order to improve the microelectrochemical patterning proce-

ure we have performed the direct local reduction of the initiatorayer electrografted on Au surfaces, in opposition with its indirecteduction in the presence of a mediator. This study will also allow toompare the extent of patterning when the reduction is achieved,n the former situation, from the lower (in contact with the Au sub-trate) or from the upper (in contact with the solution) side of thenitiator organic layer.

The same setup is used, namely the microelectrode is placedear the initiator grafted gold surface. For a better control of thelectrochemical processes, a 2-electrode configuration was pre-erred [49]. The gold surface modified by the initiator layer is useds the working electrode in a resistive dilute electrolytic solution ofMF (5 mM of NBu4BF4) and the microelectrode is used as counter

lectrode in order to focus the reduction. The reduction of the ini-iator layer was achieved in a solution of DMF containing only 5 mMf NBu4BF4, by applying a reductive potential of −2.6 V to the goldurface vs the Pt microelectrode tip. This potential has been chosens the onset of the reduction wave observed by cyclic voltammetry

Acta 54 (2009) 5127–5136 5133

in the same 2-electrode configuration, for the reduction of the Auelectrografted initiator layer.

The preliminary experiments show that it is possible to decreasethe pattern width by using the direct reduction instead of theindirect one. For the same velocity and a higher microelec-trode/substrate distance than in the indirect mode, the patternwidth, estimated by SEM, is clearly lower as shown in Fig. 4 (+).

Moreover, interferometric measurements show that this config-uration has an effect on the evolution of the ratio �e/t when the tipis moved near the substrate. Increasing the tip velocity decreasesthe pattern depth. Furthermore, this ratio reaches a value of 0.84when the tip is moved at 5 �m/s, meaning that the major part ofthe initiator layer is reduced. When using a lower velocity, a totalreduction of the C–Br bonds should be expected.

In the direct mode, the local surface reduction is then more effi-cient than in the indirect mode, even though the direct reductionuses less reducing potentials (Esubs ∼−1.6 V vs SCE compared withthe 2,2′-dipyridyl radical anion, E0 = −2.1 V vs SCE). This suggeststhat, at least for slow tip speeds, the direct charge transfer from theAu surface to the initiator layer is more favoured than the chargeinjection from the upper side of the layer that is connected to thesolution. This could be related to the ease of transport of chargesand species within the organic layer.

3.3. Toward a picture of charge transport at diazoniumelectrografted multilayers

Contrary to what is observed for the homogeneous reduction of1 and previous patterning of insulating surfaces, the local mediatedreduction of the initiator layer electrografted on Au is not completeand reaches, at best, 60%. This means that some Br terminated func-tions are still available within the locally reduced initiator layer thatcannot be accessed by the microelectrogenerated reagent. It agreeswell with the ellipsometric characterization that indicates that theelectrografting procedure leads to the growth of a multilayer struc-ture, most likely consistent with the large nodules observed by AFM(inset of Fig. 6A).

The incomplete reduction of C–Br functions of the initiator mul-tilayer means that the Br functions buried in the core of thesenodules do not get reduced at the timescale of the patterning (<10 s).The progressive controlled growth of polymer structures demon-strates that these hindered C–Br functions are still available for thereduction step that occurs during the ATRP, however the timescaleof the ATRP polymerization is much longer (>1 h). It is not surpris-ing that the more organized self-assemblies of silanes (or thiols)are reduced more steadily than the more disordered electrograftedaryl multilayers obtained by the reduction of diazonium ions. Thepartial accessibility of the C–Br functions of the electrografted layermay be correlated to the difficulty to quantify electroactive groupspresent in a compact organic layer obtained from the electrograft-ing of diazonium ions [28,64].

For the reductive local transformation of the material during thepatterning, the electron transfer across the electrografted multi-layer may be decomposed into different parallel processes whenpermeation is neglected, as presented in Scheme 2. They corre-spond to: (1) the ET at macropores or (2) at nanopores and, (3)the irreversible reduction of the layer by the reducer. The globalET process, described by the total substrate current, iET, is the sumof all these different processes whose characteristic currents are,respectively, i , i , i :

iET = imp + inp + ired (5)

The current required for the local material transformation dur-ing the SECM patterning of the electrografted layer may be roughlyestimated from [65]

