Corrosion Behavior of 55mass%Al–1.6mass%Si–Zn Alloy in Wet

1. Introduction

The good resistance of zinc and aluminum against corro-sion and their relatively low price is the reason for theirwidespread use as a coating on steel.1) Zinc and aluminumprotect steels through original barrier layer action of thecoating, secondary barrier action of corrosion productslayer, and galvanic action of coating layer with the exposedparts of underlying steel.2) Aluminum coating provides bet-ter protection than zinc coating since aluminum itself carry-ing protective oxide is attacked very slowly.3) However,under certain mild conditions the attack on aluminum is tooslow to provide cathodic protection to steel and in such cir-cumstances zinc coating is preferred.4) In view of the ad-vantages and also the short-comings of individual zinc andaluminum coatings a new concept of combining the bestproperties of these traditional coatings has been developedby Bethlehem Steel, USA during the 1970s, better knownas Al–Zn alloy coated steel (GL) with the composition55mass%Al–1.6mass%Si–Zn.4)

GL has shown a superior anticorrosion performance thanany other zinc or Al–Zn alloy coated steel in field exposuretests.4) However, there are limited laboratory investigationsusing accelerated test methods, which give quick informa-tion about the performance of coating in the desired envi-ronment. To further widen the knowledge about the corro-sion performance of GL coated steel in marine environ-ment, in this study, wet–dry cyclic test, which simulatesmarine atmospheric environment, has been used to evaluate

the corrosion behavior of 55mass%Al–1.6mass%Si–Znalloy.

2. Experimental

55mass%Al–1.6mass%Si–Zn alloy (GL alloy) samplewas prepared from aluminum (99.99%), silicon (99.9%)and zinc (99.5%). They were melted at 883 K and cooledby pouring on a stainless steel plate so as to obtain similarstructure and chemical composition to that on the GL coat-ing on steel. The purpose of experimenting with bulk alloysample instead of coated steel was to avoid the necessity ofrepeated preparation of the cell as the bulk sample could bepolished and used repeatedly. For the monitoring of ac impedance and corrosion potential, a two-electrode cellarrangement with the exposed area of 10 mm2 each, asshown in Fig. 1, was used as a corrosion-monitoring probe.In this electrode system, one alloy plate was served as the working electrode and other as the counter electrode. The specimen was cut into small plate of dimension13�13�2 mm3 and cleaned by sonicating in acetone for120 s. One side of the plate was connected to the lead wireusing silver paste. The connection of the plate and wire wascovered by quick drying epoxy to make sure that the con-nection was secured. Two plates, separated apart by 0.1 mmwith the help of a polyester sheet, were then mounted in anambient temperature solidifying epoxy resin in such a waythat exposed surface area was 10�10 mm2. Before embed-ding the plates into the epoxy resin, they were subjected to

ISIJ International, Vol. 44 (2004), No. 10, pp. 1727–1732

1727 © 2004 ISIJ

Corrosion Behavior of 55mass%Al–1.6mass%Si–Zn Alloy inWet–Dry Cyclic Environment

Amar Prasad YADAV, Atsushi NISHIKATA and Tooru TSURU

Department of Metallurgy and Ceramics Science, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8552 Japan.

(Received on November 18, 2003; accepted in final form on July 21, 2004 )

The corrosion of 55mass%Al–1.6mass%Si–Zn (GL) alloy has been tested under a wet–dry cyclic conditionusing an AC impedance technique to clarify its corrosion behavior. The results have been compared withthat of a zinc sample exposed under identical condition. The sample was exposed to alternate conditions ofimmersion for 1 h in a 0.05 kmol/m3 NaCl solution and drying for 7 h at 298 K and 60% RH, and the corrosionrate was monitored for 10 d (30 cycles) using a two-electrode type probe. Simultaneously, the corrosion po-tential was continuously monitored using a special arrangement in combination with a Luggin probe tip.From the monitoring results of the corrosion rate and corrosion potential and the polarization curve, it wasfound that the corrosion rate of GL was decreased by inhibition of the cathodic reaction.

