Corrosion behaviour of stressed

magnesium alloys

P.L. Bonora

a

, M. Andrei

a,*

, A. Eliezer

b

, E.M. Gutman

b

a

Department of Materials Engineering, Laboratory of Industrial Corrosion Control, University of Trento,

via Messiano 77, I-38050 Trento, Italy

b

Department of Materials Engineering, Ben-Gurion University of the Negev, Beer-Sheva, Israel

Received 3 July 2000; accepted 28 May 2001

Abstract

Potentiodynamic polarisation and impedance measurements are used to examine the cor-

rosion aspects of some Mg-based alloys, which were previously stressed in order to established

the e€ect of mechanical deformation on surface electrochemical reactions. A ®rst approach

was made for the unstressed alloys. The electrochemical tests were carried out in a sodium

borate bu€er solution. Ó 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Mechanochemical e€ect; Magnesium alloys; Potentiodynamic polarisation; Electrochemical

impedance

1. Introduction

The perception of magnesium as a rapidly corroding material has been a major

obstacle to its growth in structural applications despite its other desirable physical

properties. More importantly, the problem of stress corrosion is becoming a major

one today in Mg alloys.

New Mg alloys have been developed in recent years to meet the needs of structural

applications. Many cast alloys were tested for stress corrosion cracking and their

high sensitivity to stress corrosion was demonstrated [1,2]. Some magnesium and

magnesium alloys applications, especially in the transportation industry require

www.elsevier.com/locate/corsci

Corrosion Science 44 (2002) 729±749

*

Corresponding author. Tel.: [phone]; fax: [phone].

E-mail address: [e-mail] (M. Andrei).

0010-938X/02/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved.

PII: S0010-938X(01)00101-9

good fatigue properties [1]. Corrosion fatigue is the simultaneous action of corrosion

and alternating stress on an alloy, leading to a lower fatigue limit than in the case of

fatigue without any environmental impact.

In general, the aim of every stress corrosion study is to develop a basis for

quantitative prediction and qualitative diagnostic of the incidence of cracking and

failure under impact of environments and stresses. A more realistic way is to separate

stress corrosion with in¯uencing signi®cant phenomena and to study these phe-

nomena with the hope to develop a more general mechanism in future. Such im-

portant phenomena are mechanochemical e€ects (MCEs) and creep at crack tip [3].

2. Experimental method

In this work, AM50 and AZ91D magnesium alloys were studied. The most

common die casting alloy is AZ91D with 9% Al content. The combination of ¯u-

idity, strength and ductility of an alloy with 9% Al is undeniably one of the reasons

for which AZ91D alloys usage is widespread. Nevertheless, AZ91D has some dis-

advantages for use in a die casting. Firstly, although the material has good short-

term strength at elevated temperatures, the creep resistance is poor. This has been

attributed to the large amount of the presence of low melting eutectic (Mg

17

Al

12

).

Another disadvantage of higher Al content containing alloys is that ductility is usual

sacri®ced to grain strength [4]. But Mg alloys having lower Al content, for example

AM50 (5% Al) were found to be more ductile, especially during impact situation.

However, many structural applications require an appreciable amount of energy

absorption during their operation [3]. Thus, it is essential to study corrosion fatigue

resistance of Mg alloys and to investigate the correlation of corrosion fatigue with

the mechanochemical behaviour of the Mg alloys. The Mg alloys were submitted to

the standard mechanical tests before the electrochemical ones. The behaviour of

both stressed and non-stressed magnesium alloys was investigated by using poten-

tiodynamic polarisation measurements (DC polarisation) and electrochemical im-

pedance spectroscopy (EIS).