Page 8

5134 F. Hauquier et al. / Electrochimica Acta 54 (2009) 5127–5136

S multt d redua the re

i

wolmtfdwdblaicomas

stt

k

sa[

k

wpo

dii

cheme 2. Possible paths for transport of a redox probe through an electrograftedransport of organic probes in meso- or nano-pores of radius Rd, (3) partial mediatend B = Red. In (1) and (2) discharge of B at the conductive surface occurs whatever

red = 2F0.6�Brwv (6)

here F is the faraday, � Br represents the surface concentrationf the initiator molecule (<5 × 10−9 mol/cm2 for a 2–3 nm thickayer), w is the pattern width and v is the tip writing speed. Its

aximal value is 2 × 10−8 A, which is considerably small comparedo the 5 × 10−7 A flowing through the substrate. The major pathor the electron flow into the system during the patterning anduring the approach curves is then related to the mass transportithin the electrografted layer by transport through pinholes andefects toward the Au electrode. As the layer is partly reducedy the different reducers, permeation through the whole layer isikely not favoured and transport through defects is preferred. Actu-lly, permeation through the film involves very small channels;t is likely prevented in mediated reduction owing to the diffi-ulty to transport both the reducing species and, for the respectf local electroneutrality, the electrolyte cation within the denseultilayer. This is indeed consistent with the heterogeneity in the

pproach curves recorded from sample-to-sample even though theame thickness of initiator layer was immobilized.

Following Mirkin’s formalism [41,66] the effective ET rate con-tant obtained by fitting the experimental approach curves toheory, as made in Fig. 2, then leads to the estimate of the ET for theransport through defects (macropores, kmp and nanopores, knp):

eff = kmp + knp (7)

Generally the mass transfer limited ET at pinholes of averageurface density Np and average radius Rp can be measured by SECMs it is characterized by an apparent ET rate constant given by:45,46,67,68]

p = 4NpDRp = 4�pD

�Rp(8)

here D is the solution diffusion coefficient of the redox probe (theinhole is assumed to be filled by solution), �p is the surface fraction

f the pinholes (�p = Np�R2

p), with the index p = mp or np.The lower rate observed for aqueous redox mediators would

epict transport through macropores (and/or partial conductiv-ty if possible) of the organic layer. The transport of hydrophilicons through meso- or nano-pores is prevented owing to capil-

ilayer. (1) Transport of organic and aqueous probes in macropores of radius Rd, (2)ctive local transformation (into a gray material) of the multilayer only when A = Oxdox nature of B (reducer or oxidant).

larity restrictions (hydrophobic pores inaccessible to the aqueoussolution) and/or the organic layer shrinking.

From the inspection of the Au grafted layer by cyclic voltam-metry with ferrocyanide as the aqueous redox probe, thesurface coverage of the larger defects (macropores) may beestimated. It is obtained from the ET rate constant extractedfrom cyclic voltammetry (Figures S2 and S3) at the naked,k0

hetCV,Au, and electrografted, k0hetCV,Au-R, Au electrodes as �mp =

k0hetCV,Au-R/k0

hetCV,Au ∼ 2 − 3 × 10−4. Combined with (8) and theSECM inspection of the surfaces, one obtains the averagedefect radius, Rmp = 4�mpD/�kSECM,FeCN ∼50 nm, and their densityNmp ∼3 × 106 cm−2. The aqueous redox probe transfers throughvery large pores in the grafted film that are largely distant from eachother, as the average distance between two neighbouring pores canbe roughly estimated as R0 ∼ Rmp/�1/2

mp ∼ 3 �m [67]. It is consistentwith the AFM image since such large pores could be the black regionobserved in the inset of Fig. 6A.

This picture also confirms the recent question arisen by Dow-nard and co-workers [69,70] or McDermott and Kariuki [48,55]concerning the use of redox probes for the inspection of electro-grafted multilayers. We believe that the combination of SECM andCV gives complementary characterization and interesting insightinto such complex film structures. Again, the SECM demonstratedthat this description is reasonable; indeed, the proper patterningof the surface required the perfect parallelism of the tip travelplane and the Au surface plane. If this alignment is easy to per-form with perfectly insulating surfaces, it is more delicate with theelectrografted Au surfaces because linescan of the surface showsheterogeneities in the measured feedback current along the lines-can. Such heterogeneities are commonly observed when inspectingmaterial surfaces [65,71]; here, they could be accounted for thepeculiar large pores size and distribution.