The impedance response was evaluated by using an equivalent circuit considering the charge and masstransfer consequences of the coupled corrosion processes. It was found from the time constant of the diffu-sion impedance that the corrosion product layer formed on GL alloy acted more effectively as a diffusion bar-rier against oxygen transport compared to zinc.

KEY WORDS: atmospheric corrosion; wet–dry cyclic test; galvanized steel; AC impedance.

electro-deposition coating at 170 V employing an organicresin for 180 s, which was followed by curing at 443 K for1.2 ks. The aim was to minimize the possibility of occur-rence of crevice corrosion at the sides of the specimen. Theresults of GL alloy were compared with that of a zinc sam-ple (99.5%), which was exposed together with GL alloy.

Two holes (1.7 mm diameter), as shown in Fig. 1, weredrilled at the middle of the epoxy resin to contain the saltbridge, which acted as the effective Luggin probe tip duringthe dry cycles for the Ag/AgCl, KCl reference electrode.This helped achievement of the continuous monitoring ofcorrosion potential. Prior to exposure, the electrode surfacewas mechanically polished with wet SiC papers up to grade1 000 and then ultrasonically cleaned in distilled water.Afterwards, a peripheral bank of 0.5 mm thick polyestersheet was glued on it surface. This served to control theamount of solution, which remained on the electrode sur-face at the on set of each dry cycle. A three-electrode cellconfiguration was employed for polarization measurementwith an Ag/AgCl, KCl (sat) reference electrode (SSE) andPt sheet as a counter electrode.

The corrosion-monitoring probe (sample) was set insidea temperature and humidity controlled chamber with itsmetal surface facing upwards and a water leveler was usedto ensure that the cell was placed horizontally. The wet–drycycle was conducted by exposing the probes to an alternatecondition of immersion for 1 h in a 0.05 kmol/m3 NaCl so-lution and drying for 7 h at 298 K and 60% RH. The test so-lution was introduced into the vessel from an external stor-age tank through the inflow magnetic valve programmed todeliver enough volume to submerge the test specimen atleast 20 mm under the solution for 1 h. It was then drainedthrough the outflow magnetic valve and the cell was left todry for 7 h. The cyclic test was done for 10 d comprising30 cycles. The solution was prepared from doubly distilledwater and analytical grade NaCl. Duplicate samples wereexposed in all sets of experiments to check the repro-ducibility.

A frequency response analyzer (FRA, Schlumberger SI1 280) with a multiplexer controlled by a personal computerwas used to continuously monitor the impedance. High fre-quency limit of FRA was set at 10 kHz (ZH) and low fre-quency limit was set at 10 mHz (ZL). The impedances were

simultaneously measured at these frequencies, and the po-larization resistance, Rp, was determined by subtracting ZH

from the ZL, i.e.,

Rp�ZL�ZH..................................(1)

This is based on the assumption that the high frequency im-pedance gives the solution resistance and the low frequencyimpedance gives a sum of solution resistance and polariza-tion resistance. Each single value of Rp was obtained at thetime interval of 900 s.

It is well known that the corrosion current density icorr isproportional to the value of 1/Rp.

5,6)

icorr�k/Rp ...................................(2)

The proportional constant k is theoretically calculated fromthe Tafel slopes of anodic and cathodic polarization curves.Due to difficulty in getting distinct Tafel regions in this cor-rosion system, it has been assumed that the k value does notchange during the progress of corrosion. By making this as-sumption, the obtained R p

�1 was used as the index of corro-sion rate.

In order to know the contribution of kinetic and diffusionprocesses on the corrosion of GL alloy, the whole frequen-cy impedance characteristics of corroded samples weremeasured in the range of 10 mHz to 10 kHz at 10 mV per-turbing signal using a frequency response analyzer.