All the electrochemical measurements were performed in a tetraborate bu€er

solution that is a stable solution with a pH (ˆ9.7) at which Mg can cover itself with

more or less protective oxide or hydroxide which checks the dissolution reaction

[5]. A very used solution for the corrosion common studies of Mg alloys is

NaCl ‡ Mg(OH)

2

with a pH about 10.5. This solution does not have a stable pH

value in the electrode surface and thus, the corrosion conditions are not stable. In

our study the borate solution eliminate this unstable and the undesirable presence of

anions Cl that even in small amounts, usually break down the protective ®lm on

Mg [6,10]. The electrochemical testing was employed to study the main features of

the processes taking place at the alloy/solution interface. The e€ect of di€erent ap-

plied tensile stresses on the Mg-alloy corrosion resistance was studied. The corrosion

resistance of Mg alloys was pointed out by EIS measurements performed during the

free immersion time and under polarisation and the e€ect of the di€erent tensile

stresses was studied. The evolution of the electrode/electrolyte interface at di€erent

730

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

immersion times was also studied. The corrosion rate has been calculated by the

analysis of the potentiodynamicpolarisation that was carried out by the linear po-

larisation method.

The ¯at specimens of magnesium alloys (AM50 and AZ91D) used for this study

were prepared by the casting procedure (t

cast

ˆ 660°C and 650°C, respectively).

Chemical composition (wt.%) and mechanical properties are given in Tables 1 and 2.

The electrochemical tests using both DC (potentiodynamic polarisation) and AC

techniques (EIS) were carried out in an aerated 0.05 M sodium tetraborate solution

(pH ˆ 9:7). For all measurements a three electrode electrochemical cell was used,

with an Ag/AgCl as reference electrode and a platinum counter electrode. The

working electrode was prepared from the Mg-alloy samples after the application of

the tensile strength. Mg-alloy samples were embedded in an acrylic resin to provide

electrical isolation of the sample surface. The samples were air dried at room tem-

perature.

The potentiodynamiccurves were obtained using a PAR 273 potentiostat, with a

voltage scan rate of 0.2 mV/s. The impedance measurements were carried out using a

Solartron 1250 frequency response analyser coupled with the potentiostat. All the

experiments were controlled by a PC, which was also used for the acquisition,

storage and plotting of data. The scanned frequency ranged from 6 mHz to 100 kHz

and the perturbation amplitude was of 5 mV (it was observed that a variation of the

amplitude did not change the frequency response of the electrode/electrolyte inter-

face). The impedance measurements were performed at open circuit potential (E

OC

)

and also under potentiostaticconditions.

Table 1

Chemical analysis after die casting in wt.%

Mg alloy

Al

Mn

Zn

Si

Cu

Fe

Ni

Be

AM50

5.1

0.57

0

0.013

0.0007

0.0074

0.0006

0.0013

AZ91D

8.4

0.17

0.85

0.01

0.0008

0.0013

0.0007

0.0003

Table 2

Mechanical properties of samples

Mg alloy

TYS

a

(MPa)

UTS (MPa)

Elongation (%)

AM50

Without

Loading

AM50

120.3

125.4

0.30

AM50

129.9

147.6

1.94

AM50

131.2

152.1

2.13

AM50

129.9

151.7

2.26

AM50

135.6

166.0

3.56

AM50

118.7

160.4

4.39

AZ91D

Without

Loading

AZ91D

166.4

174.2

0.29

AZ91D

166.7

180.4

0.45

AZ91D

179.2

193.7

0.71

a

TYS is the tensile yield strength.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

731

All the measurements of this study were taken after stabilisation of the open

circuit potential, i.e. 20 min after immersion of the electrode in the solution. All the

potential values are reported with respect to the reference electrode (E

Ag=AgCl

ˆ 0:210

V versus ENH).

3. Results and discussion

3.1. Potentiodynamic polarisation measurements

The potentiodynamiccurves in 0.05 M sodium tetraborate solution (pH ˆ 9:7) of

unstressed Mg alloys and pure Mg are shown in Fig. 1. As a ®rst remark we note that

the shape of the polarisation curves is almost the same for the two alloys. In addi-

tion, a current plateau begins at 1:2 and 1:3 V for AM50 and AZ91D, respec-

tively, the current values being quite similar and relatively high. We can assume that

the alloy samples present a ``pseudo-passivation'', which di€ers with respect to pure

magnesium. We also note that by alloying the corrosion potential values become

much nobler and the anodic current densities reduce.

