A

s

ict

t©

K

1

brlabtc

tliamc

0d

Electrochimica Acta 52 (2007) 7486–7495

Electrochemical study of modified bis-[triethoxysilylpropyl]tetrasulfide silane films applied on the AZ31 Mg alloy

M.F. Montemor a,∗, M.G.S. Ferreira a,b

a Instituto Superior Tecnico, ICEMS, Av. Rovisco Pais, 1049-001 Lisboa, Portugalb University of Aveiro, Department of Ceramic and Glass Engineering, 3810-193 Aveiro, Portugal

Received 5 July 2006; received in revised form 10 October 2006; accepted 17 December 2006Available online 26 January 2007

bstract

This work investigates the protective behaviour of bis-[triethoxysilylpropyl] tetrasulfide silane pre-treatments on the AZ31 Mg alloy. The silaneolution was modified by the addition of cerium nitrate or lanthanum nitrate in order to introduce corrosion inhibition properties in the silane film.

The corrosion behaviour of the pre-treated AZ31 magnesium alloy was studied during immersion in 0.005 M NaCl solution, using electrochemicalmpedance spectroscopy and the scanning vibrating electrode technique (SVET). The electrochemical experiments showed that the presence oferium ions or lanthanum ions improve the protective behaviour of the silane film. The SVET experiments evidenced that the presence cerium in

he silane film led to an important reduction of the corrosion activity.

The results demonstrate that either cerium ions or lanthanum ions can be used as additives to the silane solutions to improve the performance ofhe pre-treatments for the AZ31 magnesium alloy.

2007 Elsevier Ltd. All rights reserved.

mrouaoadaiocts

eywords: Magnesium alloy AZ31; Silanes; SVET; Cerium; Corrosion

. Introduction

The use of magnesium alloys for different applications haseen increasing during the last years. The need of lighter mate-ials for specific applications such as the electronic industryed to an increasing interest on the corrosion behaviour of Mglloys. They are used in the production of cellular phones, note-ooks and other electronic devices. The alloys are also used inhe automotive and aeronautical industry, mainly in structuralomponents.

The increasing interest on the use of Mg alloys is due tohe fact that these alloys exhibit an attractive combination ofow density, high strength, good damping capacity, castabil-ty and machinability. The AZ grades of Mg alloys containluminium and zinc. Zinc is effective to increase strength. Alu-

inium increases strength and corrosion resistance, but reduces

astability and weldability.

∗ Corresponding author. Tel.: +351 218417234; fax: +351 218419771.E-mail address: mfmontemor@ist.utl.pt (M.F. Montemor).

btrcdm

013-4686/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.oi:10.1016/j.electacta.2006.12.086

Mg alloys show a surface film composed essentially of aixture of MgO and Mg(OH)2, which provides reasonable cor-

osion protection in air, but becoming very unstable in aqueousr in high humidity environments. Thus, Mg alloys are generallysed in environments with reduced aggressiveness because theyre very sensitive to corrosion attack. To improve the lifetimef Mg alloys different chemical treatments and organic coatingsre generally applied on their surface. In addition to improvedurability, organic coatings also confer specific functionalitiesnd aesthetic properties. For such purposes, adhesion of the coat-ng to the bare substrate is an important issue for the durabilityf the coated material under service. Thus, the performance ofoated Mg alloys is strongly dependent on the corrosion resis-ance and on the adhesion properties of the coating to the metallicubstrate.

The corrosion resistance of Mg and its alloys can be enhancedy different procedures. Anodising [1–4] has proved to improvehe corrosion resistance of the alloys under aggressive envi-

onments. Anodic oxidation produces an oxide film with goodorrosion resistance and reasonable adhesion properties. Theeposition of chemical conversion layers using modified per-anganate solutions led to the formation of a nearly protective

lectro

uicsotoapf

op

wpTmmtmftc

oasswfapvpatbc

stwap(Trct

2

2

f

tl(

wwmsursu

cwi

2

seTw42

tPuaifp

sc

3

3

dtsia

spca

M.F. Montemor, M.G.S. Ferreira / E

niform coating [5,6]. The use of cerium rich conversion filmss also reported in literature [7,8]. It is generally accepted thaterium ions are effective corrosion inhibitors and that they con-titute a promising alternative to the use of pre-treatments basedn Cr (VI)-containing formulations. Brunelli et al. [7,8] reportedhat cerium conversion films increased the corrosion resistancef the Mg alloys because they ennoble the corrosion potentialnd decrease both the anodic and cathodic currents. An acidre-treatment prior film formation increased the corrosion per-ormance of the Ce conversion films.