3. Results and Discussion

3.1. Corrosion Monitoring under Wet–Dry Cycles

The results of the corrosion monitoring and the corrosionpotential variation of GL alloy under immersion for 1 h in a0.05 kmol/m3 NaCl solution and drying for 7 h at 298 K and60%RH are shown in Figs. 2(a) and 2(b), respectively. Forthe comparison, monitoring results of zinc under similarcondition are shown in Fig. 3. The Rp

�1 is taken as the indexof corrosion rate. The corrosion rate, Rp

�1, of GL alloy be-came almost constant with the progress of wet–dry cycles.Initially the surface was almost dried up during the dryingperiods, however, after 9th cycles (72 h) the surface couldnot dry due to accumulation of corrosion products, whichretain moisture. The corrosion products were thin and com-pact and the corrosion attack was homogeneous, as com-pared to thick and porous corrosion products and heteroge-neous corrosion attack on zinc or galvanized steel.7) Thehigh corrosion rate of zinc led to the formation of volumi-nous corrosion products, which enhanced heterogeneousdeposition of salt and hence the heterogeneous corrosion incontrast to GL alloy. From Fig. 2(a) and Fig. 3(a), it can beseen that the corrosion rate, Rp

�1, of GL alloy was muchlower than zinc.

To better understand the corrosion process in a singlewet–dry cycle, plots of Rp

�1 and Ecorr of GL alloy and Znwithin the 15th wet–dry cycle against time is made in Figs.4(a) and 4(b), respectively. Each cycle can be divided intothree different periods, immersion, drying and dried-up, asshown in Fig. 4, where the dried-up state was defined in thisstudy as time period when the solution conductance ob-tained from high frequency impedance was lower than1�10�3 W�1. The variation of Rp

�1 of both samples revealsthat the corrosion rates increased sharply just after the

ISIJ International, Vol. 44 (2004), No. 10

© 2004 ISIJ 1728

Fig. 1. Schematic diagram of two electrode cell used for moni-toring of impedance and corrosion potential underwet–dry cyclic condition (a) top view and (b) transversecross sectional view.

probe was submerged in the test solution. In the drying pe-riod, GL alloy (Fig. 4(a)) showed only small increase ofRp

�1 as compared to a large increase for zinc (Fig. 4(b)).

This suggested that the corrosion process of GL alloy wasdifferent from that of zinc under a thin layer of electrolyte.The reason for this difference in the corrosion behavior ofGL alloy and zinc under thin electrolyte layer will be dis-cussed later.

The corrosion reaction of Zn consists of anodic dissolu-tion of zinc (Zn→Zn2��2e�), though Al may dissolve si-multaneously in case of GL alloy, and cathodic reduction ofmainly oxygen (O2�2H2O�4e�→4OH�) in a neutral chlo-ride solution. According to a mixed potential theory, a bal-ance of these reactions of anodic and cathodic determinesthe corrosion rate (current) and corrosion potential. As canbe seen in Fig. 4(a), the corrosion potential of GL alloy be-came most negative in the last stage of the drying period,the corrosion rate Rp

�1 being decreased (Fig. 4(a)).According to a mixed potential theory, this indicated thatthe oxygen reduction was greatly inhibited in the progressof drying; nevertheless the electrolyte layer became thinner.This may be attributed to accumulation of a protective cor-rosion product layer during the drying. On the other hand,in the case of zinc, the corrosion potential during the dryingprocess did not shift to negative value and maintained eithera constant value or slightly shifted to positive value and thecorrosion rate Rp

�1 was gradually increased in the dryingperiod (Fig. 4(b)). This implies that enhancement of oxygenreduction by thinning of the electrolyte layer. Thus, the cor-rosion products accumulated on the zinc surface during thedrying did not inhibit the oxygen reduction process.