Figs. 2 and 3 show the potentiodynamiccurves obtained for both Mg alloys as a

function of the applied stress. The e€ect of mechanical deformation on the corrosion

behaviour can be seen. The corrosion parameters estimate the MCE, which means

the increasing of the anodic current densities and the disennobeling of the potential

values caused by the deformation.

The corrosion rate (i

corr

) was determined for small deviations from the corrosion

potential. The obtained values are plotted as a function of strain in the Fig. 4. The

graphs in Fig. 4 also contain the corrosion potential values. First of all, we remark

that both corrosion current and corrosion potential depend strongly on the applied

stress. So, we can note that i

corr

increases with the elongation up to a value of 1.94%

and 0.29% for AM50 and AZ91D, respectively. After that, the corrosion current is

Fig. 1. Potentiodynamiccurves of pure Mg and of AM50 and AZ91D alloys at v ˆ 0:2 mV/s.

732

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

reduced. Thus, the corrosion rate passes over a maximum in relation to the increase

in plasticdeformation, according to the general theory [2]. For the same values of the

strain for that a maximum value of the corrosion current was obtained, the corrosion

potential reaches the most negative values. A signi®cant MCE for both corrosion

rate and corrosion potential manifests itself in the magnesium alloys under study. It

can be also seen that a new slower increase of i

corr

is produced at 3.56% and after

0.45% of residual strain for AM50 and AZ91D, respectively.

Experimental con®rmation of the correlation between the mechanochemical be-

haviour and the strain hardening stages (intensive strain hardening and dynamic

recovery) could be see in Fig. 5, which regroups the stress±strain curves and the

corrosion rate variations for both magnesium alloys. The observed correlation could

be explained by the change of dislocation substructure during the plastic deforma-

tion as was already done in the case of other materials alloys [7]. Under intense strain

Fig. 2. Potentiodynamiccurves of stressed and unstressed AM50 alloy at v ˆ 0:2 mV/s.

Fig. 3. Potentiodynamiccurves of stressed and unstressed AZ91D alloy at v ˆ 0:2 mV/s.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

733

hardening, stresses grow and planar dislocation pile-ups appear. This leads to a

sharp increase in the MCE. At the ®nal stage of the dynamic recovery, planar pile-

ups are destroyed due to dislocation cross-slip and partial annihilation. This leads to

a decrease in the MCE value, which, thus, should pass over a maximum in the

process of plastic deformation. The MCE should grow intensively under plastic

deformation at the stage of strain hardening. In particular, the acceleration of metal

anodic dissolution is caused by a local reduction of equilibrium potential in the

vicinity of dislocations. This e€ect is much lower at the ®nal recovery stage. At this

stage, strain hardening is suppressed because of the development of dislocation

cross-slip processes.

In a previous stress corrosion study [1] a signi®cant mechanoelectrochemical e€ect

for corrosion rate, anodic current under constant potential and corrosion potential

for the di€erent Mg alloys in alkaline media (pH ˆ 10:5) was already observed. It

was also noted that for the AM20 and AM50 alloys the mechanical e€ect show a

maximum when the strain hardening stage transfers to the dynamicrecovery stage.

It is also important that the features of mechanical behaviour are the same in both

active and pseudo-passive states, independently of surface ®lm existence. Thus, the

analysis of the potentiodynamic curves was also carried out at the cross-sections of

Fig. 4. Corrosion current and corrosion potential for AM50 (a) and AZ91D (b) magnesium alloys as a

function of strain.

734

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

curves at di€erent selected potential levels to ®nd the corresponding current densi-

ties. The potential levels were chosen in both active and ``pseudo-passive'' ranges.

The obtained values are plotted in Fig. 6. For a better illustration of the MCE, Fig. 6

also regroups the potential values under constant current densities. All the values are

plot as a function of strain. Therefore, it can be seen that the anodic current in-

crement and the decrease of potential manifest itself in both active and passive states

in the case of AM50 alloy. For the AZ91D magnesium alloy, a slight decrease in

anodic current and a slow increment of potential are observed in both active and

passive states at the beginning of the applied stress, contrary to the corrosion current

and potential.

