Influence of electrode material and roughness on iron electrodeposits dispersion by ultrasonification

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in :

http://oatao.univ-toulouse.fr/

Eprints ID : 15089

To link to this article : DOI:10.1016/j.electacta.2015.10.052

URL :

http://dx.doi.org/10.1016/j.electacta.2015.10.052

To cite this version :

Iranzo, Audrey and Chauvet, Fabien and

Tzedakis, Théodore Influence of electrode material and roughness

on iron electrodeposits dispersion by ultrasonification. (2015)

Electrochimica Acta, vol. 184. pp. 436-451. ISSN

0013-4686

Any correspondence concerning this service should be sent to the repository

administrator:

staff-oatao@listes-diff.inp-toulouse.fr

Influence

of

electrode

material

and

roughness

on

iron

electrodeposits

dispersion

by

_{ultrasonification}

A.

Iranzo

a,b

_,

_F.

_Chauvet

a,b,

_*

_,

_T.

_Tzedakis

a,b,

_**

a_{UniversitédeToulouse,INPT,UPS,LaboratoiredeGénieChimique,118RoutedeNarbonne,F-31062Toulouse,France} b_{CNRS,UMR5503,F-31062Toulouse,France}

Keywords:

zero-valentironelectrodeposition sonoelectrochemistry

adhesionenergy surfaceroughness electrodepositsdispersion

ABSTRACT

This studyrelates thesonoelectrochemical productionof metallicparticles andnanoparticles. The emphasis is on the influence of electrode material and roughness on the morphology of iron electrodepositsandtheirdispersionfromtheelectrodebyultrasonification.Ultrasonificationiseither appliedduringcyclicvoltammetrieswithsolutionstirringoraftergalvanostaticironelectrodeposition; nodispersionwasobservedwhenusingagoldelectrode,whereasdispersionwasalwaysobservedwhen usingvitreouscarbon(VC) substrates.ScanningElectronMicroscopy (SEM)imagingofthe electro-depositsshowshigherironcoverageongoldthanonVCelectrodes.IronspreadsmoreongoldthanonVC. Thevaluesofboththeinterfacialenergyoftheiron/electrodeinterfaceandtheworkofadhesionofiron ontheelectrodeareinagreementwiththepreviousobservations.DispersionkineticsonVCwerefound tobedependentontheelectrodesurfaceroughness.Resultssuggestthatdispersionfollowsafirstorder kinetics,whichiscoherentwiththeconstantactionofcavitationbubblesinthevicinityoftheelectrode surface.Enhancementofmass-transferbyultrasoundhasalsobeenobserved.

1. Introduction

Iron-based nanoparticles (Fe3O4,

g

-Fe3O4, nickel-cobalt-iron

alloy)exhibitinterestingmagneticpropertiesinthemedicalfield,

suchascontrastagentsforMagneticResonanceImaging[1,2]andfor

treatment of tumors by hyperthermia [3,2]. Recent studies [4]

indicatethatzero-valentironnanoparticles(ZVI-nP)showbetter

magnetic performancethanironoxidesformedicalapplications.

ZVI-nPcanalsobeusedforvariousotherapplicationssuchasan

effective reducing agent: i) in water treatment [5], ii) for

dechlorination[6,7],iii)nitrateremoval[7–9],andiv)for destruction ofvariousotherpollutants(see[5]foranextensivereview).

ZVI-nPcanbesynthetizedbyseveraltechniques(see[10]and

[5] for an exhaustive list):ball milling [11], thermal reduction

(reductionofironsalts[12]andreductionofoxide[13]),ironsalts

wet-chemical reduction [6,14] (using borohydride salt as a

reducing agent) and direct electrochemical reduction [15,16,9].

Themostwidelyusedtechniqueisthewet-chemicalmethod.

Scaling-upofwet-chemicalmethod,inordertoproducelarge

quantities of ZVI-nP, requires expensive reagents and specific

securityconditionsbecauseofthegaseoushydrogenproduction

[12].

DirectelectrochemicalZVI-nPsynthesisappearsasapromising

techniqueforeconomicandsafeproductionprocesses,especially

atalargescale;thereducingreagentisreplacedbyelectricityand

under controlled conditions hydrogen production is avoided.

Nevertheless, metallic iron, produced at the cathodic surface,

mustberemovedfromtheelectrodeanddispersedintheliquidat

therequiredsize.Variousworksinvolveultrasonicdispersionof

electrodepositediron, simultaneouslyor sequentiallywith

iron-precursorsreduction,allowingtherenewalofthecathodesurface.

Generally,powerultrasounds(!20kHz,usinganultrasonichorn

orbath)areusedtogeneratecavitationbubblesthat,duringtheir

violentcollapses,createfluidmotionwhichremovessoliddeposits

fromtheelectrodesurface.During ironelectrodeposition,under

sequential pulses of the appliedcurrent, Delplancke et al. [15]

applyshiftedpulsesof ultrasounds,usinga titaniumultrasonic

horn (20kHz, 50W/cm2_{), also used as the polarized cathode.}

Nanoparticles(6nm-100nm)ofpureFeandalloysofFe/Ni/Cohave

been successfully synthetized in aqueous solutions; partial

‘chemicaloxidation’ofpureironparticleshasbeenobserved.In

* Correspondingauthorat:LaboratoiredeGénieChimique,UMRCNRS5503, UniversitédeToulouseIII-PS, INPT,118routedeNarbonne,F-31062Toulouse, France.Tel.:+33561557468.

**Correspondingauthorat:LaboratoiredeGénieChimique,UMRCNRS5503, UniversitédeToulouseIII-PS, INPT,118routedeNarbonne,F-31062Toulouse, France.Tel.:+33561558302.

E-mailaddresses:chauvet@chimie.ups-tlse.fr(F. Chauvet),

anotherstudy[16],cathodeisassembled bothwithahigh

(0.2-2MHz, 5W/cm2_{) and a low (20kHz, 100W/cm}2_{) frequency}

ultrasonictransducerswhoirradiatethecathodicarea;constant

current electrolyses werecarried out intetrahydrofuran, under

ultrasonification,and10nmsizedZVI-nPareproducedwhenthe

lowandthehighfrequencytransducersareusedsimultaneously.

Chenetal.[9],claimthatZVI-nPrangingbetween1-20nmwere

synthetized using aqueous 1M FeCl3 in the presence of

cetylpyridiniumchlorideas a dispersingagent; atwo platinum

electrodes‘classicalelectrochemicalcell’,‘entirely’immersedinto

anultrasonicbath(20kHz),wasusedandgalvanostatic

electrol-yses were carried out under ultrasonification. Other metallic

nanoparticles(silver,palladium,platinum,zinc,nickel,gold)have

beensynthesizedbythesonoelectrochemicalmethod,see[17]for

areview.Previousstudiesweregenerallydevotedtodetermining

thebestoperatingparameters(pulsetimes,current,temperature,

ultrasoundsintensity,stabilizers,etc.)foragivenset-upallowing

thesynthesisofdesirednanoparticles.

Effectsofpowerultrasoundonvariouselectrochemicalsystems

have also been extensively studied. Ultrasounds are generally

appliedusingatitaniumultrasonichornactingastheultrasound

generator placed in front of the working electrode (‘face-on’

configuration); the working electrode can also be directly

integratedontheultrasonichorn([18]).Anotherpossibilityisto

immersetheelectrochemicalcellinanultrasonicbath[19].Among

other effects and regardless of the experimental configuration,

ultrasounds induce cavitation bubbles and acoustic streaming

whichenhancemass-transferleadingtoanincreaseofthecurrent

[20–25,19].Currentfluctuationsarealsoobservedduetoviolent bubblecollapses[21,26,27].

Theabilityofthefluidmotionandalsotheshocksinducedby

ultrasoundto‘clean’theelectrodesurfacehasbeeninvestigatedin

severalworks.Cavitationbubblecollapsesinducedbyultrasound

werefoundtobeabletoactivateelectrodesavoidingpassivationby

‘eroding/roughening’ the electrode material [28,19]. Coupling

mercury electrodeposition and ultrasonification simultaneously

on a vitreous carbonelectrode has led to a ‘steadystate’regime where

the quantity of electrodeposited Hg on the electrode remains

constant(electrodepositionflux=ablationflux)[29].Under

ultra-sonification,theelectrodepositionofmetalssuchasZn,Co,Pb,and

Hg, was investigated on a vitreous carbon substrate by cyclic

voltammetryin[30](seealso[31]).Theauthorsanalyzedtheratioof anodictocathodicchargesasafunctionoftheultrasoundintensity,

for theirparticular experimentalconditions (potential scanrate,

sonoelectrochemicalsystem,electrodematerial...),andshowed

thatitdependsonboththeultrasoundintensityandthemetalused.

Thissuggests theimportance of metal/substrate affinityon the

electrodepositsdispersion.Consideringthecaseofaparticlelyingon

asubstrateandsubmittedtoultrasonificationin[32],theauthors

discussedthecompetitionbetweenthehydrodynamicforcesand

theadhesionforceasafunctionoftheparticlesize;theyshowedthat

ultrasoundsshouldbeabletoremovesubmicrometerparticlesbut

not smaller particles such as molecular adsorbates (adhesion

outweighshydrodynamicsforces).