Literature [9] also reports that the corrosion performancef painted samples pre-treated with phosphate exceeds that ofainted Cr conversion layers.

Conversion films lead to the formation of a protective barrier,hich delays or inhibits corrosion onset. However, the adhesionroperties of these layers are somewhat poor and often they fail.hus, adhesion promoters have also been tested. Surface treat-ents based on organofunctional solutions became an attractiveethod to reduce the corrosion rate of metallic substrates and

o enhance adhesion. Functional silanes [10] and self assembledonolayers [11] have been successfully tested as pre-treatments

or Mg alloys. These pre-treatments provide corrosion pro-ection in addition to surface functionality, improving theompatibility of the metallic substrate with the painting systems.

Silane based pre-treatments lead to the formation of a thinrganic coating that provides corrosion protection because itcts as a physical barrier that delays the ingress of aggressivepecies towards the metallic substrate. The performance of theilane coatings can be improved through the addition of speciesith corrosion inhibition properties to the silane solutions used

or the pre-treatments. Among the possible additives, ceriumnd lanthanum salts show promising properties, since they haveroven to inhibit the corrosion processes on steel [12], gal-anised steel [13,14] and aluminium and its alloys [15,16]. Inrevious works it was demonstrated that cerium can be useds dopant in hybrid sol–gel and mulftifunctional silane pre-reatments [17–21]. The presence of this additive improved thearrier properties of the pre-treatment layer and decreased theorrosion rate of galvanised steel and aluminium substrates.

In the present work bis-[triethoxysilylpropyl] tetrasulfideilane pre-treatments modified with cerium nitrate or with lan-hanum nitrate were applied on the AZ31 Mg alloys. The samplesere tested during immersion in 0.005 M NaCl, which is a very

ggressive environment for the Mg alloys. The pre-treated sam-les were studied via electrochemical impedance spectroscopyEIS) and via scanning vibrating electrode technique (SVET).he results showed that silane based pre-treatments provide cor-

osion protection of the AZ31 Mg alloy and that the presence oferium or lanthanum induced a decrease of the corrosion rate ofhe substrate.

. Experimental procedure

.1. Materials and solutions

AZ31 Mg alloy was obtained from Goodfellow. The sur-ace of the alloy was polished with SiC paper of grit 2400. All

soat

chimica Acta 52 (2007) 7486–7495 7487

he substrates were ultrasonically cleaned using acetone. Fol-owing cleaning the panels were washed with deionised waterMillipore) and dried in air.

The bis-[triethoxysilylpropyl] tetrasulfide silane (BTESPT)as obtained from Sigma/Aldrich. The pre-treatment solutionas prepared by dissolving 5% (v/v) of silane in a mixture ofethanol (90%, v/v) and the aqueous solution 10−3 M of nitrate

alt (5.0%, v/v). Two aqueous solutions of nitrate salts weresed: Ce(NO3)3 or La(NO3)3. The final concentration of theare earth cation in the silane solution was 5.0 × 10−5 M. Theilane solution was stirred during 1 h and kept for 3 days beforese.

The pre-treatment consisted in the immersion of the metallicoupons in the silane solution for 10 s. The excess of solutionas removed by using air stream and then the panels were cured

n a Memmert oven at 120 ◦C for 40 min.

.2. Electrochemical measurements

The scanning vibrating electrode technique (SVET) mea-urements were performed using the Applicable Electronicsquipment, controlled by the ASET program (Sciencewares).he vibrating electrode was made of platinum–iridium coveredith polymer, leaving only an uncovered tip with a diameter of0–50 �m. The distance of the tip to the surface was kept at00 �m.

The EIS measurements were performed at room tempera-ure in a Faraday cage, using a Gamry FAS1 Femtostat plus aC4 controller board. A three-electrode electrochemical cell wassed, consisting of the working electrode (3.15 cm2 of exposedrea), the saturated calomel electrode as reference and the plat-num as counter electrode. The measuring frequency rangedrom 5 × 105 Hz down to 5 × 10−3 Hz. All the experiments wereerformed at the corrosion potential.