The two alloys under study contain the intermetallic compound Mg

17

Al

12

. A low

content of Al leads to a reduced amount of phase Mg

17

Al

12

[4].

It was observed that the b-phase is inert to the chloride solution in comparison to

the magnesium matrix and acts as a corrosion barrier depending on the manner that

the phase Mg

17

Al

12

is distributed in the alloy matrix [8]. So, the b-phase present in the

cast alloy is generally more resistant than the surrounding matrix alloy. Corrosion

resistance of the b-phase is related to its passive behaviour within a much wider pH

range than its pure components.

Fig. 5. Anodic current (under constant potential) in the active state and stress curves as a function of

strain for AM50 (a) and AZ91D (b) magnesium alloys.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

735

Corrosion rate measurements of binary alloys showed that Al was the only ele-

ment which caused a decrease in the corrosion rate of Mg; the corrosion rate de-

creased with increasing Al content [9].

Indeed, in our study realised in a free chloride media it was seen that in the non-

stressed conditions, the corrosion rate of the alloy with higher Al content is lower

than that for the alloy with less Al. This is largely due to the presence and distri-

bution of the b-phase (Fig. 7a) that better protects in the AZ91D alloy with a higher

content of Al and so with an elevated amount of b-phase. For AM50 with low Al

content, the fraction b-phase is small and is only formed as discrete islands in the

structure (Fig. 7b).

But hard secondary phase promotes strain hardening and thus, increases chemical

potential of atoms, i.e. they create the necessary conditions for mechanochemical

dissolution [2]. Consequently it is expected that most resistance to developing MCE

will be obtained for the alloys with highest Al contents. Indeed, deep pitting cor-

rosion occurs and the lifetime of the alloy decreases as the amount of b-phase in-

creases [10].

Our potentiodynamicstudy demonstrated that the AZ91D had high corrosion

rate in deformed state that AM50 under stress. In addition, some recent studies

Fig. 6. Anodic current densities (under constant potential) and potential values (under constant current)

in both active and pseudo-passivation states for AM50 (a) and AZ91D (b) magnesium alloys, as a function

of strain.

736

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

proved that the highest sensitivity to corrosion creep and corrosion fatigue [11] in a

corrosive environment is observed in the alloy with highest Al content. It was also

shown that the borate anions act as a corrosion inhibitors at the ®rst stage of creep

[10].

3.2. Electrochemical impedance spectroscopy measurements

A corroding metal is, as a ®rst approximation, modelled as a simple electro-

chemical system consisting of a double-layer capacitance, a solution resistance and a

charge transfer resistance. A system like this can be studied by using an AC signal,

which can provide more information than a DC polarisation. Thus, applying a 5 mV

sinusoidal potential through a potentiostatic circuit, the potential±current response

plots provide the impedance values.

Fig. 7. Typical microstructures of die casting AZ91D (a) and AM50 (b) alloys by optical microscope.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

737

The impedance diagrams are recorded at the initial time (t ˆ 0 h) immediately

after the stabilisation of the steady-state potential.

In the present study, EIS measurements were performed at open circuit and also

under polarisation. Thus, the typical Nyquist impedance plots showed in Fig. 8 are

obtained for unstressed Mg alloys in the tetraborate solution at free immersion

potential and under anodicand cathodicpolarisation.

Fig. 8. Nyquist plots at free immersion potential and under anodicand cathodicpolarisation for AM50

(a) and AZ91D (b) magnesium alloys.

738

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

The Nyquist plots of both magnesium alloys at open circuit exhibit two capacitive

loops, one for high and intermediate frequencies and the other, the small one, for low

frequencies. The ®rst capacitive loop is attributed at the charge transfer process.

Thus, for the frequencies higher than 1 Hz, a resistor R

p

and a capacitor C

dl

in

parallel can model the electrode/electrolyte interface. A partial data ®tting made with

the Boukamp circuit equivalent software [12] for the charge transfer process pro-

duced the R

p

(polarisation resistance) and C

dl

(double-layer capacitance) values.