When considering the organic redox probes, higher feedbacksare observed. It is then expected that the defects contribution will

be the sum of the transport through the same large defects andthe transport through smaller pores because of their better wettingand swelling by the organic solvent [70]. The effective heteroge-neous rate constant measured by the SECM reads as the sum ofboth contributions (7), among which that of the macropores may

Page 9

imica

b

k

weto(tfeaNhaeocitOHoto

atpsptodgotfipi

tween

rpoiatwcsr

ttpatti

F. Hauquier et al. / Electroch

e neglected, and:

eff = 4NmpDRmp + 4NnpDRnp ∼ 4�npD

�Rnp(9)

ith Np, �p and Rp, respectively the surface density, surface cov-rage, radius and heterogeneous rate constant for the transporthrough the two different defects; p = mp, large defects or macrop-res, p = np smaller mesopores. As the less reducing redox probeanion radical of terephthalonitrile) presents a keff very close tohat of the 2,2′-dipyridyl anion radical, despite a 600 mV E0 dif-erence, the value for the former probe could represent a roughstimate of the contribution of the transport within both largend smaller pores, and one could then estimate the productnpRnp ∼ kSECM/4DPht ∼75 cm−1. If one considers that those poresave an average radius of Rnp ∼10 nm, it is expected that they havesurface density of Nnp ∼7.5 × 107 pores per cm2 or a surface cov-rage of �np ∼2.5 × 10−4. From the cyclic voltammetry of ferrocenexidation at the Au electrografted electrode, a value of the surfaceoverage of all the defects of �d ∼4 × 10−4, may be obtained (assum-ng that on naked Au electrode k0

het ∼ 4 cm/s). This value is close tohat observed with the aqueous redox probe for the macropores.wing to the film reconstruction upon the solvent change (from2O to DMF), it is difficult to know whether the pores structurebserved in the aqueous environment is maintained and thereforeo conclude if �d = �mp + �np. Anyway, the nanopores implied in therganic redox probes are likely smaller than 16 nm.

The higher ET rate observed for the more reducing redox probesnd for the ferricinium cation could depict some diffusion limita-ions related to the dimension of the redox probe compared to theores dimension. Indeed, even though all species investigated haveimilar diffusion coefficients, the ion pair they will form to trans-ort charges within the pores could be very different. For example,he Fc+,BF4

− pair is much smaller than the M•–,NBu4+ one. More-

ver, the negative charge in the different radical anions is mainlyelocalized around the N atoms giving rise to dipoles that couldenerate ion pairs with different sizes. Finally, the apparent increasef keff when the E0 of M decreases could be a manifestation ofhe potential drop within the pores between the film–solution andlm–electrode interfaces. The more negative the E0, the larger theotential drop and the more efficient the migration transport of the

on within the pores.The direct mode reduction only requires transport of the elec-

rons through the (partly conductive [72]) aryl layer in conjunctionith the injection and transport/permeation of the cation of the

lectrolyte of this layer. In opposition, the indirect reduction by thelectrogenerated reagent requires the transport of a more volumi-ous molecular assembly and therefore requires larger channels.

Finally, the great dependence of the pattern depths on the tipate (or the reduction time) reveals the kinetics of the charge trans-ort into the electrografted layer structure. For a tip scan rate of therder of 10 �m/s, the initiator layer is half-reduced, the correspond-ng time of flight of the substrate by the tip, � = 2 a/v ∼ 2 s, givesn estimate of the characteristic time of this phenomenon. Thisransport is likely limited by the diffusion of the electrolyte cationsithin the 3 nm thick initiator layer with an apparent diffusion

oefficient of the order of 5 × 10−14 cm2/s and suggests con-trained diffusion into very thin channels or very slow permeationates.

In the indirect mode, the pattern depth does not depend on theip rate (or the reduction time); owing to the fastest tip rate used,his shows that the limiting process that characterizes the trans-

ort of the mediated reduction of the upper initiator layer occurst a characteristic time shorter than 0.05 s, two orders of magni-ude faster than for the direct reduction. This also supports thathe mediated reduction engages only the superficial layer of thenitiator and transport into larger channels.

