Copper Electrodeposition on Alkanethiolate Covered Gold Electrodes*

Zeitschrift fürPhysikalischeChemie,Bd. 208,S. 107-136(1999)

© _byR._{Oldenbourg Verlag,} München 1999

Copper Electrodeposition

on

Alkanethiolate Covered

Gold Electrodes

By

O.

_Cavalieri,

A. M.

_Bittner**,

H.

_Kind,

. Kern

InstitutdePhysique Expérimentale, Ecole

_{Polytechnique}

Fédérale de Lausanne, CH-1015Lausanne, Switzerland

and T. Greber

Physik-Institut,UniversitätZürich, CH-8057 Zürich, Switzerland

(ReceivedDecember 1, 1997;accepted December23, 1997)

Cyclic voltammetry

I Film

_growth

I

Scanning tunnelling microscopy

(STM)

I

_Single

_crystal

electrode I

Surface

structureI Thiols I

_{X-ray photoelectron}

_spectroscopy

_(XPS)

We have

_investigated

the structure and thermal dynamics of alkanethiolate

_layers

on Au(lll)withvariable temperaturescanning tunnelling

microscopy

(STM),X-ray

photo-electron spectroscopy(XPS) and voltammetry.The results buildthe basis fora study of

electrodeposition

of copper on alkanethiolate-covered _Au(lll). Electrodeposition has been studied as afunction ofthe thiolate chain

_length,

the _deposition

_potential

and the

temperature. Time-resolved in situ STM and

_voltammetry,

both at _{temperatures up} to

345K,and XPS ofemersed_{samples showed that copper}can

—

dependingonthe

deposi-tion _potential

—

either form nanometer-sized islandsor layers, both without destruction

of the thiolate. _{We propose}amechanism_{where copperpenetrates}the thiolatelayer. The slow _deposition rate isdeterminedonly by kinetic factorssinceCu/Cu2+ exchange

pro-cessescannotoperate. Finallywe discuss therole of thiolatesaspreadsorbedsurfactants.

1.

Thiolate

_layers

and the

_{electrodeposition}

of

metals

Certain chemical elements or

molecules,

especially

soft bases such as

iodide,

bromide,

sulphide,

carbon monoxide and

_cyanide,

areknowntobind with

_{exceptional strength}

to_{noble metal surfaces. In electrochemical terms,}

the substances adsorb

_{specifically.}

Such

_adlayers

are often

densely packed,

hence the atoms or molecules

prefer (quasi)-hexagonal

_arrangements

[1-7].

* _Presented

at the 5. Himer Elektrochemische Tage on "Fundamental Aspects of

Electrolytic

Metal

_Deposition",

June23-24, 1997.

**

The

_adlayer

_geometry

is determined

_by

the

_interplay

of

adsorbate-sub-strate forces that tend to direct the

_adsorbing

atoms to

_places

that offer a

strong

bond

(often

hollow

_sites)

and interadsorbate forces thatcanbe

attrac-tive or

repulsive.

More

complex,

but also much more

versatile,

are thiolate adsorbates. Alkanethiols

CH,(CH2)„_ ,SH ('C„

thiols')

form

_strongly

adsorb-ed,

ordered thiolate

_layers

CH,(CH2)„_,S/substrate [8-10].

Although

the

adsorption

process does not differ

_principally

from the above-mentioned

cases,theterm

'self-assembly'

is

employed

to

point

outthatthiolate

interad-sorbate forces are _very

_strong

—

long

adsorption

times are _necessary to

assemble an ordered structure where the

alkyl

chains

align parallel. Up

to now,

mainly

the

Au(lll)

substrate was

used;

its reconstruction is

readily

lifted

_during

the

_adsorption.

Several studies of vacuum metal

deposition

on _{thiolate-covered metal} substrates were

conducted;

Jung

and Czanderna

compiled

avaluable review

[11].

The

_{electrodeposition}

of

metals,

e.g. copper,onthiolate

layers

is

tech-nically

much

_simpler,

but moredifficultto

interpret:

onehas toconsider the electrode

_{potential. Especially}

the Cu2+/Cu

_equilibrium

(Nernst)

potential

is

important.

Positive of that

_point,

_copper can

only deposit

in a

single layer

and

_only

on some

special

substrates

(underpotential

deposition,

UPD) [12

—

17].

Negative

of

_it,

normal copper bulk

_deposition

takes

_place

on _any

sub-strate

(overpotential

deposition,

OPD).

Electrodeposition

of Cu2+ in the

OPD range often results in 3D

growth

ofcopperon _{metal substrates} as_e.g. found with in situ STM

[12a, 18-23].

A similar situation shows up when copper is

_deposited

on

alkanethiol-ate-covered

Au(lll)

deep

in the OPD range:

large

copper nodules with diameters in the 100nm range

_develop

[24, 25].

In contrast, we worked _at moderate

_{overpotential}

which results in a

complex

scenario that can be

investigated

with in situ STM

[26—29].

We found a

pseudo-layer-by-layer

growth

[27],

i.e. one

layer

is almost finished before the next one starts to

grow. Such abehaviour is _very rare on _{bare metal electrodes}

_(some

excep-tions of this rule were found with in situ STM

[12a,

19,

30—32]).

On the other

hand,

it is well known that a _{metal surface covered}

by strongly

ad-sorbing

substancescanindeed incite sucha 2D

growth.

Such substancesare called 'surfactants'; in the

simplest

case, one of the above-mentioned

strongly adsorbing

atomic ions is a

surfactant,

as known for the

elec-trodeposition

of silver on iodide-covered

Pt(lll) [33].

Usually,

the metal

penetrates

the

_halogenide

_layer;

the

_topmost

structure can be addressed as a

metal-halogenide

coadsorbate

[17, 33—36].

In the presence of a

large

surfactant

_molecule,

_crystal

violet,

in the

_electrolyte,

an STM

study

indeed

proved

the 2D

_growth

scenario

_{[20, 22].}

This situation is

_already

close to

the technical

_application

where

_{'brighteners'}

are added to metal

_plating

baths to

_yield

smooth

_deposits.

In the UPD range where island formation on bare metal substrates is a rare

phenomenon

(but

see

[37]

and

[38]),

we were able to

deposit

_copper

Copper ElectrodepositiononAlkanethiolate Covered Gold Electrodes 109

islands with diameters in thenm _range on alkanethiolate-covered

Au(lll).

This process is a

relatively simple

method for

building

nanometer-sized

metal structures

_[29].

An often

_neglected

_parameter in electrochemical STM studies is the

temperature on which we shall concentrate _{in this paper. As} we _will show in section

3,

heating

induces several

stages

of disorder in

thiolates,

pro-gressing

from the

_(methyl)

_{end group towards the substrate.}

First,

the

methyl

rotation is activated at > ₁₀₀

asrevealed

by

helium diffraction studies

_[10].