Electrochemical experiments were performed during immer-ion in 0.005 M sodium chloride solution. Reproducibility wasonfirmed on a minimum of two identical samples per condition.

. Results and discussion

.1. Electrochemical study

The EIS plots obtained for the bare metal (no pre-treatment)uring immersion in 0.005 M NaCl are presented in Fig. 1. Dueo the high corrosion rate of the blank substrate in this aggressiveolution the EIS spectra were taken only during the first 24 h ofmmersion. It can be observed that the impedance increased bybout one order of magnitude during the immersion period.

The EIS results obtained for samples treated with the silaneolutions are presented in Figs. 2–4. For the silane treated sam-les the EIS measurements were performed during 1 week. Theomparison of the EIS spectra obtained in the different samplesfter 1 day of immersion is depicted in Fig. 5. The presence of the

ilane film increased the low frequency impedance by more thanne order of magnitude comparatively to the blank sample. Theddition of cerium to the silane film increased the impedance bywo orders of magnitude.

7488 M.F. Montemor, M.G.S. Ferreira / Electrochimica Acta 52 (2007) 7486–7495

Fi

wltcr

sdiafiootibcbfis

Fa

acicspwsficttapuptake and the silane coating capacitance and resistance could

ig. 1. EIS bode plots obtained for the bare AZ31 Mg alloy during 1 day ofmmersion in 0.005 M NaCl.

For the blank sample, the EIS spectra (Fig. 1) revealed aell-defined time constant around 10 Hz and, another one, at

ower frequencies (<0.1 Hz). The high and the low frequencyime constants were attributed to the response of charge transferontrolled processes and mass diffusion-controlled reactions,espectively.

The EIS spectra obtained for the samples treated with theilane solution modified with cerium nitrate (Fig. 2) showedifferent shapes during immersion. During the first 2 days ofmmersion the spectra revealed the presence of a time constantt frequencies around 100 Hz, which was attributed to the silanelm properties (i.e., capacitance and resistance) and anotherne at lower frequencies (10 Hz), in the same frequency rangef that observed for the blank sample, which was attributedo charge transfer controlled processes at the metal/silane filmnterface. Latter on the shape of these spectra changed andecame more identical to that obtained for the blank sample,onsisting of one time constant around 10 Hz and another one

elow 0.1 Hz. For the systems, treated with BTESPT modi-ed with La (Fig. 3) and non-modified BTESPT (Fig. 4) thehapes of the spectra were similar to those of the blank sample

not

ig. 2. EIS bode plots obtained for the AZ31 Mg alloy pre-treated with BTESPTnd Ce(NO3)3 during 1 week of immersion in 0.005 M NaCl.

nd, contrarily to the spectra obtained for the system containingerium, the silane layer did not reveal a defined time constantn the high frequency range of the EIS spectra. From Fig. 5 itan be seen that the spectra obtained for the BTESPT treatedample are very close to those of the blank sample. The sam-le treated with BTESPT and La shows a broader phase angle,hich seems to include two overlapped time constants in the

ame frequency range of those observed for the Ce-containinglm. However, the response of the silane film could not belearly distinguished. The shape of the EIS spectra suggestshat the barrier effect of the silane film on the BTESPT andhe BTESPT plus La pre-treated samples were not as markeds that observed for the cerium-containing films. Despite theresence of the silane layers there was a very fast electrolyte

ot be clearly separated from the response of the processesccurring at the interface metal/silane coating. The absence ofhe time constant characteristic of the silane coating was also

M.F. Montemor, M.G.S. Ferreira / Electrochimica Acta 52 (2007) 7486–7495 7489

Fa

rl

ntfiTtihagbb[

a

h

Fd

CuctfaoM

•

ig. 3. EIS bode plots obtained for the AZ31 Mg alloy pre-treated with BTESPTnd La(NO3)3 during 1 week of immersion in 0.005 M NaCl.

eported for galvanised steel substrates treated with aminosi-anes [22].