Acta 54 (2009) 5127–5136 5135

4. Conclusion

The local etching of polymer brushes onto conducting substratesmay be achieved by a method previously described for insulat-ing substrates. It is based on the overall on atom transfer radicalpolymerization and scanning electrochemical microscopy. After thesurface functionalization by a brominated initiator, SECM allowsthe local reduction at the micrometer scale of the initiator layerpreviously grafted on the gold surface. As already described foran initiator grafted layer onto an insulating substrate, the initiatordecomposition consists of the reductive debromination of the ini-tiator by a radical anion electrogenerated at the microelectrode tipof the SECM. ATR polymerization on these surfaces leads to the for-mation of a local patterning whose morphologies depend on the tipvelocity. The pattern width decreases as the tip velocity increases.Moreover, the closer the microelectrode to the surface, the lowerthe width. Nevertheless, contrary to an insulating substrate, suchas glass or silicon oxide, the local reduction of the initiator layeris incomplete. Furthermore, the use of a more reducing mediatorhas no influence on the efficiency of this local reduction. The persis-tence of some C–Br bonds into the reduced area enables the growthof some polymer chains. Inside the pattern, the lower density ofinitiator decreases the polymer thickness.

An alternative method, using the gold substrate as the workingelectrode and the microelectrode as the counter electrode, gave abetter pattern resolution. This local electrochemical etching, thatdoes not require a mediator, could be a promising alternative forthe patterning of conducting substrates. On the contrary to whatwas observed on insulating substrates, the incomplete reductionof the initiator layer is quite frustrating. Indeed, it means that acomplete debromination of the initiator layer is hardly obtainedby an electrochemical means owing to the complex structure ofthe initiator multilayer. In the present case, this would make themasking technique superior to the SECM alternative proposed here.

Finally, the combination of SECM, CV and observation of thearyl-grafted layer transformation allows to evidence the differentchannels involved in charge transport within these multilayers. Itseems that aqueous redox probes are vehicled within large defectsof about 50 nm while the transport of organic electroactive speciesexploit 5–10 times smaller meso- or nano-pores. The slow perme-ation of smaller species is likely possible during the direct injectionof charges from the electrode surface.

Acknowledgments

F. Hauquier and T. Matrab acknowledge the ANR for financialsupport via the ANR-06-BLAN-0368 project. J. Ghilane from ITODYSis thanked for help for the acquisition of the AFM images.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.electacta.2009.01.059.

References

[1] Y. Tsujii, M. Ejaz, S. Yamamoto, T. Fukuda, K. Shigeto, K. Mibu, T. Shinjo, Polymer43 (2002) 3837.

[2] U. Schmelmer, R. Jordan, W. Geyer, W. Eck, A. Golzhaüser, M. Grunze, A. Ulman,Angew. Chem., Int. Ed. 42 (2003) 559.

[3] I.S. Maeng, J.W. Park, Langmuir 19 (2003) 4519.[4] S.J. Ahn, M. Kaholek, W.-K. Lee, B. LaMattina, T.H. LaBean, S. Zauscher, Adv. Mater.

16 (2004) 2141.

[5] Q. He, A. Küller, M. Grunze, J. Li, Langmuir 23 (2007) 3981.[6] M. Husemann, M. Morrison, D. Benoit, J. Frommer, C.M. Mate, W.D. Hinsberg,

J.L. Hedrick, C.J. Hawker, J. Am. Chem. Soc. 122 (2000) 1844.[7] P. Iwata, P. Suk-In, V.P. Hoven, A. Takahara, K. Akiyoshi, Y. Iwasaki, Biomacro-

molecules 5 (2004) 2308.[8] F. Zhou, L. Jiang, W. Liu, Q. Xue, Macromol. Rapid Commun. 25 (2004) 1979.

Page 10

5 imica

[

[[

[

[

[[

[

[[[

[[[

[

[

[[[[[

[

[

[

[

[

[[

[[

[[[[

[

[[[[[

[

[[

[[[[

[[[

[[

[[

[[[[[[

136 F. Hauquier et al. / Electroch

[9] L. Andruzzi, W. Senaratne, A. Hexemer, E.D. Sheets, S.B. Ilic, J.K. Kramer, B. Baird,C.K. Ober, Langmuir 21 (2005) 2495.

10] F.J. Xu, Y. Song, Z.P. Cheng, X.L. Zhu, C.X. Zhu, E.T. Kang, K.G. Neoh, Macro-molecules 38 (2005) 6254.

11] R. Konradi, J. Ruhe, Langmuir 22 (2006) 8571.12] V.S. Khire, A.W. Harant, A.W. Watkins, K.S. Anseth, C.N. Bowman, Macro-

molecules 39 (2006) 5081.13] R. Dong, S. Krishnan, B.A. Baird, M. Lindau, C.K. Ober, Biomacromolecules 8

(2007) 3082.14] R.R. Shah, D. Merreceyes, M. Husemann, I. Rees, N.L. Abbott, C.J. Hawker, J.L.