Above ₂₀₀ the infrared

_probing

of the

_{méthylènes'}

scissors vibration

_suggests

_{progressive unlocking}

of the

_alkyl

twist

(a

rotation around either anS—CoraC-C

_bond)

and

_development

of

_gauche

defects in the otherwise all-trans

_configured

chains

_[39].

_{This behaviour compares} well with that of self-assembled thiolates on

gold nanoparticles

[40]

and that of

_melting

bulk alkanes

[41].

Heating

above 300 as weused for this

study

can even affect the Au—S bond: boundariesbetween different ordered thiolate domainsmove or

vanish,

mass

_transport

of the substrate occursand

finally

leadstothe

_healing

of the

_typical

substrate vacancies. These vacan-cies are defects

_{appearing during}

_the

self-assembly

[42—45].

A discussion of the

_{phenomena occurring}

upon

_heating

thiolate

_layers

(see

section

3.2)

will

_provide

a _{valuable basis for}

understanding

our _most

recent

_experiments

where we followed the

electrodeposition

_{of copper at} elevated

_temperatures

with in situ STM

_(see

section

_4).

Electrochemical STMswere _uptonow never

operated

above 300 albeit

deposition

studies should be of

_greatest

_importance

for the

_{understanding}

of technical metal

plating

which is

_mostly

carried out well above 300 K. We will also show that the

_temperature

canbe usedas anextra

_parameter

for

_tuning

_{the copper} as well as the substrate structure

(see

section

4.3).

_Encompassing

our

ex-periments

at 300

_[26,

27,

29]

and the recent

results,

we will focus onthe structure of the

_deposited

_{copper in section 4.2 and}

_{additionally analyse}

the

extremely

slow

_deposition

kinetics in section 4.4.

Our

_experimental

methods are in situ electrochemical STM at 300 up to 345

K,

STM in

_nitrogen

at _{300 up} to 370

K,

cyclic voltammetry

and

X-ray photoelectron

spectroscopy

(XPS)

of emersed

_{samples. They}

will be

explained

in the

_following

section.

2.

_{Experimental techniques}

The

_{gold samples}

were

prepared by evaporation

of 120 to 150nm

gold

on 570 hot cleaved mica in a 10~5 mbar vacuum and then annealed for several hours at —600

₍₂

· 10"6

mbar).

Directly

before

starting

the

self-assembly,

the

_gold

films were flame annealed at _{very dark red}

glow

and cooled in ethanol

_{(p.a., Fluka).}

For the

_{self-assembly,}

the

_samples

were transferred into 10

_pM

or 1 mMethanolic solutions of alkanethiols

(Fluka)

"beetle" STM

Fig. 1. Schematicdrawingoftheelectrochemical temperature-variableSTM setup.

CH3(CH2)„_,SH

with =

6, 8, 10, 12, 14,

16 and 18. Forthe

experiments

carried out under

_nitrogen,

the

_samples

were

kept

at 300 for at least

one

day.

For the electrochemical

experiments,

the

following self-assembly

programwas used: _{twohours 300}

K,

_two

days

ca. 325

_K,

thereafter several more

days

(or

inrare cases

weeks)

300 K. All

samples

were

emersed,

rinsed with ethanol and

_quickly

used. STM measurements were

performed

as de-scribed in

[29].

As shown in

_Fig.

_1,

a new _{feature is} a Peltier element

(operated

in the

_heating

_{configuration)}

that we

coupled

with a thin copper

spacer to the bare side of the mica

_sample,

thus

_attaining

a stable

sample

and

_electrolyte

_temperature

_(less

than 1

_change

_per

_hour).

The

gold-covered side was incontactwith the

_electrolyte,

50 mM

H2S04

_(p.a.,

Fluka)

+ 0 or 1 mM

CuS04

_(p.a.,

Fluka),

either in water

_{(Millipore®)}

or

-for the

_{high-temperature experiments}

—

in 2:1

_HOCH2CH2OH

_{(1,2-ethandiol,}

ethylene glycol,

p.a.,

Fluka): H20.

_{Ethylene glycol}

very

effectively

stops

the otherwise

_{quick evaporation.}

The reference was a _copper

wire,

etched

in

_HNO,

and rinsed with water

_shortly

beforeuse. The

sample

was mounted in a PCTFE STM cell

equipped

with a Pt counter electrode. For the XPS

sample

preparation

as well as for the redox

pair

measurements,

_exactly

the same

_set-up

was used. The redox

electrolytes

contained

Ru(NH,)6Cl3

(Alfa)

or

K3Fe(CN)6 (MicroSelect, Fluka)

with KCl

(p.a., Fluka)

and HCl

(p.a.,

Fluka)

in

_{ethylene glycol/water,}

and an

Ag/AgCl

reference was

dipped

into the STM cell. All

_potentials

are

quoted

with

_respect

tothe Cu/Cu2+

pair

in

1 mM Cu2+

(i.e.

0 mV read +250mV on the standard

hydrogen

scale and +50 mV on the

Ag/AgCl

(KCl sat.) scale).

We

always

employed

an EG&G

PAR 400

_{potentiostat.}

The

_potential

_{range above 500}mV and below —200mV was

usually

not accessed in order to avoid oxidation and re-ductive destruction of the thiolate

_{layer, respectively}

_{(see 4.3.2).}

STM

im-Copper ElectrodepositiononAlkanethiolate Covered Gold Electrodes 111

ages are shown derivatised: _steps

_descending

_{from left} to

_right

_{show up} as black

lines,

ascending

ones as white lines. The

X-ray

photoelectron

_spectra

have been

_performed

in a VG ESCALAB 220. After removal from the

liquid

we mounted the

samples

with electrical _contact and _{introduced them}

via a fast _entry lock into the vacuum _system. In order to minimize

_X-ray

tube induced

_damage

the

_spectrometer

transmission was maximized with lowest

_angular

resolution andanoverall energy resolution of 1.4 eV FWHM on the Au

4f7/2

peak

at 84.0 eV

_binding

_{energy. The}

_X-ray

twin anode that

provides

non-monochromatized

_Mg

_{(1253.6 eV)}

radiationwas retracted and run _{with 140}W

_input

_{power. With}

_respect

_{to the maximum flux these}

settings

lower the

_resulting

electron flux from an aluminum

sample by

a factor of 0.3 to 15 nA/cm2. The emission

_angle

was 0°

(normal emission)

and 80°

_(grazing

_emission).

Fromthe

_comparison

ofthetwoemission direc-tions the vertical structure of the chemical

_composition

of the

_(Cu)/

CH3(CH2)„-iS/Au(lll)

system

canbe inferred.

3.

Alkanethiolate

_layers

on

Au(lll)

Inorder to

_develop

abasis for

understanding

the

complex

_copper

deposition

process we shall here

_present

studies of alkanethiolate

_layers

_on

_Au(lll)

undervacuum, in

nitrogen,

air and immersed in

electrolyte,

all in absence of copper.