Thesepreviousstudiesshowthattheefficiencyofultrasoundto

remove/disperseelectrodepositedmetalsshoulddependon:

-theultrasound intensity (and theexperimental configuration

used)[30,31]

-thesize/morphologyandthespatialdistributionofdeposited

metallicparticles[32]

-theadhesionenergyofelectrodepositedmetalontheelectrode

material

Note that, todate, theeffect of the adhesion energy of the

electrodepositedmetalontheelectrode substrate,hasnotbeen

directlyinvestigated.

In the present work, the dispersion by ultrasound of iron

electrodepositsisstudiedforvariouselectrodematerials(goldand

vitreouscarbons)whichallowstovarytheadhesionenergy.The

effect of the electrode material on the morphology of the

electrodepositedironisalsostudiedbySEMimaging.Theinfluence

of the electrode surface roughness is investigated using VC

electrodeshavingdifferentlevelsofpolishing.

Nomenclature

A geometricsurfacearea(m2₎

Aelect Hamakerconstantof electrode material, elect=AuorVC(J)

A_elect=L=Fe Hamakerconstantforelectrodematerial

(elect = Au or VC) and iron interacting

acrosstheliquid(J)

Ar Argon

AFe;AL Hamaker constants of the iron and the

liquid(J)

C_Hþ Protonsconcentration(mol/m3)

d Separationdistance(m)

D Diffusioncoefficient(m2_/s)

e Thicknessofthedeposit(m)

E_adh AdhesionenergyofirononVCelectrode=

WFe=VCSFe=VC (J)

E Electrodepotential(V)

F Faradayconstant(96,500C/mol)

h Localsurfaceheight(m)

jlim Limitingcurrentdensity(A/m2)

I,Iapplied,Ilim Current, applied current and limiting

current(A)

MFe Molecularweightofiron(kg/mol)

n Electronsnumber

Qc,Qa,andQaref Respectivelythecathodic,theanodicand

theanodic referenceamountof charges

(C)

r Potentialscanrate(mV/s)

ra Arithmeticroughness(m)

rw Wenzel's roughness = ratio between

actualsurfaceandgeometricalsurface

s Dispersionrateconstant(s#1₎

Sdisk Surfaceareaofthediskelectrode(m2)

t Time(s)

tUS Ultrasonificationduration(s)

WFe=elect Workofadhesionofironontheelectrode, elect=AuorVC(J/m2₎

x,y Cartesiancoordinates(m)

Greekletters

g

a Surfacetensionofmediuma(J/m2)

g

a/b Interfacialtensionbetweenbothmediaaandb(J/

m2₎

g

elect Surfaceenergyoftheelectrode,elect=AuorVC1

orVC2(J/m2₎

g

dand

g

p Respectively thedispersiveand thepolar

com-ponentsofthesurfaceenergy(J/m2₎

h

Overpotential(V)

r

Fe Volumetricmassofiron(kg/m3)

n

Kinematicviscosity(m2/s)

F

a/b Adjustmentparameterwhichdependson

inter-actionsbetweenbothinteractingmediaaandb

Thestudyisbasedontheanalysisofelectrochemical

measure-ments:voltamperometriesandgalvanostaticelectrolysescoupled

simultaneously or sequentially with ultrasonification.

Experi-mentswererealizedinaclassicalelectrochemicalcellimmersed

in an ultrasonic bath. Two different salts (FeCl2 and (NH4)2Fe

(SO4)2)wereusedasironelectrodepositprecursors.

Experimentalresultsonironelectrodepositsmorphologyand

adhesion are discussed on the basis of a theoretical analysis,

enablingtheestimationoftheworkofadhesionandtheinterfacial

tensionbetweenironandthesubstrate,whichispresentedinthe

Appendix.

2. Materialandmethods

2.1.Chemicalsandexperimentalset-up

Allsolutions,preparedusingultrapurewater(18.2M

V

.cm),were deaeratedbeforeexperiments(Argon,1bar),foraduringofatleast

15min; argon sparging duringthe experiments whichwereachieved

atroom temperature(18<T($_{C)<22),exceptfor experimentsunder}

ultrasonicirradiation,forwhichthetemperatureslightlyincreases

(

D

T<5$_{C).Solutionscontaining0.01Mofiron (II)werepreparedby} dissolvingNormapursolid FeCl2or(NH4)2Fe(SO4)2 (Mohr’ssalt)

supplied by Sigma-Aldrich. Solutions also contain respectively

0.1MKClor0.05MK2SO4assupportingelectrolytes,andtheirpH

wasadjustedto4.0,usingHClorH2SO4respectively.

The iron(II)was preferredtoiron (III)inordertoavoidthe

oxidationofelectrodepositedironbythelatter.Moreover,using

two different counter-ions (chlorideor sulfate)allows tostudy

theireffectonthemorphologyoftheelectrodepositediron.

FeCl2wasselectedbecauseitisasimple,solubleandcheapiron

(II) salt;it has already been used to study iron nucleationon

vitreouscarbonin[33].Furthermore,anotherwork[9]hasshown

thatironnanoparticlescouldbesynthetizedby

sonoelectrochem-istryusingFeCl3.

(NH4)2Fe(SO4)2complex,preferredtothesimplesaltFe(SO4)2,

was alsostudiedthankstoits‘chemical stability’againsttothe

oxidationbyoxygen.

Electrochemical experiments were performed in a classical

threeelectrodescell(50mL),eitherbycyclicvoltammetricscansor

bygalvanostaticelectrolyses;arotatingdiskwasusedasaworking

electrode (Radiometer,EDI101T),a platinumfoilconstitutesthe

counter electrode, and a saturated calomel electrode (SCE)

immersedwithinaLuggincapillaryisthereferenceelectrode.

Solutionwasalwaysstirredbytherotationofthediskelectrode

(angular velocity

v

=1000rpm). For cyclic voltammetry, the

potential scan rate r is equal to 20mV/s except for particular

cases where lower scan rates were used. For galvanostatic

electrolyses, theappliedcurrent isequal to90%of thelimiting

current correspondingtoFe(II)_{reduction which ismeasured on}

cyclicvoltammetryscans(withoutultrasound).

Depending onthe experiment, the electrochemical cell was

immersedinanultrasonicbath(FisherScientificFB15057,37kHz,

750W,6.9L).ApotentiostatVoltalabPGZ100from

Radiometer-Analytical isusedtoachievetheelectrochemicalmeasurements.

Voltamaster4istheelectrochemicalsoftwareusedtocontrolthe

potentiostat. Working electrode material is either gold (2mm

diameter,purity>99.99%)orvitreouscarbon1(VC1,3mmdiameter,

purity=99.96%)orvitreouscarbon2(VC2,3mmdiameter,purity=

99.99%). VC1 was provided by Carbone Lorraine Company and

VC2wasprovidedbyOrigalysCompany.Chemicalcompositionof

these VCelectrodesareslightlydifferent:0.04%ash contentand

50ppmsulfurandboretracesforVC1, 0.0042%ashcontentand

13.5ppmmetaltracesforVC2(dataprovidedbysuppliers).Themain

differencebetweenVC1andVC2isthatVC1hasfabricationdefects

(smallvoidinclusionsinitsvolume)andthusitssurfaceexhibits

holeswith diametersof about1-10

m

m(even afterpolishing at

0.3

m

m), while the VC2 has no defects. VC1 and VC2 surfaces

characterizationisgiveninsection3.3wheretheeffectofelectrode

surfaceroughnessonelectrodepositadhesionisdiscussed.

Electrodeswerepolishedusingaluminaaqueoussuspensionof

9,5,1and0.3

m

monarotatingpad.Theelectrodesweresonicated

for 5minutes in a 50:50 ethanol/water mixture between each

polishing and 5minutes more in ultrapure water to remove

remainingaluminaparticles.

2.2.Electrodesurfacecharacterization

Thesurfacesof theelectrodes werevisualized byanoptical

microscope(ZeissAxiolab)andsurfaceprofilesweremeasuredby

aninterferometricsurfaceprofiler(Zygo3D).Usingdataprovided

byZygo3Dmeasurements,Wenzel’sroughnessrw(definedasthe

ratio between actual surface and geometrical surface) can be

estimatedusingthefollowingequation:

rw¼1_A Z A ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1þrhðx;yÞ2 q dxdy _ð1Þ

whereAisthegeometricsurfacearea,histhelocalsurfaceheight

andx;yaretheCartesiancoordinates.rwwascomputedusinga

home-madeMatlabprogram.Arithmeticsurfaceroughnessrawas

computedbytheZygosoftware.