Hedrick, Macromolecules 33 (2000) 597.15] A.Y. Sankhe, B.D. Booth, N.J. Wiker, S.M. Kilbey, Langmuir 21 (2005) 5332.16] C. Slim, Y. Tran, M.M. Chehimi, N. Garraud, J.-P. Roger, C. Combellas, F. Kanoufi,

Chem. Mater. 20 (2008) 6677.17] S. Nishiyama, A. Goto, K. Saito, K. Sugita, M. Tamada, T. Sugo, T. Funami, Y. Goda,

S. Fujimoto, Anal. Chem. 74 (2002) 4933.18] F.J. Xu, Q.J. Cai, Y.L. Li, E.T. Kang, K.G. Neoh, Biomacromolecules 6 (2005) 1012.19] S. Ko, J. Jang, Biomacromolecules 8 (2007) 1400.20] T. Matrab, M.M. Chehimi, C. Perruchot, A. Adenier, V. Guillez, M. Save, B.

Charleux, E. Cabet-Deliry, J. Pinson, Langmuir 21 (2005) 4686.21] for example see: P.M. Diakowski, A.S. Baranski, Electrochim. Acta 52 (2006) 854.22] W.H. Yu, E.T. Kang, K.G. Neoh, Langmuir 20 (2004) 8294.23] M. Hikita, K. Tanaka, T. Nakamura, T. Kajiyama, A. Takahara, Langmuir 20 (2004)

5304.24] M. Cerruti, S. Fissolo, C. Carraro, C. Ricciardi, A. Majumdar, R. Maboudian, Lang-

muir 24 (2008) 10646.25] M.W.J. Beulen, M.I. Kastenberg, F.C.J.M. van Veggel, D.N. Reinhoudt, Langmuir

14 (1998) 7463.26] K. Boukerma, M.M. Chehimi, J. Pinson, C. Blomfield, Langmuir 19 (2003) 6333.27] A.M. Nowak, R.L. McCreery, Anal. Chem. 76 (2004) 1089.28] C. Combellas, F. Kanoufi, J. Pinson, F.I. Podvorica, Langmuir 21 (2005) 280.29] D. Jiang, B.G. Sumpter, S. Dai, J. Am. Chem. Soc. 128 (2006) 6030.30] T. Matrab, M. Save, B. Charleux, J. Pinson, E. Cabet-Deliry, A. Adenier, M.M.

Chehimi, M. Delamar, Surf. Sci. 601 (2007) 2357.31] T. Matrab, J. Chancolon, M. Mayne L’Hermite, J.-N. Rouzaud, G. Deniau, J.-P.

Boudou, M.M. Chehimi, M. Delamar, Colloids Surf. A 287 (2006) 217.32] T. Matrab, M.M. Chehimi, J.-P. Boudou, F. Benedic, J. Wang, N.N. Naguib, J.A.

Carlisle, Diamond Relat. Mater. 15 (2006) 639.33] T. Matrab, M.N. Nguyen, S. Mahouche, P. Lang, C. Badre, M. Turmine, G. Girard,

J. Bai, M.M. Chehimi, J. Adhes. 84 (2008) 684.34] M. Husseman, E.E. Malmström, M. McNamara, M. Mate, D. Mecerreyes, D.G.

Benoit, J.L. Hedrick, P. Mansky, E. Huang, T.P. Russell, C.J. Hawker, Macro-

molecules 32 (1999) 1424.

35] K. Matyjaszewski, P.J. Miller, N. Suhkla, B. Immaraporn, A. Gelman, B.B. Luokala,T.M. Siclovan, G. Kickelbick, T. Vallant, H. Hoffmann, T. Pakula, Macromolecules32 (1999) 8716.

36] K. Matyjaszewski, J. Xia, Chem. Rev. 101 (2001) 2921.37] M. Kamigaito, T. Ando, Chem. Rev. 101 (2001) 3689.