3.1 Structure and

_morphology

Letus first focus on the chemical

composition

_of alkanethiolate

_layers

_on

Au(lll)

which canbe detected with XPS. In

Fig.

2 theoverview

_spectrum

of

_{CH3(CH2)17S/Au(lll) (sample}

_A)

is shown for normal emission. The

spectrum

is dominated

_by

the

_strong

emission from the

_gold

substrate and the C Is

_peak.

It can be seen that the

samples

remain almost oxygen free

although they

have been

_exposed

to airfor more than 100s. In all

details,

such a

_spectrum

is

typical

of the

_{thiolate/gold}

_system

[46, 47].

In

_Fig.

3 the Au

4f,

C

Is,

O Is

_(and

for

_comparison

Cu

_2p3/2)

XP

_spectra

areshown for

sample

A. The Au

4f7/2

peak

is set to 84.0eV

_binding

_energy. The C Is

_peak

is foundat285.0eV and

_corresponds

tothat of

_polyethylene

chains

[48].

We show as well the

_spectra

from

grazing

(80°)

emission

(data

for carbon and oxygen are not shown since the

_sample

holder contributes

to _{the oxygen and} some of the carbon

emission).

As the

intensity

from

sulphur

(not shown),

the

_intensity

of the

_gold

_signal

is

_{strongly suppressed}

;

this is observed in presence

(as

in

_Fig.

3)

and absence

_{(not shown)}

of copper. The attenuation

by

more than one order of

magnitude

can

only

be

explained by

a

layer

model in which the thiolatecovers the

gold

and binds with the

_sulphur

end tothe substrate.

SampleA Au 41 Sample Cu2 1000 800

ElectronBinding Energy (eV)

Fig.

2. Normal emission XPS ofaC,„thiolatelayeron Au(lll); X-ray source Mg .

Sample Acorresponds toan untreated sample, whileon sample _copper was

electro-chemically

deposited.

In

_Fig.

4 the

_grazing

emission

_intensity

of the C

_Is,

the Au 4f and the Cu

_2pv2

levelare shownas afunction of exposure timeto the

X-rays.

While the carbon emission remains constant, the

gold

and the copper emission increasewith the exposuretime. For

_samples

withoutcopper

deposit

we do observe thesame

aging

kinetics

_(not

shown).

This variation

clearly

indicates a

change

of the

sample.

The

overlayer

appearstobecome more

_transparent

to the emission from the substrate. The

gold

andcopper intensities

perfectly

fit a _{first order kinetics} ansatz I «

(1

—

ße"r)

where

_ß

> 1 is a

sensitivity

factor and where is the time constant. Within the

_uncertainty

of the

_data,

the

_sensitivity

_ß

and the time constant are

independent

whether copper has been

_deposited

on the

sample

or not. The

_{sensitivity ß}

is 0.45 ±0.07 for copper and

gold,

is

_{inversely proportional}

to the

_secondary

electron flux from the

_sample.

From the value of of 2 · 104 s and the

secondary

electron emission flux we find that the

secondary

electron current

density

from an aluminum

sample

times the time constants is in the order of

2· 1015 e~/cm2. This indiates that the thiolate films are very

susceptible

to

ionizing

radiation as also

reported

by Jäger

etal.

[49] (other

groups found much

_larger

time constants

_[50]).

Turning

the attention towards the interfacial structure

probed by

STM,

let us first note that the

_tunnelling

currents

employed

(some

tenths of

nA)

are

essentially

sensitive tothe

sulphur

atom

-the

_tip

_easily

_penetrates

the thiolate chains which cannot offer local electronic states in the _proper

CopperEleclrodepositiononAlkanethiolatcCoveredGold Electrodes 113

(Fig.

5),

can thus be ascribed to the

_sulphur

atoms,

_yielding

a _{coverage of} 0.33. Note that thesamestructurewas_{found fora}

sulphide

_adlayer

_adsorbed on an

Au(lll)

electrode and for

S/Au(lll)

in vacuum

[7]. However,

we

-1-1-1-r

C 1s

-Au 4f

X-ray

exposure-time(103

s)

Fig. 4. _Grazing

_{photoelectron}

emission intensities of C Is, Au 4f and Cu 2p,/2 (C,8 thiolate)as afunction of_exposuretotheMgKaX-raysource.The solidlines in the data

of_{copper and}goldarefits tofirst order kinetics.

a)

b)

Fig.

5. Molecular resolution STM

_images

takenin

_{nitrogen showing}

thec(4X₂₎

_pinwheel

(a)and thec(4X₂₎

_zig-zag

_{(b) superstructures}ona_C,„thiolatesample.

Tunneling

param-eters: (a) I = 1.0nA, V = 1000mV; (b) I = 0.6nA, V = 1000mV. Image size:

3.5nmX3.5nm.

bution

_(Fig.

5)

of the

₍₇₅

X

_75)R30°

elements

(in

proper notation

(3X2

75))

[10, 52—55].

At lower coverage

(attainable

_by

vapour

phase

Copper Electrodepositionon Alkanethiolate Covered Gold Electrodes

1

115

Fig.6. STM

_image

ofaC,„thiolate-coveredAu(lll) surface in air. The darklines are

thiolate domain boundaries which are

_{pinned by}

the substrate vacancies (dark spots). Tunneling parameters: I =_0.3 nA,_V = 900_mV. Image_size: 320nmX320nm.

chain thiolates we _detected

striped

_structures

[58] (see

_also

_[56,

_57,

_59,

60]).

Allthese structurescanexist indifferent domains

_{separated by}

_domain walls. These walls show up as dark grooves of

_roughly

half a

monolayer

depth

and 1 to3nm width. In

Fig.

6 the walls often extend between

neigh-bouring

substrate vacancies

_(large

dark

_spots),

but

_they

can also _terminate

at other walls or at

_steps.

_{During heating}

in

_nitrogen,

the number of grooves in an STM

image

diminuishes,

i.e. the average domain size in-creases

[61].

Further

_healing

_{processes that} can be followed involve Ostwald

ripen-ing,

coalescence and annihilation

_(at

_steps)

of the vacancies

[43,

45].

The latter are ascribed _to

_etching

_processes

[62, 63]

_{and the}

_lifting

_{of the}

Au(lll)

reconstruction and concomitant surface diffusion

_during

the

self-assembly

process

[64, 65];

their surface is covered

by

thiolate as are all otherterraces

_[58,

66,

_67].

Our latest

_experiments

show that the

₍₇₃

X

73)R30°

_lattice,

domain walls and vacancies arefound also when the

samples

are immersed in

elec-trolyte

and

_kept

under

_potential

control

_(in

situ

_STM).

When we use the

modified

_electrolyte

basedon an

ethylene

glycol/water

mixture,

we can now

additionally

scanthe surface at elevated

_temperature

_[28].

As visualized in the

_image

series of

_Fig.