The R

p

of the charge transfer process is 207.7 and 374 Xcm

2

for unstressed AM50

and AZ91D alloys, respectively. The obtained capacitance values are 22.6 and

68 lFcm

2

, for AM50 and AZ91D, respectively and they are associated at a double-

layer capacity, C

dl

, (50 lFcm

2

). The slightly lower value of C

dl

for the AM50 alloy

implies the formation of a thick, protective ®lm on the electrode surface; the much

lower C

dl

values being already reported for other Mg-based alloys [13]. The second

small capacitive loop is generally attributed at the masse transfer in the solid phase

[14], which consists of the oxide/hydroxide layers.

The EIS spectra (Fig. 8) obtained under anodic polarisation inside the potential

range of the MgO formation exhibit one capacitive loop followed by a linear part for

both magnesium alloys. As a ®rst remark we note the increase of the R

p

, which is

signi®cant in the case of AZ91D alloy. The increment of R

p

suggests the layer

growing on the electrode surface. The linear part of the Nyquist diagrams suggests

the di€usion process.

The impedance data obtained under anodic polarisation was simulated with the

Boukamp equivalent circuit software. Thus the values of R

p

are 236:8 Xcm

2

for

AM50 and 596:4 Xcm

2

for AZ91D, whereas the C

dl

values are 63.84 and 127.31

lFcm

2

for AM50 and AZ91D, respectively. The equivalent circuit consists of a

resistor (R

p

) in series with a constant phase element (CPE), the two being connected

with a capacitor (C

dl

) in parallel. The CPE can be assumed to be a Warburg di€usion

according to the n values close to 0.5. Thus, under anodic polarisation, the corrosion

process is controlled by the mass transfer of the corrosion products through the

oxide layers [15].

The Nyquist plots for both Mg alloys obtained under cathodic polarisation

present one loop capacitive (Fig. 8), which is attributed at water reduction.

The EIS data under polarisation show that both magnesium alloys have quite

similar behaviour in free immersion conditions and under anodic polarisation.

Moreover, for the AZ91D alloy, the charge transfer parameters obtained under

anodic polarisation in the active state indicate that the formation and growth of

corrosion layer products is more signi®cant than in the case of AM50 alloy.

The EIS measurements are further performed at the free immersion potential.

For the stressed alloys, the Nyquist diagrams (Fig. 9) obtained at the potential of

open circuit present a capacitive loop at high and intermediate frequencies as in the

case of unstressed magnesium alloys. In addition, at low frequencies a small ca-

pacitive loop was always observed for all the samples, this loop being more or less

reproducible. For a good clarity of the pictures in Fig. 9, the small capacitive loop

was not plotted. The EIS data for the ®rst capacitive loop associated at the transfer

charge process can be ®tted with the Boukamp circuit equivalent as in the case of

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

739

unstressed alloys, the equivalent circuit consisting of a R

p

and a C

dl

in parallel. The

obtained 1=R

p

(which is proportional to the corrosion rate) and C

dl

values are plotted

as a function of strain (Figs. 10 and 11), for both AM50 and AZ91D alloys.

Moreover, Fig. 10 regroups the corrosion current values determined on the poten-

tiodynamic polarisation curves (DC measurements). It can be seen that the maximal

1=R

p

value is obtained for a strain of 1.94% and of 0.29% for AM50 alloy and

Fig. 9. Nyquist plots at free immersion potential for stressed and unstressed AM50 (a) and AZ91D (b)

magnesium alloys.

740

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

AZ91D, respectively. Thus the 1=R

p

values obtained on the EIS measurements pass

over the maximum as the amount of increase in the plastic deformation, in the same

way as the corrosion current determined on the DC polarisation. Moreover, the C

dl

values plotted as a function of strain (Fig. 11) also pass over a maximum. Fig. 11

also plots the stress±strain curves for both magnesium alloys, in order to emphasise

the correlation between the strain hardening stages and the electrochemical

parameters. The strain values for the maximum of C

dl

are of 1.94% and 0.29% for

AM50 and AZ91D, respectively. The corrosion rate, proportionally with 1=R

p

, and

Fig. 10. Corrosion rate (1=R

p

) from EIS measurements and corrosion current from potentiodynamic

polarisation as a function of strain for AM50 (a) and AZ91D (b) magnesium alloys.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

741

also the C

dl

values manifest a signi®cant MCE. These results show that the EIS

measurement could be a rapid and ecient test to highlight the MCE.