Detailed interpretation of the EIS plots was performed byumerical simulation, using an equivalent circuit composed ofwo-time constants (Fig. 6). This equivalent circuit reasonablyts the experimental results as can be confirmed in Fig. 6b and c.his equivalent circuit included constant phase elements (CPEs)

o simulate the capacitive response of the EIS spectra. The CPEs generally attributed to distributed surface reactivity, surfaceeterogeneity, roughness or fractal geometry, electrode porosity,nd to current and potential distributions related with electrodeeometry. The CPE is associated with an exponent (α) and it haseen considered to represent a circuit parameter with limitingehaviour as an ideal capacitor if α = 1 and a resistor for α = 023].

Fitting parameters for the blank sample are presented in Fig. 7nd for the silane treated samples are presented in Figs. 8 and 9.

Fig. 7a depicts the evolution of the low frequency CPE andigh frequency CPE for the blank sample. The high frequency

•

ig. 4. EIS bode plots obtained for the AZ31 Mg alloy pre-treated with BTESPTuring 1 week of immersion in 0.005 M NaCl.

PE was nearly constant and around 4 × 10−5 F/cm2. Such val-es indicate that for the blank sample the high frequency timeonstant is due to charge transfer controlled reactions. The resis-ance associated with this time constant increased with time,rom 6 × 103 to 7 × 104 � cm2. This increase can be explainedssuming that the corrosion products formed at the surface devel-ped a protective layer. The following reactions occur when theg substrates are immersed in neutral NaCl solution:

Cathodic reactions:

O2 + 2H2O + 4e− → 4OH− (1)

2H2O + 2e− → 2OH− + H2↑ (2)

Anodic reactions:

Mg → Mg2+ + 2e− (3)

7490 M.F. Montemor, M.G.S. Ferreira / Electrochimica Acta 52 (2007) 7486–7495

FaL

•

Thdtfapicl

b(H

Fig. 6. Equivalent circuit used for numerical simulation of the EIS spectra (a);fiNF

rcaN

ft(nho(om

ig. 5. EIS bode plots obtained for the blank AZ31 Mg alloy and for the samelloy pre-treated with BTESPT, BTESPT plus Ce(NO3)3 and BTESPT plusa(NO3)3 after 24 h of immersion in 0.005 M NaCl.

Consequently the following reaction also occurs:

Mg2+ + 2OH− → Mg(OH)2↓ (4)

he precipitation of corrosion products led to the formation of aydroxide layer, which thickens with time and progressively hin-ers the corrosion processes. According to literature [24] whenhe corrosion rate decreases, a balance may occur between theormation and dissolution of the Mg(OH)2 layer, which reachesconstant thickness. However, this layer is porous and not fullyrotective, allowing the diffusion of charged species involvedn the corrosion processes. The low frequency response isharacterised by CPE values around 1 × 10−3 F/cm2, which areikely to be associated with mass transfer controlled processes.

For the blank sample, during the first hours of immersion,oth the high frequency and low frequency resistances increasedFig. 7b) reflecting the build-up of the corrosion products film.owever, after about 5 h of immersion the low frequency

boB1

tted spectra for BTESPT and Ce(NO3)3 after 3 days of immersion in 0.005 MaCl (b) and respective fitting parameters obtained using ZView2 (c). CPE units:/cm2; R units: � cm2.

esistance (RLF) started to decrease, reflecting an increase of theonductivity of the corrosion products layer. Identical trendsre reported in literature for anodised Mg alloys immersed inaCl [25].For the silane treated samples, the evolution of the high

requency parameters is depicted in Fig. 8. The BTESPTreated sample showed CPE values that increased with timefrom 7 × 10−7 to 3 × 10−6 F/cm2), being one order of mag-itude lower than those observed for the blank sample. Theigh frequency CPE decreased even more in the presencef lanthanum or cerium. Comparatively to the BTESPT filmwithout additive) the presence of lanthanum or the presencef cerium decreased the CPE values for more than one order ofagnitude. These trends suggest that the barrier layer created

y the modified silane film presented a decreased numberf conductive pathways comparatively to the non-modifiedTESPT film. The high frequency resistance was above× 106 � cm2 for the BTESPT film with cerium and one order

M.F. Montemor, M.G.S. Ferreira / Electrochimica Acta 52 (2007) 7486–7495 7491

FM

otwbaostifi

toFiopbCF

Fp

afir

toeB

iat

rt(ito understand the effect of inhibitors on localised defects. Thus, it

ig. 7. Evolution of the capacitance (a) and resistance (b) for the blank AZ31g alloy sample during immersion in 0.005 M NaCl.

f magnitude lower for the silane films without additive. Forhe BTESPT film-containing lanthanum the resistance valuesere lower than those of cerium and decreased with time,ut did not reach the values observed for the film withoutdditives. The increased resistances and decreased capacitancesbserved in the high frequency range of the EIS spectra for theilane pre-treated samples are mainly due to the presence ofhe silane film, which creates a protective barrier. This effects more pronounced for the cerium or lanthanum modifiedlms.