7,

_depicting

a _{vacancy that is}

rapidly

filled with

material from a

_nearby

_step,

we found

highly

mobile surfaces

already

at 335 K. The surface

_appeared

to be more mobile than in

_{nitrogen, probably}

a result of the electrochemical

polarization.

Some further

_properties,

_especially

the

_ability

to transfer electrons

through

athiolate

layer,

can be

probed

with

cyclic voltammetry.

As

electro-lytes

we

routinely

used

sulphuric

acid either in water orin

ethylene

glycol/

water. In

_passing

we note that we

proved

_that

ethylene

glycol

does neither affect the

_{electrochemistry}

of

thiolate/Au(lll)

nor that of the copper re-duction at

thiolate/Au(l 11)

and bare

Au(lll) [28].

The

voltammograms

are

typical

for nonreactive surfaces since the double

_layer

current is smaller thanforabare metal surface. In

general,

thecurrentfor short chain thiolates is

_larger

than for

_long

chain thiolates. At

_higher

_temperatures

the current

rises

_{considerably.}

Thus electronscan be transferredacrossthe thiolate eas-ier when the chain is short and the

_temperature

_high.

However,

one should

keep

in mind thatthe double

_layer

could have a structure and thus

_capacity

differing

from thaton bare

gold.

Hence we also

probed

the surfaces with a redox

pair,

Fe(CN)ft'4".

The

voltammogram

in

_Fig.

8 reveals the

_typical

current waves that result from kinetic and diffusional

limitations,

but with less

_peak

current than on bare

Au(lll).

Wefound thesame behaviourasfor the double

layer

current men-tioned above: the redox waves' current rose at

high

_temperatures

and for

short chain

_lengths.

Note that the

_temperature

behaviour canbe more

com-plex

[68].

Experiments

with

Ru(NH3)6/2+

normally

showed a much

higher

current;

however,

the same

dependence

onthe thiolate chain

length

is found

(Fig.

9).

To assess the

findings

we note that the diffusional limitation must be the same _at bare and thiolated

Au(lll).

The rate of the slowest

(rate

determining)

process

responsible

for the thiolates'

blocking

behaviour can

be the

_charge

transfer across the thiolate or

possibly

a

penetration

of the

complexes

into the thiolate.

In the

_following,

we shall

interpret

the structure and

_morphology

of alkanethiolate-covered

_Au(lll)

at 300 K. We _{will progress from} a

single

molecule

(sub-nm)

to the

_mesoscopic

scale

_(pm).

First,

our XPS data

con-firm the

_accepted

structure model of a

_gold

surface covered

by

sulphur-bound thiolates. The small oxygen contamination is

likely

due toair

expo-sure. It may consistofadsorbed oxygenatoms

_(e.g.

at thiolate-free

defects)

and/orwater

(difficult

toremove

completely

invacuum

chambers).

However,

oxygen isaminor

species

on athiolate

surface,

evenafter exposuretoair.

Fig.

7.In situelectrochemical STMimagesequence(time

separation

ca. 1 min)

showing

the _temperature induced

_mobility

ofaC(,thiolate/Au(lll) interface at335 K. Electrode

potential:

300mV vs. Cu/Cu2+, i.e.

_positive

from the UPD _{range, thus the surface is}

copper-free. Electrolyte:

50 mM H2S04 + 1 mM CuS04 with the solvent

_being

a 2:1

mixtureofHOCH2CH2OH: , .

Tunneling

parameters: I =0.5 nA,_V= _{120 mV.}Image

Copper ElectrodepositiononAlkanethiolate Covered Gold Electrodes 117

0 200 400 E(mVvs.Cu/Cu2+)

Fig.8.Cyclic voltammogramsforan Au(lll)electrodeat300 _(a),foraC,6 thiolate-coveredAu(lll)electrodeat300 (b,dashedline)andat335 (b, solidline).

Electro-lyte: 1000 mM KCl+ 1 raMK,Fe(CN)„with the solvent_beinga2:1 mixture ofHOCH,

CH,OH:H20. Scan rate: 50 mV/s. Potentialswere _measured vs. theAg/AgClelectrode

and are reported vs. a

hypothetical

Cu/Cu2+ (1 mM) reference electrode for an easier

comparison

with the previouslyshown voltammograms.

Extending

the view from a

single

molecule to

_neighbouring

_molecules,

wenote that the coverageof

_sulphur

atoms

_(and

thus of the

alkanethiolates)

is in most cases

1/3;

_the

sulphur

atoms _occopy a

(73

X

73)R30°

lattice

(see above).

Exceptions

are encountered for the short-chain thiolates

(e.g.

= 6 where this lattice and _a

striped

_structure

coexist)

_or for the low

coverages attainable either for

samples

annealed above 370 or

by

vacuum

deposition

of thiols. As we

already pointed

out in section

1,

the

(73

X

_73)R30°

_geometry

is also found for the

S/Au(lll)

system.

This and other

_{(quasi)hexagonal}

structures

_(e.g.

found when soft bases are adsorbed

at noble metal surfaces; even when the substrate structure isnot

hexagonal

[1, 2])

are

typical

fora balance between substrate and adsorbate-adsorbate interactions.

Clearly,

this balance has to be

_additionally

affected when

_alkyl

chains interact with each other. It is well known that this results in the all-trans

CopperElectrodepositionon_{Alkanethiolate Covered Gold Electrodes} 119 12(1 -180F

I_ _ _ _

I

-300 -200 -100 0 E(mVvs.Cu/Cu2+)

Fig.

9. Cyclic

voltammograms

fora C10thiolate (solid line) and a C14thiolate-covered (dashed line) Au(lll) electrode at 300 K.

_Electrolyte:

50 mM _H2S04 + ImM

Ru(NH,)6Cl3. Scan rate: 50mV/s. Potentials were measured vs. the Ag/AgCl electrode

and are reported vs. a

_hypothetical

Cu/Cu2+ _(1mM) reference electrode for an easier

comparison

with the

_previously

shown

_{voltammograms.}

S Au _OS · C

Fig. 10. Models of the CH3(CH2),S/Au(l11) interface: (a)

alkyl

chains in the all-trans

conformation,(b) alkyl chainspartiallyin_gauche conformation.

configured alkyl

chains

_being

_{tilted away from the surface normal}

_by

about

30°

(see

_Fig.

_10(a)).

Theazimuthal tilt direction liesbetween the directions of the next and nearest next

sulphur neighbours.

The above-mentioned

c(4

X

₂₎

_{lattice appears}tobe basedon an ordereddistribution of

alkyl

chain twist

_{angles (presumably}

a twist around the S—C

axes);

structures based on ordereddistributions of

alkyl

twist

angles

exist also in bulk alkanes

[4].

We can conclude that the thiolate structure is based on the _strong Au—S

bond;

the maximum coverage is

_{governed by}

S —S and interchain interac-tions

—

each one exhibits an _energy

minimum,

and their

interdependence

yields

an overall minimum with the

optimum

_geometry

described above. The

_{densely spaced alkyl}

chains form an effective spacer

layer

that blocks electron transferas evidenced

by

the voltammetric data.