To have a better understanding on the corrosion process,an optical photograph of GL alloy, observed normal to thesheet surface, was taken and is shown in Fig. 5. A cored


1729 © 2004 ISIJ

Fig. 2. Monitoring results of (a) the reciprocal of polarization re-sistance, Rp

�1, and (b) the corrosion potential, Ecorr, of GLalloy exposed under wet–dry cyclic condition consistingof 1 h-immersion in 0.05 kmol/m3 NaCl solution of 7 h-drying at 25°C and 60%RH.

Fig. 3. Monitoring results of (a) the reciprocal of polarization re-sistance, Rp

�1, and (b) the corrosion potential, Ecorr, ofzinc exposed under wet–dry cyclic condition consistingof 1 h-immersion in 0.05 kmol/m3 NaCl solution and 7 h-drying at 25°C and 60%RH.

Fig. 4. Variation of corrosion rate, Rp�1 and corrosion potential,

Ecorr, of (a) GL alloy and (b) zinc in a single wet–drycycle. The immersion state, the drying up state and driedup state are indicated. The probe surface was assumed tobe in dried up state when the solution conductance ob-tained from high frequency impedance was lower that1�10�3 W�1.

dendrite aluminum-rich matrix can be seen with zinc-richinterdendritic alloy filling in the interstices. The results ofcorrosion monitoring and corrosion potential variation sug-gested that the zinc-rich inter-dendritic phase of the alloycorroded preferentially. According to Zoccola et al.,4) thezinc corrosion product gradually fills up the interdendriticinterstices and the alloy is transformed into a composite ofan aluminum-rich matrix and zinc corrosion products me-chanically keyed into the interdendritic labyrinth. The lowcorrosion rate of GL alloy as compared to zinc or galva-nized coating4) was because of the relatively small area ofexposed zinc and that aluminum rich dendrite was coveredwith aluminum oxide. With the progress of corrosion nobreakdown of aluminum oxide film occurred as can bejudged from the low corrosion rate. However, as is evidentfrom Fig. 4(a) pitting of the aluminum oxide film seemed tohave occurred during the drying process.

The corrosion products were identified using an X-raydiffractometer at the end of wet–dry cyclic test and the X-ray pattern is shown in Fig. 6 together with zinc sampleunder identical condition. Since the amount of corrosionproducts of GL alloy was very limited, it was not possibleto use powder sample. The corrosion products identifiedfrom X-ray peaks were ZnO and ZnCl2.4Zn(OH)2 both inGL alloy and zinc though there were large number of peaksin zinc sample. There were no peaks for Al2O3. However,

the low corrosion rate suggested the presence of Al2O3 inthe aluminum-rich dendrite.

3.2. Polarization Behavior

The polarization behavior of the as-polished as well ascorroded GL alloy and zinc was investigated in0.05 kmol/m3 NaCl at a scan rate of 0.5 mV/s. The current-potential response is shown in Figs. 7(a) and 7(b), respec-tively. The cathodic current density due to oxygen reductionon GL alloy was approximately 5 mA cm�2 at �1.1 V, whichwas more or less 5 times lower than those on zinc (Fig.7(b)) and galvanized coating.3,5) Accordingly, the improve-ment of corrosion resistance of zinc coating by the additionof aluminum could be attributed to the inhibition of theoxygen reduction. The cathodic side of the polarizationcurve, Fig. 7(a), showed a peak or a shoulder at around�1.20 to �1.25 V (vs. SSE). This range of potential is veryclose to that obtained by deferential capacitance measure-ments8) for the formation/removal of a zinc oxide/hydroxidefilm. In the later stage of cathodic polarization (�1.5 V vs.SSE), reduction of H2O started with the progressive in-crease in the current density.

Contrary to the inhibition of cathodic reaction, the anod-ic current was found to increase abruptly after the potentialwas made slightly positive. To know the reason, a laser mi-crophotograph of the specimen surface was observed beforeand after the anodic polarization and is shown in Fig. 8. Itis evident that a selective dissolution of zinc from interden-dritic phase has occurred, which account for the sharp in-crease of anodic current. The anodic Tafel slope had a typi-cal value observed among galvanized coatings.8,9)

3.3. Impedance Characteristics

Typical impedance spectra measured at different expo-


© 2004 ISIJ 1730

Fig. 5. Optical photograph of GL alloy, used in this study, show-ing dendritic structure. The interdendritic phase consistsof zinc.