The lowest frequency processes observed for the silanereated samples are likely to be associated with the developmentf corrosion activity and growth of porous corrosion products.or the BTESPT sample, the low frequency CPE values (Fig. 9)

ncreased from 8 × 10−5 to 8 × 10−4 F/cm2 during the first hoursf immersion, reflecting an increase of the active area and theresence of diffusion-controlled mechanisms as observed for the

lank sample (Fig. 7). Fig. 9 also shows that the low frequencyPE increased with time, being lower for the modified films.or the Ce-containing film, the low frequency resistance was

wfie

ig. 8. Evolution of the high frequency capacitance and resistance for the silanere-treated samples in 0.005 M NaCl.

bout two orders of magnitude above that of the non-modifiedlm and kept constant with time, revealing that the corrosionate was strongly reduced.

During the first 2 days of immersion, the low frequency resis-ance values for the La-modified film were identical to thosebtained with cerium, but after that period they decreased. Nev-rtheless, they were always higher than those obtained for theTESPT film.

The trends observed in Figs. 8 and 9, indicate that the mod-fied films are more protective then the non-modified films andlso indicate that cerium is the most effective additive in reducinghe corrosion rate of the AZ31 substrate.

The EIS measurements provide information on the averageesponse of the exposed surface. The localised electrochemicalechniques, such as the scanning vibrating electrode techniqueSVET) allow local electrochemical information to be obtainedn the form of current density maps. These maps are very helpful

as decided to investigate the protective properties of the silanelms with and without additives by using the scanning vibratinglectrode technique (SVET).

7492 M.F. Montemor, M.G.S. Ferreira / Electro

Fp

afpci

wia2o

boaaAdcs

oArd

oHti

wsbrcmattto

sdTdaettwadsw

bcafiTsSiltomatt

a

ig. 9. Evolution of the low frequency capacitance and resistance for the silanere-treated samples in 0.005 M NaCl.

To understand the performance of cerium or lanthanum ionss corrosion inhibitors, the pre-treated samples were immersedor a period of 24 h in the 0.005 M NaCl solution and after thiseriod an artificial defect was made on the surface. This defectreates a preferential anodic site and therefore the corrosionnhibition properties of the additives can be studied.

For the BTESPT treated sample localised corrosion activityas noticed after 6 h of immersion (before defect). After 24 h of

mmersion a defect was created on the surface and a rise of thenodic activity could be observed (Fig. 10a and d). After more4 h of immersion, the anodic currents decreased by about onerder of magnitude (Fig. 10b and d).

A quantitative approach of the SVET results was performedy plotting the anodic and cathodic current densities measuredver an horizontal line that crossed the defect (Fig. 10d). Thenodic current densities measured after defect formation showedsharp increase, reaching values around 70 �A/cm2 after 1 h.

bout 2 h after defect formation the anodic currents started toecrease, approaching stable values around 10 �A/cm2. Theathodic currents were around −10 �A/cm2. The line scanhows that preferential magnesium oxidation occurs at the centre

a

ts

chimica Acta 52 (2007) 7486–7495

f the defect, whereas reduction reactions surround the defect.n interesting feature is that the cathodic current densities

eached a maximum at a distance around 600–700 �m from theefect centre.

In addition to the artificial defect more anodic areas appearedn the defect vicinity, becoming larger with time (Fig. 10b and c).owever, both the anodic and cathodic currents decreased due

o the protective behaviour of the corrosion products produced,n good agreement with the previous EIS results.