2.3.Measurementofsurfaceenergiesofelectrodes

Electrodesurfaceenergyorsurfacetension

g

elect(elect=Auor

VC1orVC2)isanimportantparameterallowingtheestimationof

both the work of adhesion of metallic iron on the electrode

materialinelectrolyticsolutionWFe=elect and alsotheinterfacial

tensionbetweentheelectrodeandtheiron

g

Fe=elect.Thesurface

tensionsofthegoldandVCelectrodes,

g

Au,

g

VC1and

g

VC2,were

obtainedbymeasuringthecontactanglesofaseriesofliquidsof

known surface tensions directly on the polished electrodes

surfaces(Owens-Wendt‘oneliquidmethod’[34]).Contactangles

ofultrapurewater, glycerol,dimethylsulfoxideanddecane were

measuredwith a GBXDigidrop goniometer DGD fast/60 using

!2

m

Lliquiddropletinambientconditions.Themeasurementof

contact angles was reproducible in the range )5$_{. Absolute}

uncertaintyon

g

electmeasurementislowerthan4mJ/m2.

2.4.Electrodepositmorphologyanalysis

MorphologyofironelectrodepositswasobservedbySEMwith

aMEBFEGJEOLJSM7100FTTLSoraMEBFEGJEOLJSM7800F

Prime-EDS. Before observation, the sample was rinsed with

ultrapure water (to avoid electrolyte crystallization) and dried

underambientconditions.Toavoidelectronsbeamdeviation(due

to the presence of insulating material surrounding the disk

electrode),thewholesamplesurfacewascoatedbyananometric

layer of gold by vapor deposition, if required. An electric

connectionbetweenelectrodesandtheSEMsupportwasinsured

byabrassstudadaptedtotheelectrode.

3. Resultsanddiscussion

3.1.Ironelectrodepositionbycyclicscanvoltammetryundersilent conditions

3.1.1.Fe(II)_{electrochemicalbehaviourongoldandvitreouscarbon}

electrodes

The electrochemical behavior of Fe(II) _{salts used (FeCl}

2 and

electrolytesand onthreedifferentelectrodespolishedwiththe

finestaluminasuspension(0.3

m

m):VC1,VC2andgold.

Voltam-perometriccurvesobtained(understirring)withoutFe(II)_(residual

currents,seeinsetsinFig.1aandb)revealadiffusion-limitedwave

attributedtothereductionofthefreeprotonsatpH=4:

Fig. 1.CyclicvoltammetryscansandthecorrespondingSEMpictures.a)andb):curves(forwardandbackward,seearrowsina))obtainedonarotatingdiskelectrodeforthree differentmaterials:VC1,VC2andgold.a):Mohr'ssaltsolution,b):FeCl2solution.Insets:Residualcurrents(forwardcurves).SEMpictures(takenattheendofthecathodic cycle,seethearrowona)andb)):1)to3)ironelectrodepositsusingMohr’ssaltsolutiononrespectivelyVC1,VC2andgoldelectrodes;4)to6)ironelectrodepositsusingFeCl2 solutiononrespectivelyVC1,VC2andgoldelectrodes.

H+_+e#

!1/2H2. (2)

Curves obtained with the electrolyte K2SO4 alone (inset in

Fig.1a)exhibit a resolute diffusion-limited wave,of which the

limitingcurrent,evenlowmagnitude(0.65mA/cm2_{),appearstobe}

constantforthethreeexaminedmaterials(samehydrodynamics).

ConcerningthecurveobtainedinpresenceofKCl(insetinFig.1b),

it doesnot showa clearsignalfor theH+ _{reductionbeforethe}

reductionofwater(2H2O+2e#!H2+2OH#),exceptforthegold

cathode.IndeedtheH+_{reductionontheVCisnotvisible(seeinset}

in Fig.1b)becauseitis shiftedtomorenegativepotentialsand

screenedbywaterreduction.Thisphenomenonhasalreadybeen

observedbyGrujicicandPesic([33])whostudiediron

electrode-positionfromsulfateandchloridesolutionsonVC.Moreover,for

the gold cathode, thecorresponding limiting current (0.34mA/

cm2_{)ishalfofthemagnitudeofsulfatesystemones(0.68mA/cm}2_).

Thisratioof2betweentheselimitingcurrentvaluesisattributed

tothedi-acidnatureofH2SO4comparedtothemono-acidnature

of HCl. Indeed, a solution of sulfuric acid and K2SO4 at

pH=4 contains thefollowing species: H+_{, HSO}

4#, SO42#and K+,

soafter theconsumptionof thefreeH+_,HSO

4#dissociates and

supplyadditionalH+_{whichcouldbereduced.Thisleads,forthe}

sulfatesystem,toanapparentH+_{concentrationtwotimeshigher}

thantheoneforthechloridesystem,andexplainswhyalimiting

current(whichisproportionaltoconcentration)thatistwotimes

higherhasbeenmeasuredforthesulfatesystem.

Valuesforlimitingcurrentdensity,jlim,canbeestimatedusing

Levich'sequation([35]): j_lim¼0:62:n:F:D2=3_:v#1=6_:C

Hþ:

v

1=2; ð3Þ

wheren,F,D,v,C_Hþand

v

arerespectivelytheelectronsnumber,

the Faraday constant, the diffusion coefficient, the kinematic

viscosity,theprotonsconcentrationandtheangularvelocity.

TakingthevaluesforD,C_Hþ,and

n

ofrespectively9.3+10#9m2/

s([36]),10#4_Mand10#6_m2_/sleadstoj

lim=0.33mA/cm2whichis

in accordance withthe measuredvalueof jlim for the chloride

system(0.34mA/cm2_,C

Hþ=10#4M).

Inthepresenceofironsalts(Fig.1aandb),allcurves(1000rpm,

20mV/s) exhibit an additional signal attributed to the Fe(II)

reduction:

Fe(II)_+2e#_!Fe(0)_{. (4-a)}

Forallexaminedelectrodematerials(gold,VC1andVC2),the

firstscan(startingat+0.2V/SCEtowardcathodicpotentials)clearly indicatesaslowredoxsystem,reductionsofFe(II)_{toFestartingat}

around-1.2to-1.3V/SCEontheinitiallybareelectrodes.Duringthe

firstscan,vitreouscarbonsandgoldarepartiallycoveredby

zero-valentiron,sooncetheswitchingpotentialisreached,substrates

tendtoactasanironelectrode.Indeedatthepotentialscanrate

used, cathodiccurves obtainedduring the return scan (-1.4 to

-0.4V/SCE),exhibitaresolutediffusion-limitedwave,

correspond-ingtotheFe(II)_{reductionsonthenativeiron surface.It canbe}

noticedthatforMohr'ssaltsolutions(Fig. 1a),thediffusion-limited

wave(pseudo-plateau)isnotwelldefinedduetotheco-reduction

of free protons on electrodeposited iron followed by water

reduction.Thisdistinctionbetweenchlorideandsulfatesystems

forironelectrodepositionhasalreadybeenobservedbyGrujicic

andPesic([33]).

Forthebackwardscans,thecathodiccurrentwascanceledat

theequilibriumpotentialswhichlieintherange[!-0.9to!-0.8V/

SCE]fortheKClmediumand[!-1to!-0.9V/SCE]fortheK2SO4

medium.ThesepotentialsremainrelativelyfarfromtheFe2+_/Fe

Nernst'spotentialE=-0.44-0.25+0.03x(log(10#2_/1))=-0.74V/

SCE.Thiscouldbeexplainedbythefollowingreasons:

-theFe(II)ionscomplexedwithchlorideorbyammonium/sulfate

ions(Mohr'ssalt),sotheredoxsysteminvolvedisdifferentfrom thesimpleFe2+_/Fesystem

-co-reduction of free H+ _{(even at the equilibrium potential)}

disturbstheredoxpotential oftheFe2+_{/Fesystem duetothe}

increaseofthepHatthecathode.

The anodic curves obtained during the backward scan for

potentialshigherthan-0.8V/SCE,exhibitpeaksattributedtothe

ironelectrodepositoxidation(forbothKClandK2SO4mediaand

forthethreesubstrates,Fig.1aand1b): Fe(0)

!Fe(II)_+2e#_{. (4-b)}

The sharp decrease in current, clearly indicates complete

oxidationoftheelectrodepositandtheregenerationoftheinitially

baresurfaceoftheworkingelectrode(gold,VC1andVC2).Thisis

theendofonecompletecycleoftheoperatedexperiments.

3.1.2.Influenceofelectrodematerialonironelectrodeposit morphology

Ironelectrodeposits,obtainedonthethreesubstratesandfor

bothchlorideandsulfatemedia,wereobservedbySEMattheend

ofthecathodiccycle.TheresultingpicturesareshowninFig.1,

pictures1–6.

Inchloridemedium,a clearmorphologicaldifferencecanbe

noticedbetween iron electrodeposits obtained on goldand on

vitreouscarbonelectrodes(Fig. 1,pictures4–6).Indeed,onthegold

electrode (Fig. 1, picture 6), the SEM image reveals iron

electrodepositasacompactlayerthatcoverswelltheelectrode

surface.Thislayerconsistsofcubicstructuressizedintherangeof afewhundrednanometers.Conversely,ontheVCsubstrates(Fig. 1,

Fig.2.SEMpicturesofironelectrodepositsformedafteracathodiccyclebyvoltamperometry(startingat0.2V/SCE,reversedat-1.3V/SCE,stoppedat-0.8V/SCE)using Mohr’ssaltsolutionforthegoldelectrodea)andfortheVC2electrodeb)respectively.

pictures4and5)ironelectrodepositsappearassmallmicrometric

dendriticstructureswithheightsnotexceeding1

m

m.