On the

_length

scale of 10nm, we find the grooves that form the bound-aries between

_differently

oriented molecular domains.

_{Unfortunately,}

it is

impossible

todeduce a structural model for them:

firstly,

the mechanism of

tunnelling

on thiolates is still under

debate,

hence the

depth

is difficult to

interpret; secondly,

molecular resolution of the grooves is very difficult. The

_simplest

idea is an

antiphase boundary

where the S —S interdomain distance is

_larger

than the intradomain distance of

J3

atomic

_gold

distances,

but also

_missing

rows of thiolate molecules have been

suggested

[55].

Finally

let us

shortly

focus on the

mesoscopic

scale

(pm):

this is the

typical length

of

_gold

substrate

_steps.

_{Very long}

and

_straight

_steps

are

pre-sumably

due to the

healing

of volume defects

during

the

gold

flame

an-nealing (gliding along

internal

₍₁₁₁₎

_planes);

hence

_they

can cross each

other. Shorter and more round

_steps

should

develop

via

_transport

_processes

during gold

flame

_annealing

and

_quenching.

3.2 Thermal

_dynamics

Substantial

_changes

of the surface

_morphology

manifest themselves in

_step

movements whichcanbe ascribedtothe diffusion of

_gold

adatomsor

single

atom vacancies. Indeed we use the

smoothening

effect of surface diffusion

in both the vacuum

heating

and flame

annealing processing

of our

gold

samples. Naturally,

such processes are

thermally

activated. On thiolate-covered

Au(lll),

their rateis much

_higher

than on bare

Au(lll)

since the

Au —Au interaction isweakened

_by

the

_strong

Au—S bond. Substrate atom

diffusion was also recorded for

similarly strongly

bound

halogen

atoms on

Au(100)

[69a], Au(lll) [69b], Au(110] [70]

and

Pt(110) [71, 72].

High

temperature

dosing

of

_halogènes

can even

produce virtually

defect-free sur-faces

_{[2, 73]}

as can

prolonged annealing

of thiolate-covered

Au(lll) [42].

For this

_annealing

_procedure,

the vacancies and

_step

_edges

move

by

dif-fusion of

_single

substrate atom vacancies

[26, 45].

This leads to Ostwald

ripening

before the vacancies reach each other

_(vacancy

_coalescence)

and

finally

other

_steps

_(vacancy

annihilation).

We can _{group all such} events

under the

_{heading 'strongly}

activated': the

thermodynamic driving

forces have to be so

large

that

they

result in

desorption

of more

weakly

bound adsorbates. An

_example

for a

driving

force is the

temperature

that should be above ca. 800 for

gold

flame

annealing

(acting

to desorb

impurities

and

_resulting

in

_large

scale

_step

movements)

orabove ca. 340 forvacancy coalescenceon

C10

thiolate covered

Au(lll) [61].

CopperElectrodepositiononAlkanethiolate Covered Gold Electrodes 121

Another

_example

for

_'strong

activation' is the electrode

_potential;

changing

it _{to either} more

positive

or more

negative

values can result in

desorption

of adsorbed substances and surface diffusion. In most _cases, an electron is transferred fromor to the adsórbate

_during

the

_{desorption. E.g.}

only

below -HOOmV thiolates desorb from

_Au(lll)

_{[74, 75],}

demon-strating

that

_they

bind

_strongly;

below

—

500mV

_sulphide

desorbs from

Au(lll) [7];

below -300 mV iodide desorbs from

Pt(110),

causing

surface mass

_transport

of the substrate

[72];

-200mV sufficetodesorb the weaker bonded

_sulphate

ion from

_platinum

surfaces

_[76]).

In the

_{positive potential}

range, the situationcan _be much more

complex;

often anadsórbate is first oxidized and the

_product

can

desorb,

_{e.g. alkanethiolates} are oxidized to

alkanesulphonic

acids. In any cases,

changing

the

potential

will

supply

the

CH3(CH2)„_,S/Au(lll)

interface with additional electrons

(or

draw

elec-trons from

_it)

and

_thereby

weaken the Au-Au and Au-S interactions. The firstcase leadsto substrate

_mobility,

the lattertoadsórbate surface diffusion or

desorption.

However,

the

_analogy

to

halogenide

or

sulphide

adlayers

reaches its limits when we consider that in the course of surface diffusion not

only

Au-Au,

Au—S and S-S

_{interactions,}

but also interchain forces come into

play. Clearly,

for

_longer

_alkyl

chains

_they

are

larger

and

_present

an ad-ditional

_barrier;

hence the mobilities will be lowered

_(as

observed

[43]).

Whether thiolate molecules remain bound toa

moving

gold

adatomornot,

during

a diffusion event interchain van-der-Waals bonds are broken and reestablished. Thus thethiolate has toreorient which canentrain formation ordestruction of the all-trans conformation and

change

of azimuth and twist

angle.

At

_temperatures

above ca. 330

K,

a

large

_part

of the thiolate molecules

may exist in an

energetically

_(slightly)

_disfavoured state such as distorted twist

_angles

or

gauche

conformations;

the

sulphur

_atoms remain in a _fixed

position.

Let us name such a situation

'moderately

activated'.

Locally

_the effective thickness will be

lowered,

i.e.

_'spots'

_{appear in the otherwise} dense 'forest' of

_alkyl

chains

(see

Fig.

10(b)).

Such

_'spots'

are not to be confused with

_missing

molecules

_(which

are rare if the

self-assembly

was

given

sufficient

time).

Herean ion in the

electrolyte

can

approach

_the

gold

surface much more

closely.

It can

thereby

at least

_partially

_penetrate

the

thiolate,

facilitating

anelectron transfertoorfrom the

gold

surface. For this reason we detect

higher

double

layer charging

orredox currents when we

increase the

_temperature.

The currents also increase when we use shorter

alkyl

chains

—

in thiscase,the molecular spacer

layer

becomes

thinner,

and anioncan

(even

without

penetration)

approach

the

gold

surfacemore

close-ly

than in the case of

long

chains

(see

Fig.

9).

We note that

_'moderately

activated' processes cannot be

brought

about

_{by potential changes}

—

this means that we can _{vary the}

potential

in a rather wide range

_(between

a) 0.2

1

0 v.

1

-0.2 -0.4

b)

< 3. -100 100 200 E(mVvs.Cu/Cu2+) 300 400

Fig.

11.

_{Cyclic voltammograms}

forC,„thiolate-coveredAu(ll1)at300 (a)and 335

(b) showing the temperature

dependent

blocking behaviour towards copper

deposition.

Electrolyte: 50mM H,S04 + 1 raM CuSOj with the solvent

_being

a 2:1 mixture of

HOCH,CH,OH:FEO. Scanrate: 10 mV/s.