The SVET maps obtained on the BTESPT films modifiedith lanthanum nitrate were similar to those of BTESPT. Corro-

ion activity was detected during the first 24 h of immersion (i.e.,efore defect formation). After defect formation significant cor-osion activity was observed. At the defect the maximum anodicurrent density reached a maximum around 20 �A/cm2 after fewinutes and than decreased for longer immersion times. During72 h immersion period, other smaller anodic areas appeared on

he surface, revealing that the corrosion inhibition properties ofhe lanthanum ions were not completely effective. Nevertheless,he anodic or the cathodic currents were always lower than thosebserved for the BTESPT film without additives.

The SVET results obtained on the Ce-modified film are pre-ented in Fig. 11. During the first 24 h of immersion (i.e., beforeefect formation) no activity could be observed on the surface.he currents were around zero over the scanned surface. Afterefect formation, a small anodic spot could be detected and thenodic currents reached values around 3 �A/cm2 (Fig. 11a and). These values were much lower than the values measured forhe BTESPT film without additives. For longer immersion timeshe anodic and the cathodic activity vanished. This behaviouras observed until the end of the tests (72 h of immersion period)

nd illustrates that the corrosion activity was completely hin-ered (Fig. 11c–e). A random distribution of anodic and cathodicites was detected across the line scan and the current densitiesere around zero.The results obtained by the SVET technique can be explained

y the following assumptions: the SVET technique measuresathodic or anodic currents, based on the local flow of cationsnd anions. Defect formation disrupted the protective silanelm, exposing the metallic substrate that started to dissolve.he anodic currents are due to the oxidation of the magne-ium and flow of magnesium ions produced during reaction (3).imultaneously, the cathodic reactions release hydroxyl ions that

ncrease the cathodic currents—reactions (1) and (2). The metal-ic cations (Mg2+) react with hydroxyl ions (OH−) leading tohe formation of negatively charged complexes or precipitationf Mg(OH)2 (reaction (4)). As previously observed in the EISeasurements the build-up of corrosion products seems to haveprotective character. This mechanism explains the decrease of

he anodic currents observed in the SVET results obtained onhe BTESPT treated sample.

In the presence of cerium the anodic activity is much weakers well as the cathodic activity. This means that the flux of anions

nd cations is strongly hindered and corrosion could not proceed.

The presence of cerium improves the barrier properties ofhe silane film as demonstrated by the EIS and introduces corro-ion inhibition properties in the silane films as demonstrated

M.F. Montemor, M.G.S. Ferreira / Electrochimica Acta 52 (2007) 7486–7495 7493

F aftert form2

bsp(ea

•

•

Trficocbebfitttcfit[

i

ig. 10. SVET maps obtained on the sample treated with BTESPT: (a) 15 minhe end of the tests (48 h of immersion) and (d) line scan obtained after defectmm × 2 mm.

y SVET and EIS. To understand this behaviour the overallilane system must be considered. When the silane solution isut in contact with the metallic substrate, the silanol groupsR–Si–OH) formed during hydrolysis of the silane moleculesstablish covalent bonds with the native surface hydroxides,ccording to the reactions depicted below:

Hydrolysis:

3R–Si(OR′)3 → R–Si(OH)3 + 3R′OH (5)

Condensation/polymerisation:

(6)

caTo

defect formation; (b) 24 h after defect formation; (c) image of the surface atation along the line depicted in (c). Currents are �A/cm2 and the scan area is

he assembling of the silane molecules creates a stable Si–Oich network, which grows leading to the formation of the barrierlm. Additionally, the assembling of the silane film modifies thehemical composition of the metallic surface. Thus, the nativexide film, which is originally a mixture of MgO/Mg(OH)2an be converted into a Mg–O–Si interface, which anchors thearrier silane film. This barrier is hydrophobic and resistant tolectrolyte uptake. The number of conductive pathways in thisarrier defines the capacitance and the resistance of the silanelm and therefore its protective character. The presence of addi-

ives, especially cerium increases the resistance and decreaseshe capacitance of the barrier film and therefore cerium enhanceshe barrier properties. The increased resistance and decreasedapacitance suggests that the porosity and/or conductivity of thelm decrease and that the thickness increases. Increased film

hickness was found for sol–gel coatings doped with cerium26].