Fig. 6. X-ray diffraction pattern of corrosion products of (a) GLalloy and (b) zinc after 30th cycle exposure of samples.The peaks are denoted by s, o, a, z and f forZnCl2· 4Zn(OH)2, ZnO, Al, Zn and substrate, respective-ly.

Fig. 7. Polarization curves of (a) GL alloy and (b) zinc in aerated0.05 kmol/m3 NaCl solution at scan rate of 0.5 mV/s. Thesamples were removed from the wet–dry cyclic test afterdifferent exposure time.

sure time for GL alloy and zinc are shown in Figs. 9(a) and9(b), respectively. The points in the plots indicate the exper-imental value and the lines are the results of curve fittingusing the equivalent circuit shown in Fig. 10. The main fea-tures of these Nyquist diagrams of GL alloy involved a ca-pacitive semicircle in the high frequency region followed inthe low frequency part by an approximately linear behaviorwith a slope close to unity corresponding to a phase angleof 45° to the real axis. At short exposure time (1 cycle), thehigh frequency locus was dominated by an apparent resis-tive-capacitive system that contained information on thecharge transfer kinetics. The capacitive contributionmerged as a slight distorted loop as a consequence of sur-face inhomogeneities. The impedance spectra recorded atincreasing exposure time (30 cycles) exhibited the samegeneral characteristic features, but the magnitude of theoverall impedance gradually decreased. The kinetics of in-terfacial reaction became faster (decreasing Rct) with theprogress of corrosion. The shape of the impedance dia-grams at low frequencies was clearly related to the relax-ation of a mass transport process that can be attributed tothe oxygen reduction reaction.

In the case of zinc, the Nyquist plots were characterizedby two depressed capacitive loops expanding towards lowerfrequencies. The capacitive loops were fairly well resolvedto obtain the individual time constants. However, with theprogress of corrosion (after 9th cycle exposure), the twotime constants were no longer separable and a single timeconstant obtained.

From the results of impedance measurements, the inter-

face of GL alloy and zinc can be best described in terms ofan equivalent circuit shown in Fig. 10, where CPE and Rct

represent the capacitance and charge transfer resistance, re-spectively; ZW represents the infinite length Warburg im-pedance10) in the case of GL alloy and finite length Warburgimpedance in the case of zinc; and RW is a combination ofsolution and film resistances.

The constant phase element (CPE), which was used in-stead of capacitance (Cdl) due to depressed semicircle, canbe given by the expression:

Z(CPE)�[TC( jw)PC]�1 .........................(3)

where j is the imaginary number ( j2��1), TC is a constantrepresenting capacitance, w is the angular frequency, andPC�aC/(P/2), and aC is the phase angle of the CPE.11) Thefactor PC is adjustable parameter that usually lies between


1731 © 2004 ISIJ

Fig. 8. Laser photographs of GL alloy surface (a) before anodicpolarization and (b) after anodic polarization. The selec-tive dissolution of zinc from zinc-rich interdendriticphase can be seen in (b).

Fig. 9. Nyquist plots of (a) GL alloy and (b) zinc obtained in be-tween 10 kHz to 10 mHz frequency range at different ex-posure times under wet–dry cyclic condition. The pointsare the experimental values while lines are the results of fitting using equivalent circuit shown in Fig. 10. Theparameters used for fitting of Fig. 9(a) are:RW�0.042 kW cm2, Rct�4.0 kW cm2, Cdl�26 mF/cm2, andZW�10.2 kW cm2 for 1st cycle and RW�0.046 W cm2,Rct=2.5 kW cm2, Cdl�73 mF/cm2, and ZW�5.0 kW cm2 for30th cycle. Similarly, for Fig. 9(b): RW�0.048 W cm2,Rct�0.38 kW cm2, Cdl�200 mF/cm2, and ZW�0.35 kW cm2

for 1st cycle and RW�0.075 W cm2, Rct�0.95 kW cm2, Cdl

1 000 mF cm2, and ZW�1.10 kW cm2 for 30th cycle.