The electrode/electrolyte interface behaviour as a function of immersion time was

also studied. The Nyquist and Bode diagrams obtained at di€erent immersion times

are plotted in Figs. 12, 13, 15 and 16 for unstressed alloys and also for di€erent levels

of stress for magnesium alloys. For both magnesium alloys (stressed and unstressed),

the capacitive loop attributed to the charge transfer process generally increases with

time, whereas the capacitive tail of the low frequencies does not show a uniform

evolution.

For unstressed AM50 alloy (Fig. 12), for a relatively short immersion time (1 and

3 h), the electrode/electrolyte interface does not undergo important modi®cations.

For a time of over 24 h the polarisation resistance increases and so the surface

protection is increased because of the formation of corrosion products layers. In

Fig. 11. Double-layer capacity from EIS measurements as a function of the strain and the stress±strain

curves for AM50 (a) and AZ91D (b) magnesium alloys.

742

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

Fig. 12. Nyquist (a,b) and Bode-phase (c) diagrams for the unstressed AM50 alloy at di€erent immersion

times.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

743

Fig. 13. Nyquist (a,b) and Bode-phase (c) diagrams for the AM50 alloy at strain of 2.26% at di€erent

immersion times.

744

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

addition, the double-layer capacity increases as can be seen in Fig. 14(a), which plots

the R

p

and C

dl

values obtained with the Boukamp ®tting equivalent circuit. For a

time of under 96 h the corrosion process can be ®tted with only one time constant,

which corresponds at the charge transfer. For t P 96 h the Bode-phase diagram

presents two time constants and the equivalent circuit for the process ®tting changes.

The same behaviour was practically observed for all the stressed AM50-alloy sam-

ples that can be seen, for example, in Fig. 13. This behaviour suggests that after an

immersion time, which di€ers from a sample to another (but falls between 24 and 96

h), the corrosion mechanism changes; the charge transfer occurs through the cor-

rosion products layers, which can also be suggested by the drastic decrease in the C

dl

values. The equivalent circuit, which ®ts the impedance results, consists of two

parallel RC circuits connected in series. The ®rst equivalent circuit corresponds to

the oxide/hydroxide magnesium ®lm, which becomes quite protective so that the

Fig. 14. Charge transfer resistance (R

p

) and double-layer capacity (C

dl

) as a function of immersion time for

AM50 alloy at strain of (a) 0%; (b) 1.94%; (c) 2.13%; (d) 2.26%; (e) 3.56% and (f) 4.39%.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

745

charge transfer process ®tted with the second equivalent circuit reaches a high

polarisation resistance (Fig. 14).

For the AZ91D alloy (Figs. 15 and 16) the evolution of the electrolyte/electrode

interface is not similar. No change of corrosion mechanism is produced even for

longer immersion times. The values of R

p

and C

dl

of the corrosion process obtained

with a Boukamp partial ®tting are plotted in Fig. 17 for the AZ91D-alloy samples

under study. Generally, both corrosion parameters increase with time, which

suggests the formation of corrosion products layers.

A suitable equivalent circuit relevant to the lower frequencies is hardly found due

to the irregular behaviour of the Mg-alloy samples.

Fig. 15. Nyquist (a) and Bode-phase (b) diagrams for unstressed AZ91D alloy at di€erent immersion

times.

746

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

4. Conclusion

The plastic deformation e€ects on both potentiodyanmic polarisation curves and

impedance diagrams have been studied in the case of two magnesium-based alloys

(AM50 and AZ91D, respectively).

For both magnesium alloys, it was obtained that the anodiccurrent density de-

termined on the potentiodynamic curves passes over a maximum as a function of the

level of the plasticdeformation (as predicted in the theory). It was also shown that

(a)

(b)

Fig. 16. Nyquist (a) and Bode-phase (b) diagrams for the AZ91D alloy at strain of 0.29% at di€erent

immersion times.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

747

the disennobeling of potential manifests itself in both magnesium alloys. The cor-

relation between the mechanochemical behaviour and strain hardening stages (in-

tensive strain hardening and dynamicrecovery) was also related to the known

change of dislocation substructure during the plastic deformation.