500 mV and -200

mV)

without noticeable

_changes

of surface

_properties.

In other

words,

in this range the

CH3(CH,)„_ ,S/Au(lll)

system is

_ideally

polarizable.

4.

_{Copper electrodeposition}

on

alkanethiolate-covered

Au(lll)

4.1 From UPD islands to OPD

_layers

We

_begin

the

_presentation

_{of the copper}

_deposition

studies with the voltam-metric

_experiments.

Here

_voltammetry

is not

_simply

an

analytic

tool,

but also the methodto achieve

_potential

control of the

_system

(for

the STM and XPS

_experiments,

we started at a

potential

of 400 mV where copper

de-posits

neither on bare nor on thiolated

gold,

then lowered the

potential

in

steps

of

20,

50or 100

mV).

The

_{voltammograms}

of thiolate-covered

Au(lll)

samples

in contact

withCu2+ in

_electrolytes

basedon wateras wellas on

ethylene glycol/water

are

usually

featureless with low double

layer

charging

currents

(see

Fig.

11).

CopperElectrodepositionon Alkanethiolate Covered Gold Electrodes 123

In the UPD

_regime,

i.e. above the Cu/Cu2+

_equilibrium

_potential

of 0

mV,

no features were found

either,

albeit in some cases _{very shallow}

peaks

as-signed

to _{copper UPD} showed up. Thisbehaviour reflectsa certain

sample-to-sample variability

of thiolate

_layers

as _also

reported by

_{other groups}

[24, 77].

Below 0mV the current

_{drops slightly, pointing}

_{towards copper}

deposition

at amuch reduced rate

compared

to abare

gold

substrate where

currents

drop

to several ten

_pA/cm2

_(compare

_{e.g. the}currenttransients for copper

deposition

on

Au(lll) [78,

79]).

Rising

the

_temperature

increases the double

_layer

current and reveals small UPD

_peaks

for all

_samples

(see

Fig.

11).

A reasonable

_explanation

is the

_high

_temperature

defect structure

of the thiolate

(see

section

3.2)

that should accelerate the reduction ofCu2+.

Faradaic and

_charging

currents also increase with

_decreasing

thiolate chain

length.

These

_findings

_{compare well with the behaviour in}

_copper-free

elec-trolytes

(see

section

_3.1):

_higher

_temperaturesand shorter

_alkyl

_chains

gen-erally

increase the current, whether it is due to

charging

or Faradaic

cur-rents

-in both cases electrons have tobe transferred

_through

the thiolate

to the

_{adjacent electrolyte phase.}

After

_regulating

the

_temperature

and

_setting

the

_potential

to a fixed

value,

the surface structure canbe followed time resolved

by

in situ STM.

We now _{have to consider three}

_parameters

influencing

_the

morphology:

temperature,

potential

and thiolate chain

_length.

In a certain range of the

parameters

(UPD

for all thiolate chain

_lengths

and

_temperatures

; OPD

only

for > 12atlower

_{temperatures,}

seeTable

1)

the

_system

showeda

surpris-ing

topography: monolayer

high1

islands with an_{averagediameter of 2} nm cover 10to 15% of the surface

(Fig.

12).

They

are

randomly

distributed and also found in the substrate

vacancies;

they

are not

pinned

at

_steps

or at

domain boundaries. The initial nucleation of these islands couldnot be fol-lowed. Further

_{growth proceeded}

eithernot or

extremely slowly

atless than 0.7nm in twohours.

_Apart

from the fact that island formation in the UPD range is a rare

_phenomenon

(but

see

_[37]

and

_[38]),

the observed islands have the

_unique

_property

of

_changing

neither size nor

shape

for

hours,

i.e.

they

can

truly

be addressedas stable orat least metastablenanostructures. At all other conditions

(OPD

for < _{12 and OPD for all at 345}

_K)

_a

pseudo-layer-by-layer growth

was found

_(Fig.

13):

_{islands grow until}

_they

coalesce.

_Only

islands

_already

nucleated in the UPD

_regime

grow,nofurther

islands nucleate. In electrochemical terms, the nucleation is

instantaneous,

not

progressive,

in contrast to _copper

deposition

on bare

Au(lll) [79].

When a

monolayer

is almost

filled,

the next

layer

nucleates on

_top.

How-ever, the

growth speed depends strongly

on the

_temperature

and the chain

length.

At 300

K,

even the shortest chain

(n

=

6)

system

exhibits a _very

slow

_growth

(at

leastonehour per

layer).

At

high

temperatures,

all thiolate

layers

allow a

rapid growth

in the range of minutes per

layer.

1 _The

monolayer heights

of Au(lll) (0.236 nm) and Cu(lll) (0.208nm) are too

Table 1.Copper growth scenarioasfunction of alkanethiolate chainlengthn,temperature and

_deposition

_{potential range} (OPD: <0mV, UPD: >0mV).

room temperature elevated temperature short chain UPD 2D islands OPD

_{layer-by-layer}

UPD 2D islands OPD layer-by-layer long chain UPD 2D islands OPD 2Dislands UPD 2D islands OPD layer-by-layer

Fig.

12. In situ STM

_{image showing}

the 2D copper island formation on aCls

thiolate-coveredAu(lll)surface in the UPD range.Electrolyte: 50 inMH2S04 + 1 mMCuS04.

Electrode

_potential:

30mVvs.Cu/Cu24.

Tunneling

parameters:I =_0.3nA,_V =200 mV.

Image size: 220nm + 220nm.

For a

special

set of the

_parameters

_(short

chain

thiolates,

300 K,

OPD)

we observed a fractal

shape

of the

growing

and

coalescing

islands as

visualized in

_Fig.

14: each island

_develops

three 'arms' which can

again

sym-Copper ElectrodepositiononAlkanethiolate Covered Gold Electrodes 125

Fig.

13.In situ STM picture

exemplifying

the _{OPD copper} layer-by-layer

growth

ona Cothiolate-covered Au(lH)electrode at 300K. Electrode

_potential:

-90mV vs. Cu/

Cu2+.

_Electrolyte:

50 mM H,S04 + 1 mM CuS04.

Tunneling

parameters: I = 0.4nA,

V = _{110 mV.} Image _size

: 220nmX220nm.

metry

of the

underlying gold

surface and/ortothe

_symmetry

of the

_sulphur

adlattice. Such

_{growth shapes}

are

usually

associated with a

kinetically

hin-dered

_growth

_[80].

To obtain information on the chemical

composition

ofthe

interface,

we emersed

_C18

thiolate

_samples

on which copper had been

deposited

for 3 h

at50or20mV at300

K,

i.e.

_samples

covered

_by

copperislands.

They

were

investigated

with XPS. The overview _spectra

_(see

_Fig.

_2,

_sample

_B)

show the same_{features asin absence of copper, but}

_additionally

_{a Cu}

_2p3/2

emis-sion is

_present

_{(correspondingly}

the

_gold

emission

_decreases).