InMohr’ssaltmedium,ironelectrodepositedongold(Fig.1,

picture3)andonVCsubstrates(Fig.1,pictures1–2)presentquite

similarmorphologywithathinlayerofironcoveringthesurface

(verified by Energy Dispersive X-Ray Spectroscopy) with iron

micrometricstructuresonit.However,amoredetailedanalysisof

thisthinlayerofiron(Fig.2)showsthatitcoversmoreonthegold electrode(Fig.2a)thanontheVC2electrode(Fig.2b).Indeed,on

theVC2electrode,thethinlayerofironisnotcontinuous,some

holes(darkspots)arevisiblerevealingtheVCsubstratethrough

them.On contrary,onthegoldsubstrate,thislayer isfarmore

continuous.

IronelectrodepositmorphologydependsontheFe(II)_saltand

thesolutioncompositionaswellastheelectrodematerial.Ithas

beenshownthatspecificcounter-ionsadsorptioncouldinfluence

electrodepositmorphology([37]).Here,wefocusontheeffectof

theelectrodematerialontheironelectrodepositmorphology,and

ourobservationsshowthatelectrodepositedironhasmoreaffinity

withgoldthanwithvitreouscarbonelectrodes.Ironspreadsbetter

ongoldthanonVCduringitselectrodepositionforbothmedia

used.Ithastobenoticedthatnodifferenceisobservedbetween

electrodepositsonVC1andthoseonVC2.

Fromtheelectrodepositiontheory([38]),itisshownthatthe

growthmode (from2D to3D)ofa crystalstronglydepends on

theinterfacialenergyofinterfacebetweentheelectrodepositand

the electrode

g

Fe=elect (with elect corresponding either to Au

electrodeortoVCelectrodes).Crystalsobtainedforlowvaluesof

g

Fe=electarealmostflat(2Dgrowth)andconverselycrystalsgrow

preferentiallyperpendicularlytotheelectrodesurface(leadingto

poorlycoveredelectrode surface)forhighvaluesof

g

Fe=elect (3D

growth).

Contactanglemeasurementswereachievedfor goldandVC

substrates, see Table 1. Using a theoretical development, the

differencebetween

g

Fe=VC and

g

Fe=Aucanbeestimatedfromthe

measurementsofsurfaceenergies(seetheAppendix).Apositive

andnon-negligiblevaluelyingintherange[205.4mJ/m2_;58.9mJ/

m2_{]wasdeducedfor}

g

Fe=VC#

g

Fe=Au whichisinaccordancewith

themorphologicalobservationsofironelectrodeposits.

Voltamperometriccurvesindicatedin Fig.1a)and 1b),show

thattheoverpotentials

h

requiredforironreductionontheinitially

baregoldelectrode(thefirstforwardscan)

h

Auaresystematically

lowerthan

h

VC1!

h

VC2 forboth mediaused.Therefore,alower

energy(lowercellvoltage)isrequiredtocreate/increasetheFe/Au

interfacethantheFe/VCinterface.Thisisinaccordancewiththe

previous analysis showing that a higher energy is required to

increasethesurfaceareaoftheFe/VCinterfacethantoincreasethe

one of the Fe/Au interface during iron electrodeposition

ð

g

Fe=VC>

g

Fe=AuÞ.

To conclude,during a cathodiccycle onVC electrodes, iron

electrodepositsbegintogrowfollowinga3Dmode,leadingrapidly

toadendriticgrowth.Onthegoldelectrode,ironelectrodeposits

begintogrowfollowinga2Dmode(thinfilmgrowth)whichdelays

theappearanceofthedendriticgrowthmodeasitisclearlyvisible

inthecaseofchloridemedium(Fig.1,pictures4–6).

3.1.3.Influenceofironelectrodepositmorphologyonlimitingcurrent ofFe(II)_reduction

The differences in iron electrodeposit morphologies, as

discussed above, can be revealedby a detailed analysis of the

voltamperogramsforbothmedia(chlorideandsulfate).

Inchloridemedium,thegreatdifferenceinironmorphology

observed for gold and VC electrodes (see Fig.1, pictures 4-6)

explainstheslight,butsystematicdifferenceoflimitingcurrent

values, measured for both substrates (cathodic backward scan,

Fig.1b).Aspreviouslymentioned,onthegoldelectrode,theSEM

imagerevealsquasitotalcoverageoftheelectrodesurfacebyathin ironlayer.Therefore,theelectroactivesurfaceareaforthegolddisk

canbeconsideredasthegeometricalsurfacearea(seeaschematic

representationoftheelectrodepositsinFig.3aforgoldand3bfor VC).Conversely,ontheVCelectrodes(Fig.1,pictures4and5)iron

electrodeposits consist of relatively small dendrites of which

heights (<1

m

m) are lower than the thickness of the diffusion

boundary layer thickness (estimated at !14

m

m). For the VC

electrodescase,electroactivesurfaceareaislimitedtothesumof

surface areas of the growing dendrites tops(projected surface

area) which is lower than the geometrical surface areaof the

electrode(seeFig.3b).ThefactthatthelimitingcurrentsofFe(II)

reductioninchloridemedium,usingVCelectrodes,arelowerthan

thoseusinggoldcanbesimplyexplainedbya reduced

electro-activesurfaceareaassociatedwithirondendriticgrowthontheVC

electrodes,butnotonthegoldelectrode.

In Mohr’s salt medium, similar magnitudes of the

pseudo-limiting current were measured for gold and VC electrodes

(cathodicbackwardscansinFig. 1a,correspondingtothereduction

ofFe(II)_{onthesubstratecoveredbytheironelectrodeposit).This}

differentbehaviouris explainedbytheveryslightdifferencein

iron electrodeposit morphologies (comparatively to chloride

medium)obtainedongoldandonVCelectrodes(Fig.1,pictures

1–3).

3.1.4.Fe(II)_/Fe(0)_{kineticsonthegoldelectrode}

Inadditiontothedifferencesobservedinthelimitingcurrent

forthethreeelectrodematerials,differencesalsoappearbetween

goldandVCelectrodesforlowcurrents,onthebackwardcathodic

curves (Fe(II) _{reduction) on electrodeposited iron; while these}

curvesdonotoverlapinchloridemedium(Fig.1b),theyoverlapin sulfatemedium(Fig.1a).Moreover,inchloridemedium(Fig.1b),

thebackwardcurvesforVC1andVC2overlapandexhibitlower

overpotential(activationareastartingat!-0.9to!-1.05V/SCE)in

comparison to the backward curve obtained using the gold

electrode (higher overpotential, activation area starting at

!-0.95 to !-1.15V/SCE). The iron electrodeposits morphology

(Fig.1, SEM pictures 4–6) could explain the surprising lower

overvoltageobservedonVCelectrodes;indeed,theiron

electro-depositonthegoldelectrode(picture6inFig.1)ismoreuniform

but not really ‘massive’, while iron electrodeposited on VC

electrodes (pictures 4 and 5 in Fig. 1) presents micrometric

dendritesinametalliciron‘massive’orbulkform.

Theeffectofthepotential scanrate onthecurrent-potential

curvesshape,aswellasonthedepositstructure,forthereduction

of Fe(II) _{on gold substrate, was examined in Fig. 4. Decreasing}

potential scan rates(20!8!2mV/s)ledtoan increasein the

amount of iron produced during the cathodic scan

(5.7!14.7! 62.7 mC). Consequently, SEM pictures in Fig. 4b

and4cofirondeposit,producedonthegoldsubstrateduringa

cathodicscan,showthat increasingtheamountof irondeposit

leadstoachangeinitsstructure.

Aspreviouslymentioned,irondepositsproducedongoldduring

thecathodicscan,spreadontheelectrodesurfaceasathinlayer.The

SEM picture (Fig. 4c) reveals that for a scan rate of 20mV/s,

thethicknessofthedepositisabout90nmwhichisclosetothe

Table1

SurfaceenergiesofVC1andVC2electrodes.gd and gp are respectively the dispersive and the polar components of the total surface energy

gelect¼gdþgp.

gd(mJ/m2) gp(mJ/m2) gelect(mJ/m2)

VC1 19)2 10)2 29)2

calculated thickness: equivalent compact layer thickness e¼ðMFeR I:dtÞ=ð2F

r

FeSdiskÞ=67nm. This confirms the spreading

ofironanda2Dgrowthofthedepositonthegoldelectrode.A3D

growthofthedepositisobtainedonlywhenalargerquantityof

ironisdeposited(Fig.4b)forlowscanrates8mV/sand2mV/s.Asa

firststep,ironspreadsonthegoldsurfaceandoncethesurfaceis

totallycovered,deposit startstogrowthfollowinga 3Dgrowth

modeanddendriticmorphologyappears.Indeed,decreasingthe

scan rate from 20mV/s to 8mV/s leads to a change from a

homogeneousandultrathindeposit(Fig.4c)toamicrometricand

dendriticdepositthatcanactasabulkironelectrode(Fig.4b). TheI=f(E)curvesobtainedforthethreescanrates(Fig.4a)are

similartothoseindicatedinFig.1b;theforwardcathodicscans

exhibit a diffusion-limited plateau at -1.2V/SCE for which the

magnitudeofthelimitingcurrentisnotaffectedbythepotential

scanrate.Thismeansthatforthethreecurvesthesystemoperates

inasteadystate;thelimitationofthecurrentisduetotheconstant

agitation applied. Concerning the backward cathodic scan, the

curvesshowthat,decreasingthepotentialscanrate(20!8!2

mV/s)causes thehalf-wavepotential of theFe(II)_{reduction on}

native iron, to shift to the anodic values (-1.014!-0.992!