Fig. 10. Equivalent circuit used for fitting the impedance resultsof Fig. 9. In the case of GL alloy diffusion impedance,ZW, represents infinite length diffusion, while in the caseof zinc, it represents finite length diffusion.

0.5 and 1. The value of PC has been found to be almost con-stant at the value of 0.72.

The Warburg impedance, ZW is given by

ZW�R*tanh([ jwTD]PD)( jwTD)�PD ................(4)

The effective diffusion resistance R* is obtained at w → 0limit and the parameter TD is a time constant, which de-notes TD�(L2/D), L and D being the effective diffusionthickness and effective diffusion coefficient, respectively.PD�aD/(P /2), and aD is the phase angle of the ZW. Fromthe polarization curves of both samples in Fig. 7, the ZW

seems to be a diffusion impedance due to oxygen reductionin the cathodic process. The TD value for GL alloy wasmuch larger than that for zinc, if compared in the samecycle. This indicates that a corrosion product layer formedon GL alloy acts more effectively as a diffusion barrieragainst oxygen transport.

4. Conclusions

From the present study, the following conclusions weremade on the corrosion behavior of GL alloy in wet–drycyclic marine environment:

(1) The results of corrosion monitoring and corrosionpotential variation in wet–dry cyclic marine environmenthave shown that the corrosion process of GL alloy was dif-ferent from that of pure zinc during drying process. Thecorrosion potential of GL alloy shifted to less noble valueduring the drying periods in contrast to shift in the noble di-rection in the case of zinc.

(2) The corrosion rate of GL alloy was approximatelymuch lower than that of zinc. The low corrosion rate of GLalloy as compared to zinc was due to the inhibition of thecathodic process.

(3) The corrosion attack on GL alloy was more or lesshomogeneous with the formation of thin and compact layerof corrosion products in contrast to heterogeneous corro-sion on zinc. With the progress of corrosion, the kinetics ofinterfacial reaction became faster and the overall imped-ance also decreased.

Acknowledgement

The authors thank Nippon Steel Corporation for supply-ing the 55%Al–1.6%Si–Zn alloy sample. This work waspartially supported by The Iron and Steel Institute of Japan.

REFERENCES

1) X. G. Zhang: Corrosion and Electrochemistry of Zinc, Plenum Press,New York, (1996), 241.

2) A. P. Yadav, A. Nishikata and T. Tsuru: Corros. Sci., 46 (2004), 169.3) H. Katayama, Y. C. Tay, A. S. Viloria, A. Nishikata and T. Tsuru:

Mater. Trans. JIM, 38 (1998), 1089.4) J. C. Zoccola, H. E. Townsend, A. R. Borzillo and J. B. Horton:

ASTM STP, 646 (1978), 165.5) A. P. Yadav: Ph. D. Dissertation, Tokyo Institute of Technology,

(2003).6) M. Stern and A. L. Geary: J. Electrochem. Soc., 104 (1957), 56.7) A. P. Yadav, A. Nishikata and T. Tsuru: Corros. Sci., 46 (2004), 361.8) L. M. Baugh: Electrochim. Acta, 24 (1979), 657.9) W. Walter: Corros. Sci., 32 (1991), 1331.

10) A. Nishikata: Denki Kagaku, 64 (1996), 266.11) B. A. Boukamamp: Solid State Ionics, 20 (1980), 31.


© 2004 ISIJ 1732

Documents

Corrosion Behavior of 55mass%Al–1.6mass%Si–Zn Alloy in Wet