Literature reports that Ce3+ ions [27–29] can be convertedn Ce4+ ions in solution in the presence of oxygen. These ions

an be incorporated in the silane film, substituting some of the Sitoms, leading to the formation of a modified Si–O–Ce network.hus, a modified film as the one depicted in Fig. 12 is likely to bebtained. Identical assumption was proposed in literature [26]

7494 M.F. Montemor, M.G.S. Ferreira / Electrochimica Acta 52 (2007) 7486–7495

F th cera line sc�

tc

ma

sr

ig. 11. SVET maps obtained on the sample treated with BTESPT modified wifter defect formation; (d) image of the surface at the end of the tests; and (e)A/cm2 and the scan area is 2 mm × 2 mm.

o explain the improved corrosion behaviour of sol–gel coatingsontaining different amounts of cerium nitrate.

The formation of a defect in the silane film exposes theetallic substrate and originates localised anodic and cathodic

ctivity. The increase of pH at the cathodic sites disrupts the

Fig. 12. Incorporation of cerium ions in the silane film.

tta

C

C

Tfish

ium: (a) 15 min after defect formation; (b) 48 h after defect formation; (c) 72 han obtained after defect formation along the line depicted in (d). Currents are

iloxane network, releasing cerium ions. These become free toeact with the hydroxyl ions (reactions (7) and (8)) and lead tohe formation of very insoluble hydroxides, which precipitate inhe cathodic sites that surround the defect, hindering corrosionctivity as confirmed by SVET (Fig. 11):

e3+ + 3OH− → Ce(OH)3 (7)

e4+ + 4OH− → Ce(OH)4 (8)

he presence of Ce4+ and Ce3+ in the silane film was con-rmed by XPS measurements performed on galvanised steelubstrates pre-treated with the same silane, but doped with aigher concentration of cerium (1 × 10−2 M) [17].

lectro

atleths

actaaaiiLaimniioct

4

stt

sfic

aCas

A

R

[

[

[[

[

[

[

[

[

[

[

[

[[

[[

[

[

[

M.F. Montemor, M.G.S. Ferreira / E

On the other hand, when the silane film is immersed in theggressive electrolyte there is deterioration of the outer layers ofhe film as reported in previous works [20–22]. This may causeeaching of cerium ions incorporated or trapped in the outer lay-rs of the siloxane matrix to the vicinity of the anodic areas wherehey precipitate in the form of Ce(III) or Ce(IV) hydroxides,indering the corrosion reactions. Leaching of cerium ions fromilane films was also reported by van Ooij and co-workers [30].

The beneficial effect of lanthanum ions is not so markeds that of cerium. Although these ions are known to haveorrosion inhibition properties too, in the AZ31 substrateshey could not inhibit the corrosion process in the same extents cerium. The effects on the barrier properties of the filmre also poorest than those of cerium. The differences can bettributed to the different chemistry of cerium and lanthanumons. Both rare-earth ions precipitate when the pH increases ast happens at the cathodic sites. Lanthanum ions form mainlya(III) hydroxides, whereas cerium is likely to form Ce(IV)nd Ce(III) hydroxides and oxides. These species are morensoluble then the La(III) hydroxides and therefore they can be

ore effective in hindering the cathodic activity. Similar expla-ation has been proposed to understand the superior corrosionnhibition performance of cerium ions in AA2024 substratesmmersed in NaCl solutions [31]. Nevertheless, the presencef lanthanum enhances the film properties and decreases theorrosion rate of the substrate, making it an interesting dopanto improve the protective properties of silane coatings.

. Conclusions

Pre-treatments for the AZ31 Mg alloy using bis-[triethoxy-ilylpropyl] tetrasulfide silane solutions provide corrosion pro-ection and can be considered a promising technical alternativeo the use of chromates.

The addition of cerium nitrate or lanthanum nitrate to theilane solutions improved the barrier properties of the silanelms and decreased the corrosion rate of the metallic substrate,omparatively to the non-modified silane film.

The presence of cerium led to an important drop of the anodicnd cathodic activity when defects are formed in the silane film.erium ions could inhibit the corrosion processes at the damagedreas, conferring corrosion inhibition properties to the barrierilane film.

cknowledgement

POCTI/CTM/59234/2004.

[[[

chimica Acta 52 (2007) 7486–7495 7495

eferences

[1] L.-H. Chiu, C.-C. Chen, C.-F. Yang, Surf. Coat. Technol. 191 (2005)181.