-0.938V/SCE).Thisfactsuggestsasystemwhichtendstobecome

morereversible(toreachthebehaviourofa‘bulkiron’electrode)at

low scan rates, because for longer electrolysis durations the

amountofirondepositedishigherand adendriticstructure(of

micrometricsize)isobtained.

Thereforetheshapeofthecathodiccurvesisdictatedbythe

structureofirondepositedandthereforebythequantityofiron

produced.

Potentialscanratealsoaffectsthepotentialoftheoxidationof

theiron-depositsignal;themorethescanratedecreases(20!8

!2mV/s),themoretheoxidationoftheirondepositedonthegold

electrode becomes easier (-0.550, -0.566 and -0.598V/SCE),

suggestinga‘bulkironbehaviour’.

Itcanbeconcludedthattheelectrodepositobtainedat20mV/s

issothin(67nm)thatitselectrocatalyticpropertiesareaffectedby

the electronic collector (gold) and the Fe(II) reduction (and

oxidation)appearsasaslowersystem.Thiseffectisnotobserved

onVCelectrodesbecause,evenat20mV/s,thecorrespondingiron

electrodepositsconsistsofmicrometricirondendrites,whichare

inabulkironform(Fig.1,SEMpictures4and5).

3.2.Ironelectrodepositionbycyclicscanvoltammetryunder

ultrasonification

TheelectrochemicalbehaviourofFe(II)_during

ultrasonification

wasinvestigatedinchlorideandsulfatemediaandforthethree

electrodespolishedwiththefinestaluminasuspension(0.3

m

m):

VC1,VC2andgold.Theelectrochemicalcellwasimmersedinthe

Fig.3. Schematicrepresentationoftheironelectrodepositmorphologiesobtainedinchloridemedium.a):onthegoldelectrode(relatedtotheSEMpictures6inFig. 1),b):on theVC1andVC2electrodes(relatedtotheSEMpictures4–5inFig.1).

Fig.4. a)Potentialscanratedependenceofthevoltamperometrycurves(forwardandbackward,seearrowsina))obtainedonarotatinggolddiskelectrode,immersedinthe FeCl2solution.SEMpicturesofironelectrodepositobtainedongoldafteracathodiccycle(startingat0.2V/SCE,reversedat-1.25V/SCE,stoppedat-0.8V/SCE)atb)8mV/sand c)20mV/s.

ultrasonicbathandspecialcarewastakentoalwaysputthecellin

thesameplaceinthebathsoastogeneratethesameultrasound

intensity into the cell. Cyclic voltamperometry was performed

underultrasonificationorundersilentconditionsandthesolution

was continuously stirred(

v

=1000rpm, potential scan rate=20

mV/s).ResultsaregiveninFig.5.Aspreviously,forwardscanstarts

fromtheopencircuitpotential(OCP!+0.2V/SCE)andgoestothe

cathodicdirection(to-1250mV/SCEforthegoldelectrodeandto

-1350mV/SCEforVCelectrodes);thenthescanwas reversedto

potentialsofaround0V/SCEinordertogetthebackward‘cathodic

andanodicparts’oftheI=f(E)curve.

3.2.1.Effectofultrasoundonmasstransportduringironelectrodeposit growth

Current-potential curves obtained under ultrasonification

(Fig.5)exhibithigherFe(II)_{reductionlimitingcurrents(obtained}

duringbackwardscan)thanthoseobtainedundersilentconditions

(withtheexceptionofcurrent-potentialcurvesinFig.5cwhich

willbediscussedbelow).Thisisduetoultrasoundsthatinduce

cavitation bubbles and acoustic streaming, causing additional

convection near the electrode and then increasing the

mass-transport flux of the electroactive species [20–25,19].Current

fluctuations,that arehighly visible,especially onthe

diffusion-limited plateau of Fe(II) reduction (see Fig. 5a, b, eand f),are

attributedtosuccessiveeventsofthecollapseofcavitationbubbles

Fig.5.Cyclicvoltammetryscans(forwardandbackward,seearrowsina))obtainedonarotatingdiskelectrodeforthreedifferentmaterials:gold(a)andd)),VC1(b)ande)) andVC2(c)andf)).Curvesina),b)andc)obtainedwithMohr'ssaltsolution;curvesind),e)andf)obtainedwithFeCl2solution.Dashedlines:silentconditions,solidlines: ultrasoundactivated.TheratiobetweenthecathodicchargeQc, and the anodic charge Qa, is given for each casein the legend.

closetotheelectrodesurfaceinducinglocalandtransientvigorous stirring([21,26,27]).Itisinterestingtonotethatforcurrentlower

thanthelimitingcurrent,intheactivationarea(duringbackward

scans, reduction of Fe(II) _{on Fe}(0)_{), current}

fluctuations are not

observed(Fig.5a,b,c,eandf)andthecurves,withandwithout

ultrasound,tendtooverlap(especiallyinFig.5bande).Thisisin

agreementwithpreviousstudiesshowingthatultrasoundsaffect

electrochemical processes mainly via mass-transport

enhance-mentandonlyslightlytheelectrodekinetics[39,20,30].

TheincreaseinlimitingcurrentofFe(II)_{reductiononthegold}

electrode in chloride medium, when ultrasounds are activated

(Fig.5d),islow(relativeincreaseof7%), comparedtotheother

onescorrespondingtocurvesinFig.5a,b,eandf(relativeincrease

from15to60%).Onthebasisofelectrodepositedironmorphology

obtainedbySEMimaging(Fig.1,pictures1–6),itappearsthatthe

electrodepositthat hasgrownonthegoldelectrode inchloride

mediumandundersilentconditions(Fig.1,pictures6)istheonly

onetogrowasathinfilm.ForallotherFe(II)saltsandelectrode

material combinations, electrodeposited iron presents

micro-metric dendrites reaching about 1

m

m height (Fig. 1, pictures

1–5).Submittingthesolutiontoultrasoundscausesanincreasein

thelimitingcurrentassociatedwiththedecreaseindiffusionlayer

thickness [21,22]. Furthermore, the fast growth of the iron

electrodeposits probablyleadstohigherdendrites withheights

greaterthan1

m

m,similartothethicknessofthediffusionlayer.

Consequently,theelectroactivesurfaceareaincreasesleadingtoa

currentmagnitudeincrease(seetheschematicrepresentationin

Fig.6).Forelectrodepositgrowthasafilm(thedendriticgrowthis

delayed and such a couplingis notexpected), theelectroactive

surfacearearemainsequaltothegeometricalsurfaceareaofthe

electrodeandthecurrentincreasesonlyduetotheenhancement

ofmasstransportbyultrasounds.Therefore,thisanalysisallowsto

giveanexplanationfortheobserveddifferencesin thelimiting

currentincreaseswithultrasoundsbetweenthethinfilmgrowth

modeandthedendriticgrowthmode.

3.2.2.Effectofultrasoundonironelectrodepositsdispersion

Theintegrationofbothcathodic(forwardandbackward)waves

andanodicpeak(Fig.5)allowtogettheamountofchargeforFe(II)

reduction,Qc,andforFe(0)_{oxidationQa.Thecomputedvaluesof}

theratioQc=QaaregiveninlegendsofFig.5.Qc=Qavalues,higher than1inallcases,areusedbelowascriteriontodiscusstheeffects

of adhesion and ultrasound on iron electrodeposits dispersion

fromtheelectrode.

Inabsenceofultrasonification,Qc=QaobtainedinFeCl2(!1.10)

islowerthanthoseobtainedinMohr'ssalt(!1.35),becauseofthe

co-reduction of bothH+_{and partially ofwater during theFe}(II)

reduction(asalreadysaidinsection3.1.1).Notethatthecurrent

consumptionbyH+_{andwaterreductionconstitutesa drawback}

(lossinfaradicyieldforFe(II)_{reduction),butalsotwoadvantages:i)}

consumption of H+ _{in the diffusion layer, so limitation of the}

corrosionofzero-valentirondeposit,andii)productionofalittle

partofH2whichallowslimitationoftheoxidationbytheresidual

oxygen.