_Intensity

curves

(Fig.

4)

indicate a

rapid

destruction with

X-ray

_{exposure. The} oxy-gen contamination is

slightly higher

than for the

copper-free samples,

but still

_corresponds

_only

to a minor

portion compared

to _{that of copper. From}

the

_intensity

ratio,

the

_{photoemission}

cross sectionsand the

binding

_energy of Cu

_2p3/2

(932.1 eV)

weconclude that the

deposited

_{copper didnotoxidize}

although

the

_samples

had to endure more than one minute ofcontact with

laboratory

air which shouldsuffice for

_complete

oxidation ofa _copper sur-face. Hence the _{copper is}

_{protected against}

oxidation. In addition to the

Fig. 14. In situelectrochemical STM

_image

of fractal islands that nucleatedduring the

growth

of the first copper

layer

on a_C6thiolate-coveredAu(1ll)electrodeat300K. The

image wasobtained after80min of OPD

polarization

at-110mVvs.Cu/Cu"

.

Electro-lyte:

50mM FESO4 + 1 mM CuSGv

Tunneling

parameters: I = 0.5 A, V = 120mV.

Image size: 175nmX175nm.

grazing

emission

(see

Fig.

3).

This

_emphasizes

_{that the copper}

_penetrated

the thiolate

_{layer during}

the

_deposition.

4.2 Structure of

_{electrodeposited}

_copper

The formal electrode

_potential

(measured

here with the Cu/Cu2+ reference

electrode)

we_quoteisnottheoneof the

_{thiolate/electrolyte}

interface. Hence

theoretically

one cannot

_expect

a _{copper bulk}

deposition

at 0mV

_(even

at

highest

Cu2+

concentrations).

However, the

_{potential drop}

in the thiolate

layers

should not exceed 100mV since the

_{overpotential}

_{for copper} bulk

deposition

of about 150 mV is the same as on bare

Au(lll)

and

Au(100)

[18].

The reduction of Cu2+ in acidic media

_produces

copper metal. In our case it grows in

islands,

and the first

question

we havetoposeconcerns the copper islands' distance from the

gold

surface. We

already disproved

the idea of _copper

_residing

in thiolate-free surface areas

[29]

since such areas would result in clear

Cu/Au(lll)

UPD

_peaks

for all

_samples.

Fig.

15 illustrates the two further

principal possibilities

that arise:

first,

the copper islands could be located between thiolate and

_gold

(CH3(CH2)„ ,S/Cu/Au(lll)), second,

they

could be on

_top

of the thiolate

layer

(Cu/CH,(CH2)„_,S/Au(lll)).

We here discard other models because

un-Copper ElectrodepositiononAlkanethiolate Covered Gold Electrodes 127

a)

b)

Au Cu _OS · C

Fig.

15.Schematicrepresentationof the

_possible

structuresof the

_{Cu/thiolate/gold}

inter-face: _copper

_{electrocrystallized}

below_(a)and above(b)the thiolate

_layer.

likely

(e.g. alloying

into the

_gold

surface that is

_only

known for the

(110)

face

[81]).

The second model would

_postulate

_{isolated copper islands} con-nectedto the

_methyl

endgroups

(see

Fig.

15(b)).

Aswe

already pointed

out

[27],

one

might

think that the necessary electron transfer to the

_top

of the chains should

_{nicely explain}

the slowformation

_(but

seesection 4.4

!).

How-ever, several

_arguments

speak against

the model: ifoneobserves

UPD,

the

growing

metal should have a

_stronger

interaction with the substrate than with_{itself. For copper}on

_top

of thethiolate

layer,

this is

impossible

because

the Cu—H

_(or

Cu

_—CH3)

interaction is _very weak. The islands even with-stand

_rinsing

_{with water, i.e.}

_they

are

strongly

bound.

Also,

heating

involv-ing

the formation of

_{alkyl gauche}

defects _should shorten some

(but

not

_all!)

of the thiolate chains below an island. Then the island that behaves as a

rigid

copper

layer

'disk'

(compared

to the thiolate

_layer)

would remain attached

_only

to the intact

thiolates,

but the distance to the

_gold

surface would be the same as at low

_temperature

(when

all molecules are

intact).

Hence the electron transfer

_probability

would be the same at low and

_high

temperatures,

and copper should

deposit

at the same rate; this is not ob-served.

Furthermore,

let us assume an island that

additionally

has somecontact to the substrate

_(i.e.

a _{copper column}

through

the

thiolate).

Such an island of copper atoms

should,

in absence of

_strongly

bound coadsorbates

(SO2-does not

_qualify!),

continue to _{grow 3D}

[22]

as

Cu/Au(lll),

in contrast to

the observations.

_Clearly,

if the thiolate

_layer

is

_{damaged by large}

_negative

Copper placed

below the thiolate

_(Fig.

_15(a))

is in accordance with the

following

facts:

_firstly,

the XPS data show an

_{angular dependence}

of the Cu

_2p

emission which mustbe duetoconsiderable amounts of

material,

i.e. thiolate

molecules,

on

_top

of the copper.

Secondly,

the

resulting

Cu-S bond is

_strong

andcan

easily explain

the

passivation

of copper

[29]

:it is

impossi-ble to rinse off the

islands,

_they

withstand very

positive potentials,

copper doesnot reactwith air which would

_{produce higher}

Ols XPS

_signals.

Note that sucha

thiolate/copper

structure can be calleda

coadsorption

structure, and

_examples

like

_{I/Ag/Pt(lll),}

detected after

_{electrodeposition}

of silveron

I/Pt(lll) (not

adsorption

of iodide on

Ag/Pt(lll)!)

are well documented

[33, 34, 73].

The

_question

we haveto_{address is whether copper} atoms can

penetrate

a thiolate

layer;

in section 4.4 we will show that this is indeed

possible.

Here we mention that silver

evaporated

on

thiolate/Au(l

11)

in vacuum has indeed been observed to

_penetrate

the

_layer

[11,

82].

Let us

also mention that thiolates

_readily

adsorb on

Cu(lOO) [83]

and that even

multilayer

copper thiolates can form

[84].

All these

_arguments

lead us to

propose a

CH3(CH2)„_IS/Cu/Au(lll)

structure.

Further structural details are much more _difficultto observe. This con-cernsthe lattice of the copperatoms which could be

(1

X

1)

asina

complete

UPD

_layer

[15,

85,

86],

a

honeycomb (73

X

73)R30°

lattice as in the

incomplete

Cu + SO2"

_coadsorption

UPD

_layer

on

Au(lll)

[14,

86-88]

or a normal

(73

X

73)R30°

structure. All three would offer reasonable

adsorption

places (e.g.

hollow

sites)

for the

_sulphur

atoms. A more intricate feature is the structure of the islands' rims. Here one could

probably

find a

topography

that is

_distinctly

different from that in the island

-thus one_may

explain

the

_surprising

_stop

of

_growth

andhencethe instantaneousinstead of

progressive

nucleation. Since this is

_only

the case for

long

thiolate chains

and low

_{temperatures,}

we can

speculate

that

especially

the

alkyl

chains

adjust

themselves ina

special

structure.