It was also shown that the AZ91D had a high corrosion rate in deformed state

that AM50 under stress, while in the non-loading state, the corrosion rate was found

to be higher for the AM50 alloy. This behaviour con®rms the MCE theory and also

the alloys behaviour at creep.

In addition, the inverse of resistance polarisation …1=R

p

† and the double-layer

capacitance (C

dl

) determined on the EIS spectra pass also over the maximum as the

amount of increase of the plastic deformation for both Mg alloys. The obtained

results indicate that the EIS measurement could be a suitable and ecient exper-

imental test in order to evidence the MCE.

Thus experimental con®rmation of the correlation between the MCE and the

strain hardening stages by both electrochemical techniques (DC polarisation and

EIS, respectively) was provided. Only DC polarisation was generally used in order to

de®ne the mechanochemical properties of di€erent alloys that will be employed

under stress corrosion conditions.

References

[1] A. Eliezer, E.M. Gutman, E. Abramov, E. Aghion, Mater. Sci. Forum 289±292 (1998) 517.

[2] E.M. Gutman, Mechanochemistry of Solid Surfaces, World Scienti®c Publishing, 1994, pp. 1±322.

Fig. 17. Charge transfer resistance (R

p

) and double-layer capacity (C

dl

) as a function of immersion time for

AZ91D alloy at strain of (a) 0%; (b) 0.29%; (c) 0.45% and (d) 0.71%.

748

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

[3] A. Eliezer, E.M. Gutman, E. Abramov, Ya. Unigosky, G. Agiv, E. Aghion, in: E. Aghion, D. Eliezer

(Eds.), Magnesium 2000, Proceedings of the Second Israeli International Conference on Magnesium

Science and Technology, 2000, p. 356.

[4] C.F. Baker, New alloys and Shot Delivery System Developments, in: B.L. Mordike, F. Hehmann

(Eds.), Magnesium Alloys and their Applications, Informationsgesellschaft, 1992, pp. 77±87.

[5] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, Cebelcor, Brussesls, 1974,

p. 141.

[6] Metal Handbook, ninth ed., vol. 13, ASM, Metal Parks, Ohio, 1979, p. 742.

[7] E.M. Gutman, G. Solovio€, D. Eliezer, Corros. Sci. 38 (7) (1996) 1141.

[8] O. Lunder, J.E. Lein, T.Kr. Aune, K. Nisancioglu, Corrosion 45 (1989) 741.

[9] G.L. Makar, J. Kruger, J. Electrochem. Soc. 137 (2) (1990) 414.

[10] E.M. Gutman, A. Eliezer, Ya. Unigovski, E. Abramov, Mechanoelectrochemical behavior and creep

corrosion of magnesium alloys, Material Science Engineering A, vol. A302, 2001, p. 63.

[11] A. Eliezer, E.M. Gutman, E. Abramov, Y. Unigovski, E. Aghion, Corrosion fatigue and corrosion

creep of magnesium alloys, Proceedings of the International Congress: Magnesium 2000, Magnesium

Alloys and their Applications, 27±28 September, 2000, Munich, Germany, p. 498.

[12] Boukamp, Equivalent Circuit Software, Users Manual, The Netherlands University of Twente, 1988,

pp. 6±26.

[13] R. Udhayan, D.P. Prokash Bhatt, J. Power Sources 63 (1996) 103.

[14] S. Turgoose, R.A. Cottis, The impedance response of ®lm-covered metals, in: Scully, Silverman,

Kendig (Eds.), Electrochemical Impedance: Analysis and Interpretation, 1993, pp. 173±191.

[15] C.H. (Raymond) Tsai, Analysis of EIS data for common corrosion processes, in: Scully, Silverman,

Kendig (Eds.), Electrochemical Impedance: Analysis and Interpretation, 1993, pp. 37±52.

P.L. Bonora et al. / Corrosion Science 44 (2002) 729±749

749