Concerningtheadhesion/dispersionoftheironelectrodeposit

onthegoldelectrode(Fig.5aandd),theeffectoftheultrasounds

appears minor in both FeCl2 and Mohr’s salt media; indeed

comparisonoftheratioQc=Qa(seelegendsinFig.5aandd)with

ultrasonification(1.31inMohr'ssalt;1.08in FeCl2)andwithout

(1.36 in Mohr’s salt; 1.08 in FeCl2) shows very similar values,

demonstratingthatallironelectrodepositedduringFe(II)_reduction

remains on the gold surface and oxidizes totally during the

backwardanodicscan.Conversely,forVCelectrodes(seeFig.5b,c,

eandf),ironelectrodepositsdispersionisobservedineverycase,

Qc=Qaobtainedwithultrasoundisatleastequalto1.6anditis

alwayshigherthanQc=Qavaluesobtainedundersilentconditions

(Fig.5b,c,eandf).

NotethatfortheelectrolysesusingtheMohr'ssaltsolutionand

theVC1electrode(Fig.5b),eveniftheremovalofthedepositis

observedQc=Qa=1.6),theanodicpeak(andalsotheamountof

charge),whenultrasoundsareactivated,ishigherthantheone

obtainedundersilentconditions.Asdiscussedin[30]and[31],this

isduetotheenhancementofmass-transferbyultrasoundleading

tolargerdepositionrateandthenalargerquantityofironisformed

ontheelectrodecomparedtothesilentconditioncaseevenifsome

partofthedepositisdispersed.

Ironelectrodepositappearstoadheremuchmoreonthegold

electrodesurfacethanontheVCelectrodessurfaces,becauseof

thehighadhesionenergybetweenbothmetals;thesoliddeposit

requires higher ultrasonification power to be removed and

dispersed in the liquid. These results are in agreement with

estimationofworksofadhesionpresentedintheAppendix:work

of adhesion of iron on gold WFe=Au=148.7mJ/m2 and work of

adhesionofirononVCWFe=VC=17.3mJ/m2.

ConcerningVC1electrode(Fig.5bande),similarcurvesand

Qc=QaratioareobtainedinMohr’ssaltandinFeCl2.ForMohr’s

salt,Qc=Qareaches1.6withultrasound,andthemainpartofthe

irondeposit(62.5%) remainsontheVC1electrode surfaceeven

underultrasonification.ForFeCl2,ultrasonificationappearstobe

Fig.6.Schematicrepresentationsofpossiblediffusionlayerconfigurationsforthetwomaintypesofironelectrodepositmorphologyencountered,thinfilm(a1andb1)and dendritic(a2andb2),undersilentconditions(a1anda2)andwithultrasonification(b1andb2).

moreefficient,Qc=Qareaches3.45withultrasounds.Nevertheless,

ultrasonification effect on iron deposit dispersion appears less

efficient onVC1 than on VC2 (see Fig 5b, c, e and f). Indeed,

VC2electrodeledtothemostinterestingresults(Fig.5candf)

particularlywithMohr'ssaltsolutions(Fig.5c).Theanodicpeak

correspondingtotheremainingiron,electrodepositedduringthe

cathodicscan,ismissing,indicatingthatalltheironelectrodeposit

wasremovedfromtheelectrodesurfaceandQc=Qareaches240

(instead 1.4 without ultrasound)! In addition, the backward

cathodicscan(Fig.5c)tendstooverlaptheforwardcathodicscan,

meaningthatFe(II)_{reductiontakesplaceonanalmostclean/bare}

surface,whereironelectrodepositsarerapidlyandcontinuously

removedthankstoultrasounds.Ironelectrodepositsadhesionon

VC2electrodesurfaceisveryweakandtheyaredispersedeasily

duringtheirformation.Thesameeffect,evenlessimportant(in

comparisonwiththeMohr’ssalt),isobservedwithFeCl2onVC2

(Fig.5f),Qc=Qareaches4.56underultrasonification.

A similar analysis of the evolution of the ratio Qc=Qa as a

functionoftheultrasoundintensityhasbeenachievedin[30]for

the electrodeposition of metals (Zn, Co, Pb and Hg) on a VC

electrodeusingaface-onconfiguration.Theoverlappingofboth

thecathodicbackwardandtheforwardscans(asobservedinthe

presentstudy,Fig.5c)wasnotobservedinthislaststudy,evenfor

thehigherultrasoundintensity.Thissuggeststhat

electrodepos-itedmetalsparticleswerealwayspresentonVCelectrodesurface

duringthecathodicpartofthescan.Asdiscussedin[30](seealso

[31]),thedispersionratewasnotsufficientlyhightoovercomethe

deposition rate during the cathodic part of the scan. The

experimental configuration, used in the present study, clearly

shows that iron particles can be very rapidly removed by

ultrasound allowing to operate continuously (with an almost

‘bare’substrate).

Thislastobservationisimportantfromapracticalpointofview becauseitshowsthatitispossibletodrivetheelectrodepositionof

ironsimultaneouslywithitsalmosttotaldispersionthatshould

leadtothe continuoussynthesis of fineiron particles(limiting

theirgrowth).Onanotherside,asitcouldbeseeninFig.5c,the

cathodicchargeislowerthanintheothercases,leadingtoalower

synthesisrate.Asdiscussed,in[30]and[31],thecompetitiveeffect

ofmetaldepositionand itsdispersioncouldbeadjusted bythe

control of both the Fe(II) _{concentration and the ultrasound}

intensity.

Tosumup,maindifferencesinthedispersionofiron

electro-depositedongoldandVCelectrodes,areduetoa differencein

workofadhesionbetweenironandthesesubstrates.

However,differencescanalsobeobservedbetweenapparently

similarVCelectrodes:VC1andVC2.Dispersionbyultrasonification

ismoreefficientonVC2thanonVC1,suggestingthatironadheres

lessonVC2thanonVC1.Todemonstratethisassumption,contact

anglemeasurementwereachievedforVC1andVC2,andresults

wereindicatedinTable1,giving

g

VC1and

g

VC2.

ThevaluesofthesurfacetensionsofVC1andVC2electrodesare

very similar (

g

VC1!

g

VC2, averaged value=30mJ/m

2_).

g

VC2 is

foundtobeslightlyhigherthan

g

VC1probablybecauseofaslight

differenceintheirchemicalcompositions(seesection2.1)andthis

couldinduceabetteradhesiononVC2thanonVC1(seeAppendix).

Nevertheless,experimentshaveclearlyshownabetteradhesion

onVC1thanonVC2,implyinganotherparameterasresponsibleof

thisobserveddifference.TheeffectoftheVCelectrodessurface

roughnessisanalyzedinthefollowingsections.

3.3.InfluenceoftheroughnessoftheVCelectrodeonthedispersionof electrodepositediron

Theroughnessofthesurfaceof theelectrodeis aparameter

whichcouldinfluencetheadhesionofiron.BothVCelectrodes’

surface were characterized by optical microscopy and by an

interferometricsurfaceprofiler(Zygo3D),Fig.7.Thesmallvoid

inclusions withinthe VC1 (visible asholes onits surface with

diametersintherange1to10

m

m,Fig.7band7-2)correspondto

manufacturingdefects.Conversely,VC2doesnotexhibitholeson

itssurface;remainingpolishingstripesarevisibleonthesurface

profileandtheirheightsremainlessthan30nm(Fig.7-1).

Decreasing the roughness of the electrode surface enables

easier electrodeposit dispersion in the liquid. In order to

specifically study the effect of the electrode roughness on

electrodeposit dispersion, independently ofiron growth, in the

nextsectionwefocusonthedispersionkineticsofiron

electro-depositedundersilentconditions.

Iron deposition was achieved by galvanostatic electrolysis

undersilentconditionsusingtherotatingdiskelectrode(

v

=1000

Fig.7. Opticalimagesandsurfaceprofilesofthepolished(aluminasuspensionat0.3mm)VCelectrodesusedinthepresentstudy;forVC2a)opticalimageand1)surface profile;forVC1b)opticalimage,2)surfaceprofile.

rpm) and the sameelectrolytes composition. In order toavoid

significant hydrogen production, the applied current, Iapplied,

correspondsto90%ofthelimitingcurrent,Ilim,obtainedduring

thebackwardcathodicscanundersilentconditions.Theduration

of electrolysis

D

t was chosen in order to obtain an iron

electrodeposit withan equivalentiron-compact-layer of 85nm.

Afterelectrolysis, therotationofthedepolarizeddiskelectrode

(cathode disconnected from the anode), was maintained and

ultrasonificationwasimmediatelyactivated(ornot)foraspecific

durationtUS.Inordertoevaluatetheeffectofultrasonificationon

the dispersionof the deposit,an anodic scan was immediately

carried out after the ultrasonification phase (from the OCP

!-0.730V/SCE to +0.2V/SCE at 20mV/s, maintaining electrode

rotationandwithoutultrasonification).

Theamountofcharge(Qa)correspondingtotheoxidationof

theremainingdepositontheelectrodesurfacewasthenevaluated.