However,

STM observations

(which

for the currents we

employed probe

the

sulphur

atoms rather than the

chains)

did not

yield

_{any clues.}

4.3 Thermal behaviour

For an

interpretation

of

_temperature

phenomena

we have to consider the

changes

that

_develop

in the thiolate

_{layer during annealing.}

The

_gradual

development

of thiolate defect structures such as

gauche

conformations,

new twist

angles

or

possibly

azimuthal disorder

(see

section

3.2)

will

again

allow for

_increasing

_penetration

of

ions2,

this time

_including

Cu2+. As for

Fe(CN)<7,

a reduction will occur much faster.

Doubtlessly,

the

complete

penetration

is now even more facile.

However,

we also have to

explain

a

qualitative change:

islands grown on thiolates of all chain

lengths

can

co-2 _{On the}

fate of thehydrationshellwe can

only speculate.

Thisiseventruein absence of thiolate [89]!

Copper ElectrodepositiononAlkanethiolateCovered Gold Electrodes 129

alesce at

_high

_{temperatures.}

This means that the unknown barrier

against

further

_{growth (presumably}

at the island

_rim)

is overcome

by

thermal acti-vation. The

_{pseudo-layer-by-layer}

_growth

isduetothepresenceof the

thiol-atewhich

_always

_stays

on_top; otherwise the

growth

(on

_pure

copper)

_would carry on much faster and 3D nodules would result. This is the classical

picture

ofa surfactantthat hinders the

growth

in thedirectionnormaltothe surfaceas alsoobserved when copper was

deposited

on

gold

in presence of

crystal

violet

[20].

However,

our surfactant is adsorbed in a well-charac-terized

_geometry

and will neither desorb nor readsorb

during

deposition.

Thus weexcludeoneof the many processes that

complicate

the

understand-ing

of surfactant action in

_{electrodeposition.}

At

_high

_temperatures

substrate

_steps

become mobile

_('strong

activation',

seesection

3.2).

When we

_accept

_{that copper is bound tothe substrate}

_(see

section

4.2),

copper island rims — copper

step

edges

—

should be at least

as mobile as the substrate

_steps.

This is

equivalent

to an incrased island

perimeter mobility,

i.e. the fractal island

_shapes

transform to

_compact

shapes.

Obviously,

the barrier

_against

further

_growth

that

_prevails

at low

temperature

is _overcome, and further

_growth

and coalescence can occur. This

_arguments

are in line with the

theory

of surface diffusion limited

aggregation

that has

_successfully

been

_applied

to

explain

ramified metal islands grown in vacuum.

However,

the fact that

_temperature

also enhances copper

penetration

and thus

_deposition

rate makes a

quantitative

theoretical

_modelling

_very

_difficult,

sincetwo

_{interdependent}

barriers

(thiol-ate

_penetration

and the barrier at the island

_rim)

are overcome atthe same

time.

4.4 Potential

_dynamics

—

copper

deposition

rates

We shall start the discussion with a look at the

_permittable

limits of the

potential.

At the

_positive

_end,

we dissolve the _copper islands at

_potentials

several hundred mV

_positive

of the Cu/Cu2+

_equilibrium

value. Note that

we haveto wait several hours at 400mV

_{(300 K),}

_and some islands resist

dissolutioneven at1000mV

_[29]

_which

_impressively

_{demonstrates the} pas-sivation. As one

_expects,

the dissolution is much faster when the

tempera-ture is raised.

Inthe cathodic

_{regime, only}

well below -200 mV the

_{layer growth}

can

change

to 3D

_deposition

of nodules with diameters above 100nm. Since

voltammograms

show _{copper UPD}

_peaks

after such

_negative

excursions,

some bare substrate

_spots

_appear to exist which would

_clearly

cause the observed

_growth

of

_large

nodules.

For the UPD processes, no currentflow is found since the

_growth

is so

slow that the

_{resulting peaks}

would be fartoo shallow

(sub-nA

range).

In the OPD range, the detectedcurrents

_{usually predict}

a

deposition

ofa

mono-layer

of copper

during

several minutes which is

-at 300

than observed

_by

STM. The

_discrepancy

canbe

explained by

thedestruction

of the

_sample

close to the viton

_ring

where we

occasionally

found visible copper

deposits;

this means that copper had here

deposited

at least 100 times thicker than elsewhere. Thus the

_global

current is unreliable at least

when its value is small. Therefore we estimate the

deposition

rate from

STM measurements:

typical

rates are 10% of a

monolayer

in one to two

hours

(UPD),

one

layer

in two _{hours for copper}

_{layer growth}

on _{short and} medium chain thiolates

(OPD)

and one

layer

_{per minute} at

_high

tempera-tures

(OPD)\

Assuming

a Cu

(1

X

1)

_geometry

on

Au(lll)

with a

mono-layer

charge

of 440

_pC/cm2,

this

_gives

rates of0.01

pA/cm2,

0.06

µ /cm2

and 7

_{µ /cm2, respectively.}

We can set limits to the

experimental

rates:

0.005

_{µ /cm2}

< r < 10

_{µ /cm2.}

(1 )

In orderto exclude or

verify possible

mechanisms,

it is instructive to com-pare those rates to rates calculated under

typical

conditions. Let us first assume that bulk diffusion of Cu2+ is the

rate-limiting

(slowest)

_process. Then the Cotrell

_equation

would

_apply

[12]:

where F = 96485

C/mol,

D is the diffusion _constantof Cu2+ in

electrolyte

(5.7

· 10~6 cm2/s

[90]),

c0 the Cu2+ concentration

(1 mM),

t the

time,

R = 8.314

J7(mol

·

),

the

_potential

_step

_causing

the

_deposition

_(e.g.

-100

_mV)

and the _temperature. We should then find

(comparison:

440

_{µ /cm2}

accountsfor aCu

₍₁

X

₁₎

_monolayer

on

Au(lll)

deposited

inone

second)

which is about 100 times

larger

than the transient we observed

during

the first minute of

polarisation.

This means that the

diffusion is not the

_limiting

factor.

This could leave the electron transferas the slowest process. First letus

concentrate on the caseof copper

deposition

on _copper, e.g. copper islands. Then the slowest

_possible

rateis thatatzero

overpotential

where dissolution and

_deposition

currentscancel each other. Their

value,

the

exchange

current

density,

is

[91]

r, = 2

FcQ

jDhTt

[1

- exp

(2

FA

_E/RT)]

(2)

r, » 260

µ /cm2

·

JWt/sj

(2a)

r2 « ₅₀₀

µ /cm2

(3)

3 _{The estimatesare}

rough,

hence the copperatom

_density

in the islands(see section