[2] H.-Y. Hsiao, W.-T. Tsai, Surf. Coat. Technol. 190 (2005) 299.[3] Y. Mizutani, S.J. Kim, B.R. Ichino, M. Okido, Surf. Coat. Technol. 169–170

(2003) 143.[4] J.E. Gray, B. Luan, J. Alloys Compd. 336 (2002) 88.[5] M. Zhao, S. Wu, J. Luo, Y. Fukuda, H. Nakae, Surf. Coat. Technol. 200

(2006) 5407.[6] H. Umehara, M. Takaya, S. Terauchi, Surf. Coat. Technol. 169–170 (2003)

666.[7] K. Brunelli, M. Dabala, I. Calliari, M. Magrini, Corros. Sci. 47 (2005) 989.[8] K. Brunelli, M. Dabala, I. Calliari, M. Magrini, Surf. Coat. Technol. 172

(2003) 227.[9] L.Y. Niu, Z.H. Jiang, G.Y. Li, C.D. Gu, J.S. Lian, Surf. Coat. Technol. 200

(2006) 3021.10] F. Zucchi, V. Grassi, A. Frignani, C. Monticelli, G. TrabanellI, Surf. Coat.

200 (2006) 4136.11] A.L.K. Tana, A.M. Soutar, I.F. Annergren, Y.N. Liu, Surf. Coat. Technol.

198 (2005) 478.12] C. Wang, F. Jiang, F. Wang, Corros. Sci. 46 (2004) 75.13] M.F. Montemor, A.M.P. Simoes, M.G.S. Ferreira, Prog. Org. Coat. 44 (1)

(2002) 79.14] M.F. Montemor, A.M.P. Simoes, M.G.S. Ferreira, Prog. Org. Coat. 43

(2001) 59.15] A.J. Aldykewicz, H.S. Isaacs, A.J. Davenport, J. Electrochem. Soc. 142

(1995) 3342.16] A. Aballe, M. Bethencourt, F.J. Botana, M. Marcos, J. Alloys Compd.

323–324 (2001) 855.17] W. Trabelsi, P. Cecilio, M.G.S. Ferreira, M.F. Montemor, Prog. Org. Coat.

54 (2005) 276.18] M.L. Zheludkevich, R. Serra, M.F. Montemor, K.A. Yasakau, I.M. Miranda

Salvado, M.G.S. Ferreira, Electrochim. Acta 51 (2005) 208.19] M.L. Zheludkevich, R. Serra, M.F. Montemor, M.G.S. Ferreira, Elec-

trochem. Commun. 8 (2005) 836.20] M. Cabral, R.G. Duarte, M.F. Montemor, M.G.S. Ferreira, Prog. Org. Coat.

54 (2005) 322.21] W. Trabelsi, E. Triki, L. Dhouibi, M.G.S. Ferreira, M.L. Zheludkevich,

M.F. Montemor, Surf. Coat. Technol. 200 (2006) 4240.22] D. Zhu, V. van Ooij, Electrochim. Acta 49 (2004) 1113.23] J.-B. Jorcin, M.E. Orazem, N. Pebere, B. Tribollet, Electrochim. Acta 51

(2006) 1473.24] G. Baril, N. Pebere, Corros. Sci. 43 (2001) 471.25] S.J. Xia, R. Yue, R.G. Rateick Jr., V.I. Birss, J. Electrochem. Soc. 151

(2004) B179.26] M. Garcia-Heras, A. Jimenez-Morales, B. Casal, J.C. Galvan, S. Radzki,

M.A. Villegas, J. Alloys Compd. 380 (2004) 219.27] S.A. Hayes, P. Yu, M.J. O’Keefe, T.J. O’Keefe, J.O. Stoffer, J. Electrochem.

Soc. 149 (2002) C623.28] P. Yu, S.A. Hayes, T.J. O’Keefe, M.J. O’Keefe, J.O. Stoffer, J. Electrochem.

Soc. 153 (2006) C74.29] P. Yu, T.J. O’Keefe, J. Electrochem. Soc. 153 (2006) C80.30] V. Palanivel, Y. Huang, W.J. van Ooij, Prog. Org. Coat. 53 (2005) 153.31] K.A. Yasakau, M.L. Zheludkevich, S.V. Lamaka, M.G.S. Ferreira, J. Phys.

Chem. B 110 (2006) 5515.