NotethatQawascomparedwithanamountofcharge,thuscalled

Qaref,whichwasmeasuredduringa‘referenceanodicscanwithout

ultrasounds’;Qarefcorrespondstotheoxidationoftheirondeposit

not submittedtoultrasonification(tUS=0)afterelectrolysis.The

results obtained for the operating parameters investigated are

presented in Fig. 8, where the ratio (Qa/Qaref) is plotted as a

functionoftUS.Points oneachcurvecorrespondtotheaverage

valuesQa/Qarefmeasuredforatleastthreeexperiments;errorbars

correspondtostandard deviations.Asindicated above,theiron

depositonthegoldelectrodeisnotaffected(orveryslightly)by

ultrasounds; thereisnodispersion.Forthis reason,thissection

focusesonthevitreouscarbonelectrodesVC1andVC2only,andit

presentstheeffectofbothprecursors(FeCl2andMohr’ssalt)onthe

dispersionofthesolidirondeposit.ThecurvesQa/QarefversustUS

(Fig. 8) are analyzed with the help of SEM pictures of the

electrodepositedironnotsubmittedtoultrasonification(insetsin

Fig.8).

3.3.1.Ironelectrodepositsmorphology

The Morphologyof iron electrodeposits formedby

galvano-staticelectrolysis(insetsinFig.8)isdifferentfromthoseformedby

voltammetry(Fig.1).ForFeCl2andbothVC1and VC2(insetsin

Fig. 8a and b), electrodeposited iron morphologies consist of

micrometric‘rounded particles’and submicrometric iron

struc-tures covering the remaining VC surfaces, while dendritic

structures were obtained with voltammetry. It can be noticed

thattheroundedparticleshavecoalescedinsomecases,forming

biggerflakes.ThisisclearlyobservedontheVC2electrode(insetin

Fig.8b).

Themorphologiesobtainedbygalvanostaticelectrolysisappear

more‘homogeneous’ (lessdendritic)than theonesobtainedby

voltammetry.Amongothereffects,thiscouldbeattributedtothe

electrodepositionmodeandthecathodiccurrentemployed;with

voltammetry, electrodeposition is conducted under continuous

evolutionoftheappliedvoltageandtheresultingcurrent(which

reacheditslimitingvalue),whilegalvanostaticelectrolysescarried

outwitha currentclose(90%) tothelimiting current.Asoften

observedingalvanostaticelectrodeposition,electrodeposittends

to be dendritic/powdery when applied current is equal to, or

higherthanthelimitingcurrent([40–42]).Evenwhentheapplied

currentisjustbelowthelimitingcurrent,a‘lessrough’and‘less

dendritic’electrodepositmorphologycouldbeobtained[41,42].

ForMohr'ssaltandbothVC1andVC2(insetsinFig.8candd),

metallicirontakestheformofmicrometricstructuresasforthe

depositionobtainedbyvoltammetry;buthere,thesemicrometric

structures seem to have grown under the form of a ramified

Fig.8.DimensionlessironquantityQa/Qarefremainingontheelectrodeasafunctionoftheultrasonificationduration(tUS).Experimentsa)(VC1)andb)(VC2)wereachieved intheFeCl2solution.Experimentsc)(VC1)andd)(VC2)wereachievedintheMohr’ssaltsolution.PointscorrespondtotheaveragevaluesQa/Qarefmeasuredforatleastthree experiments;errorbarscorrespondtostandarddeviations.Insets:SEMimagesoftheelectrodepositsjustaftertheelectrolysis(notexposedtoultrasonification).

structurecoveringallthesubstratesurface.FortheVC2electrode (insetinFig.8d),aproportionoftheirondepositpartiallypeeled

off the electrode surface, probably solely due to the effect of

electroderotation.Comparatively,theiron depositfromMohr’s

salt on VC1 (inset in Fig. 8c) consists of the same deposit

morphologybutitremainsstuckontheelectrodesurface.

IntheFeCl2media,itseemsthattheseironparticles(especially

theflattenedones,insetin Fig.8b)alsopartiallypeeledoffthe

electrodesurfaceinthecaseofVC2electrode.Therefore,theseSEM

images, taken just before theultrasound phase, showthat the

adhesion of iron electrodeposits to the electrode seems to be

weakerfortheVC2thanforVC1.Indeed,inthefollowingsection,

thisobservation isconfirmedbyelectrochemicalmeasurements

indicatedontheplotofQa=Qaref asafunctionoftUS

3.3.2.Dispersionbyultrasonificationofiron‘electrodepositedunder silentconditions’

3.3.2.1. Ultrasound action for two VC substrates and two salt

precursors.

Theeffectofthesubstrate(VC1andVC2)ontheirondispersionfor

both FeCl2 and Mohr’s salt precursors were examined. All the

curves(exceptFig.8b)exhibitadecreaseoftheratioQa=Qaref asa

functionoftUS,meaningthatthequantityofthedispersediron

increasesprogressivelywithtimeduringultrasonification.

Forallexperiments,anultrasonificationdurationof60sallows

theremovalofmost(!90%)oftheirondeposit,anddispersesit

intotheliquid.StandarddeviationoftheratioQa=Qaref hasbeen

determinedin both mediaand with both substrates. For short

ultrasonificationtime,resultsarenothighlyreproducible,butthey

becomemorereproducibleforlongerdurations.Furthermore,the

quantityoftheirondepositremainingonthesubstratedecreases

progressively with time; this is particularly visible for the

VC1electrode(Fig.8a andc). FortheVC2electrode, dispersion

istoofasttoobservethisbehavior,especiallyinthecaseofFeCl2

medium(Fig.8b).

The progressive removal of iron from the VC1 electrode is

explainedbytheoperatedmodeofultrasonicdispersion.Indeed,

constant-power-ultrasound produces cavitation bubbles in the

liquidata constantrateperunit ofvolume.However,onlythe

collapsesofcavitationbubblesinthevicinityoftheirondeposit

surfacecancauseitslocaldispersion.Thefluxofremovedparticles

shouldthenbeproportionaltothequantityofirondepositpresent

ontheelectrodesurface.Thus,decreasingtheironquantityatthe

surface,reducestheremovedparticlesflux,asobserved

experi-mentally. A simple first order kinetics model can be built to

quantifythisbehavior:

d dt Qa Qa_ref % & ¼#s_QaQa ref ; ð5Þ

withsbeingthedispersionrateconstant(s#1_{)thatshoulddepend}

on the ultrasound power, on the adhesion energy

Eadh¼WFe=VCSFe=VC (SFe=VC beingthesurface areaof theiron/VC

interface) and on the size and distribution of particles on the

electrode. Equation (5) leads to the exponential decrease: Qa/

Qaref(t)=exp(#st).Theplotofln(Qa/Qaref)versustUSusingdata

fromFig.8aandcareshowninFig.9.Dispersionaroundthelinear

bestfitcurvesisobserved(duetotheresultsthatarenottotally

reproducible).Nevertheless,itshouldbepointedoutthatthisvery

simplemodelcapturestheoveralldynamicofthephenomenon.

This resultis in agreementwithresults fromother studies,

mentioning a progressive ‘erosion/roughing’ of Pt electrode

submitted to ultrasonification [28], or a progressive ablation

associated with cracks formation in glass surrounding a disk

electrodesubmittedtoultrasound[32].

Inbothmedia,resultsindicatethattheirondepositwasmore

easilyremovedfromtheVC2thanfromtheVC1electrode,after

15s of ultrasonification,100%of theiron deposit is dispersed,

confirmingwhathasbeenobservedpreviouslyonthebasisofSEM

imagesininsetsofFig.8.Incomparison,onVC1,!60%(inFeCl2)

and!90%(inMohr’ssalt)isremovedafter15sofultrasonification

(Fig.8).

TheaboveobserveddifferencesinironadhesiononVC1and

VC2cannotbeexplainedbydifferencesintheworkofadhesion

betweenironandVCsubstrates(WFe/VCinJ/m2).Asalsoindicated

above,anotherdifferencebetweenVC1andVC2electrodesistheir

surfaceroughness’.MicrometricholesonVC1,thatinduceahigher

roughness(Fig.7),actas‘anchors’fortheelectrodepositediron.

Roughness induces higher iron/VC interfacesurface areaSFe=VC

leading to higher adhesion energy, E_adh¼WFe=VCSFe=VC, as said

before. ‘More cavitation bubbles’ (which provide mechanical

energy)arethen requiredtodetach theironparticles fromthe

electrode, which takes more time (using always the same

ultrasound power) and leads to a slower iron removal as

experimentallyobserved.

3.3.2.2.Effectofdifferentpolishinglevelondepositdispersion.

Inordertoconfirmtheeffectoftheelectroderoughnessoniron

depositadhesion,threedifferentpolishingswereappliedtothe

VC2electrodesurface:thesubstratewaspolishedwithpapergrid

P800(electrodeVC2P800),P1200(electrodeVC2P1200)andwith

0.3

m

maluminasuspension(electrodeVC20.3

m

m).Aspreviously

stated,theelectrodessurfaceswerecharacterizedusinganoptical

surfaceprofiler(Zygo3D)andresultsareshowninFig.10.Dueto

mechanical polishing stripesare clearly visible forall levels of

polishing.Theonedimensionalprofiles,presentedinFig.10,show

a peak-to-peakamplitudeof approximately2

m

mand 1

m

mfor

VC2P800andVC2P1200respectively.OnVC2,polishedwiththe

0.3

m

maluminasuspension,thesurfaceappearstobesmoother,

presenting apeak-to-peakamplitudeofapproximately0.03

m

m.