[

[[

Acta 54 (2009) 5127–5136

38] S. Edmonson, V.L. Osborne, W.T.S. Huck, Chem. Soc. Rev. 33 (2004) 14.39] W.J. Brittain, S.G. Bowes, A.M. Granville, M. Baum, B.K. Mirous, B. Akgun, B. Zhao,

C. Bickle, M.D. Foster, Adv. Polym. Sci. 198 (2006) 125.40] W. Braunecker, K. Matyjaszewski, Prog. Polym. Sci. 32 (2007) 93.41] C. Wei, A.J. Bard, M.V. Mirkin, J. Phys. Chem. 99 (1995) 16033.42] R. Cornut, C. Lefrou, J. Electroanal. Chem. 621 (2008) 178.43] C. Combellas, F. Kanoufi, J. Pinson, F.I. Podvorica, J. Am. Chem. Soc. 130 (2008)

8576.44] C. Combellas, De-en Jiang, F. Kanoufi, J. Pinson, F.I. Podvorica, Langmuir 25

(2009) 286.45] C. Cannes, F. Kanoufi, A.J. Bard, J. Electroanal. Chem. 547 (2003) 83.46] C. Cannes, F. Kanoufi, A.J. Bard, Langmuir 18 (2002) 8134.47] T. Matrab, C. Combellas, F. Kanoufi, Electrochem. Commun. 10 (2008) 1230.48] J.K. Kariuki, M.T. McDermott, Langmuir 17 (2001) 5947.49] D. Zigah, M. Pellissier, F. Barrière, A.J. Downard, P. Hapiot, Electrochem. Com-

mun. 9 (2007) 2387.50] C. Amatore, C. Combellas, F. Kanoufi, C. Sella, A. Thiébault, L. Thouin, Chem. Eur.

J. 6 (2000) 820.51] D. Mandler, P.R. Unwin, J. Phys. Chem. C 107 (2003) 407.52] J. Zhang, C.J. Slevin, C. Morton, P. Scott, D.J. Walton, P.R. Unwin, J. Phys. Chem. B

105 (2001) 11120.53] F. Hauquier, J. Ghilane, B. Fabre, P. Hapiot, J. Am. Chem. Soc. 130 (2008) 2748.54] F. Anariba, S.H. DuVall, R.L. McCreery, Anal. Chem. 75 (2003) 3837.55] J.K. Kariuki, M.T. McDermott, Langmuir 15 (1999) 6534.56] J. Ghilane, M. Delamar, M. Guilloux-Viry, C. Lagrost, C. Mangeney, P. Hapiot,

Langmuir 21 (2005) 6422.57] for a recent review see: A. Houmam, Chem. Rev. 108 (2008) 2180.58] T. Kjaersbo, K. Daasbjerg, S.U. Pedersen, Electrochim. Acta 48 (2003) 1807.59] H. Lund, K. Daasbjerg, T. Lund, D. Occhialini, S.U. Pedersen, Acta Chem. Scand.

51 (1997) 135.60] A. Cardinale, A.A. Isse, A. Gennaro, Electrochem. Commun. 4 (2002) 767.61] see for example: C. Costentin, M. Robert, J.M. Savéant, Chem. Phys. 324 (2006)

30.62] C.P. Andrieux, A. Le Gorande, J.-M. Savéant, J. Am. Chem. Soc. 114 (1992) 6892.63] R. Fuhlendorff, D. Occhialini, S.U. Pedersen, H. Lund, Acta Chem. Scand. 43 (1989)

803.64] P.A. Brooksby, A.J. Downard, J. Phys. Chem. B 109 (2005) 8791.65] C. Combellas, F. Kanoufi, D. Mazouzi, J. Electroanal. Chem. 589 (2006) 243.66] B. Liu, A.J. Bard, M.V. Mirkin, S.E. Creager, J. Am. Chem. Soc. 126 (2004) 1485.67] C. Amatore, J.-M. Savéant, D. Tessier, J. Electroanal. Chem. 147 (1983) 39.68] F. Forouzan, A.J. Bard, M.V. Mirkin, Isr. J. Chem. 37 (1997) 155.69] A.C. Cruickshank, E.S.Q. Tan, P.A. Brooksby, A.J. Downard, Electrochem. Com-

mun. 9 (2007) 1456.70] M.G. Paulik, P.A. Brooksby, A.D. Abell, A.J. Downard, J. Phys. Chem. C 111 (2007)

7808.71] F. Kanoufi, C. Combellas, M.E.R. Shanahan, Langmuir 19 (2003) 6711.72] A. Adenier, C. Combellas, F. Kanoufi, J. Pinson, F.I. Podvorica, Chem. Mater. 18

(2006) 2021.