The values of the Wenzel's roughness rw and the arithmetic

roughnessraaregiveninTable2.

Ironelectrodepositsobtainedaftergalvanostaticelectrolysison

VC2,polishedwithP800andP1200paper,arepresentedinFig.11a

(insets). The effect of the ultrasonification duration on iron

dispersionwasexaminedbyplottingoftheratio(Qa/Qaref)asa

functionoftUS(Fig.11a),forthethreedifferentpolishinglevels.As

forVC1(Fig.8aandc,andFig.9),apracticallyexponentialdecrease

was observed, that validates the fact that iron electrodeposit

dispersion by ultrasonification follows a first order kinetics

(Equation 5). The lowest dispersion efficiencies are indeed

obtainedfor electrodesthat present thehighestroughness. For

example, after 5 s of ultrasonic duration, the quantity of iron

depositremainingonthesurfaceforVC2,decreasesfrom!45%to

<5%when polishingwas achievedrespectivelywithpapergrid

P800,P1200and0.3

m

maluminasuspension.Thisconfirmsthe

effectoftheroughnessonthedispersionkinetics.

For VC2P800and VC2P1200, thesame quantity of ironwas

depositedduringtheelectrolyses,butasrw(ratiobetweenactual

surfaceand geometricalsurface)ishigherforVC2P800thanfor

VC2P1200 (Table 2), SFe=VC is larger for VC2P800 than for

VC2P1200. This induces a larger Eadh¼WFe=VCSFe=VC for

VC2P800and then a slower dispersionkinetics, asobserved in

Fig.11a.

Depositsobtainedontheseroughsurfacescanbecompared

withdepositsobtainedonasmoothVC2electrode(polishedwith

0.3

m

m alumina suspension, inset of Fig. 8b). On the smooth

surface,ironisdepositedrandomly,formingroundedparticlesand

micrometric plates/flakes (Fig. 8). On the rough surfaces

(VC2P800andVC2P1200),micrometricplatesarealsoobserved,

buttheygrowalongthestripes,followingthetopographyofthe

substrates.Furthermore,SEMpictures,takenwith40$ _tilt,show

thatironpreferentiallygrowsatthebottomofthestripes(Fig. 11b).

Thisbehaviourdoesnotcorrelatewiththeclassical

phenome-nology of galvanostatic electrodeposition on rough substrates.

Whenthediffusionlayerthickness(!14

m

mhere)isgreaterthan

theheightofthestripes(2

m

mhere,macroprofile),due tolocal

variations of the diffusion layer thickness, metal should be

preferentiallydepositedatthetopofthestripes[41].Wecannot

explainthisobservationatthistime.However,linearvoltammetry

conducted on these three surfaces reveals that the greater

theroughness,thelessertheoverpotentialofinitialFe(II)reduction

(not shown). By analogy with the analysis undertaken in

section3.1.2,thisfactsuggeststhatpolishingstripesofferanarea

Fig.10.SurfaceprofilesofVC2electrodesobtainedbyanopticalsurfaceprofiler(Zygo3D)andtheironedimensionalprofiles(perpendiculartothestripes),a)VC2P800,b) VC2P1200,c)VC20.3mm.

Table2

VC2surfaceroughness’fordifferentlevelsofpolishing:with0.3mmalumina suspension,withpapergridP800andP1200.

Amplitudemax(mm) rw ra(nm)

VC20.3mm 0.03 1.0000 8

VC2P1200 1 1.0156 162

where

g

Fe=VCislowered,facilitatingirondeposition.

g

Fe=VCshould

thereforebeloweratthebottomofthestripesthanatthetop,

leading to an improved adhesion of these ‘anchored’ deposits

(WFe/VC=

g

Fe/L+

g

VC/L#

g

Fe/VC,seetheAppendix).

Anotherpoint is thatiron depositedontothestripescanbe

protectedfromthefluidstreamandultrasonicperturbationthat

cancausethedetachmentofthedeposit.Indeed,thesamequantity

of iron is deposited on the rough substrates (VC2P800 and

VC2P1200)asthatonthesmoothone.Butirondepositedinsidethe

stripesof theroughsubstrates hasan iron/liquidinterface less

exposedtocavitationbubbles.

These results clearly confirm the major importance of VC

electrodesurfaceroughnessontheadhesionofirondepositand

demonstratethat dispersion kineticseffectively slows down as

electrodesurfaceroughnessincreases.

4. Conclusion

Theobjectiveofthisworkwastostudyphenomenainvolvedin

thesynthesis of zero-valent iron nanoparticles by

sonoelectro-chemistry.Theroleoftheelectrodesubstrateonthemorphologyof

theelectrodepositediron and its dispersionby ultrasound was

moreparticularlyinvestigated,consideringthefollowingthreekey

parameters: 1) the interfacial energy of the iron/electrode

interface

g

Fe=elect,2)theworkofadhesionoftheelectrodeposited

ironontheelectrodeand3)theelectrodesurfaceroughness.

ChoosinggoldorVCaselectrodematerials,exhibitingdifferent

interfacialenergies

g

Fe=VC>

g

Fe=Au, allowstoobservesignificant

differencesin both themorphologyof iron electrodeposits and

theirdispersionbyultrasound.Nodispersionwasobservedusing

ultrasound when iron was reduced on the gold; the

electro-deposited iron adheres strongly and it spreads easily along

electrodesurface(2Dgrowthmode)beforetheeventualstarting

ofadendriticgrowth(dependingontheprecursorsaltused).On

VCelectrodes,electrodepositedironcoverspartiallytheelectrode

surface (3D growth mode); its weak adhesion facilitates its

dispersionbyultrasound.

A particular regime was observed, using the smoother VC

electrodeduringcyclicvoltammetryscanunderultrasonification:

electrodepositdispersioniscomplete,allowingcontinuous

regen-eration of the bare substrate; fine iron particles should be

synthetizedduringthisprocess.Theseresultsallowtohighlight

theimportanceoftheelectrodematerialchoicefortheelaboration

of sonoelectrochemical devices dedicated to the synthesis of

metallicnanoparticles.Thelowsurfaceenergyandthelowaffinity

of VC substrates against the deposited metal offer the double

benefit:

-tofavorthelowspreadingofthemetalontheelectrode (3D

growth),promotingtheformationofisolatedfineironparticles

insteadofthinfilms

-tofacilitatethe‘ultrasound-assisted’dispersionofthe electro-deposit.

Theironelectrodepositsdispersionbyultrasoundappearstobe

describedbyafirstorderkineticsbecauseoftheconstantremoval

action of the cavitation bubbles. Moreover, dispersion kinetics

slowdownwhentheelectrodesurfaceroughnessincreases,the

latterinducingenhancedadhesionoftheironelectrodeposits.

Evenifthesynthetizedquantityofdispersedironparticlesis

lowwiththedeviceused(micro-electrolyses),somepreliminary

electrolyses were carried out (without surfactant) by applying

ultrasound(VC2polishedat0.3

m

m,Mohr'ssalt).The

characteri-zation by dynamic light scattering revealed the presence of

!200nmsizedparticles.

Acknowledgments

This study was supported by the MSR Graduate Research

FellowshipandtheauthorswouldliketothankthePaulSabatier

Universityforfundingtheresearch.ThanksarealsoduetoSophie

Chambersforcheckingthemanuscript.

Appendix.Estimationofworksofadhesionandinterfacial

tensions

The objective of this Appendix is toestimate the works of

adhesionofiron(Fe)ontheelectrodematerialsusedinthisstudy,

WFe=elect, withelect correspondingeither to Au (gold) or toVC

electrodes.Furthermore,theinterfacialtensionsoftheinterface

betweenironandelectrodematerials,

g

Fe=elect,arealsoestimated. WFe=electand

g

Fe=electareestimatedusingthefollowingtheoretical

development and the values of electrode surface energies

measuredbythecontactanglemeasurementmethod.

Theworkofadhesionofironontheelectrodecanbeestimated

consideringonlythevanderWaalsinteractionsandfollowingthe

theory developed in [43]. Considering the van der Waals

interaction potential profile for iron interacting with electrode

across the liquid phase, work of adhesion corresponds to the

extremumvalue of this profile. It hasbeen shown, for several

Fig.11. Dimensionlessironquantity(Qa=Qaref)remainingontheelectrodeasafunctionoftheultrasonificationduration(tUS)forVC2inFeCl2solutionforvarious polishinglevels.Points correspondtotheaveragevaluesQa=Qaref measuredforatleastthreeexperiments;errorbarscorrespondtostandarddeviations;insets showSEMimages oftheelectrodepositsjustafter theelectrolysis(notexposedtoultrasonification)onVC2 polishedwithP800andP1200 papergrids.b)SEM imagewith 40$ tilt, of the iron electrodeposit onVC2P800; iron deposits located at the stripes bottom are visible inthe red ellipse.