Double-layered manganese phosphate conversion coating on magnesium alloy AZ91D: Insights into coating formation, growth and corrosion resistance

Surface & Coatings Technology 217 (2013) 147–155

Contents lists available at SciVerse ScienceDirect

Surface & Coatings Technology

j ourna l homepage: www.e lsev ie r .com/ locate /sur fcoat

Double-layered manganese phosphate conversion coating on magnesium alloyAZ91D: Insights into coating formation, growth and corrosion resistance

Xiao-Bo Chen a,b,⁎, Xian Zhou b, Trevor B. Abbott a,c, Mark A. Easton a, Nick Birbilis a,b

a CAST Co-operative Research Centre, Monash University, VIC 3800, Australiab ARC Centre of Excellence for Design in Light Metals, Department of Materials Engineering, Monash University, VIC 3800, Australiac Magontec Limited, Sydney, NSW 2000, Australia

⁎ Corresponding author at: CAST Co-operative ResearVIC 3800, Australia. Tel.: +61 3 9905 9297.

E-mail address: [email protected] (X.-B. Che

0257-8972/$ – see front matter © 2012 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.surfcoat.2012.12.005

a b s t r a c t

Article history:Received 29 September 2012Accepted in revised form 5 December 2012Available online 14 December 2012

Keywords:Magnesium alloy AZ91DCorrosion resistanceConversion coatingManganese phosphatePolarisationEIS

A double-layered conversion coating system, consisting of magnesium hydroxide–magnesium/manganesephosphate, was applied to magnesium alloy AZ91D using an acidic manganese nitrate and ammoniumdihydrogen phosphate solution. The coating structure, composition and morphology were characterised bySEM, EDX, XRD and XPS. A coating formation mechanism is proposed, and the effect of operating parameters,i.e. pH and temperature, on coating formation was systematically investigated, with optimised conditionsable to produce coatings of high corrosion resistance. Corrosion resistance of the coating was evaluated byelectrochemical and salt spray testing. The double-layered coating system develops in three stages: initialsubstrate dissolution, formation of a dense magnesium hydroxide layer, and then co-deposition of magne-sium and manganese phosphate film.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Magnesium (Mg) and its alloys are promising structural and elec-tronic materials for diverse applications, including transportation and3C components, due to their light weight, specific strength and con-ductivity [1,2]. The development and use of Mg alloys in industryremain in infancy due to poor corrosion resistance upon exposure toair and humid conditions. Two primary methodologies are nowadaysbeing extensively explored to improve the resistance to corrosion ofMg-alloys; these include bulk alloying and the use of barrier coatingson the Mg surface [3,4]. The latter method can decrease corrosion rateof Mg alloys by a few orders of magnitude, which is more efficient inthe short term compared to alloy development [3,4].

Increasingly extensive research is being devoted to protectivecoatings for Mg [3,4]. Chemical conversion coatings stand out fromother coating types that include anodising, electroplating, electrolessplating, ion implantation, etc., owing to low cost and efficiency [4,5].In general, no power or specific facilities are required to carry out con-version coating process, significantly reducing production cost. In addi-tion to imparting corrosion resistance, conversion coatings could alsobe an adhesive base for subsequent plating or painting [6,7].

Since being implemented by DOW (DOW Chemical Co., USA),chromate conversion coating, has been the benchmark corrosion pro-tection scheme, which also offers remarkable self-healing capability.

ch Centre, Monash University,

n).

rights reserved.

Modern concerns and legislation regarding toxicity present the needfor alternatives to the use of chromate, and research has been carriedout upon: fluoride [8], phosphate [5,9,10], phosphate–permanganate[11,12], stannate [13–15], rare earth (RE) [16–18], vanadate [19,20],titanate [7,21] and ionic liquid (IL) conversion coatings [22,23].Among these techniques, vanadate also presents an environmentalhazard; the long processing time (up to 60 min) of stannate and REcoating procedures cause significant cost increase; ILs are costly,and concentrated (and toxic) HF is indispensable for fluoride coating.

Phosphate conversion coatings on the other hand, are more environ-mentally friendly and have been successfully exploited to protect steeland zinc against corrosion in dilute aqueous solutions. Metal-phosphatelayers provide not only corrosion protection but also some specific func-tions to substrates, i.e., a coating bath with the necessary phosphateanions and metallic cations, including Ca2+, Mn2+/3+/4+/7+, Zn2+ etc.,may improve corrosion/wear resistance and sliding properties [6,24–26].Further, biocompatibility and bioactivity of calciumphosphate (Ca–PO4),high wear resistance and adhesion strength of zinc phosphate (Zn–PO4),and lubricity ofmanganese phosphate (Mn–PO4), have also been dem-onstrated. In the past decades, the applications of Ca–PO4 [5,10,27,28]and Zn–PO4 [29–31] conversion coatings have been successfully ex-panded to light-weight Mg alloys.

In contrast, very few studies have focused on protecting Mg alloyswith Mn2+–PO4 conversion coatings, in spite of being expected to bemore stable and corrosion resistant than their Zn and Ca peers. Han etal. [32] and Zhou et al. [9,33] reported a manganese hydrogen phos-phate (MnHPO4) conversion coating for AZ31D and AZ91D alloysand proposed a mechanism of the coating formation. Cui et al. also

148 X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

investigated the growing process of an MnHPO4 conversion coatingon AZ31, but presented an alternative coating growth mechanism[34]. Though these existing reports indicate that the MnHPO4 conver-sion coatings had desirable corrosion resistance, the immersion/coatingtime, up to 30 min, is much longer than the basic requirement forindustrial applications.

The performance of the phosphate conversion coatings is depen-dent on the crystal structure as well as the morphology. For example,a microcrystalline structure is usually optimal for corrosion resistanceor subsequent painting. A coarse grain structure impregnated withoil, however, may be the most desirable for wear resistance. Theseproperties can be tailored by selecting the appropriate phosphatesolution, using various additives, and controlling bath temperature,pH, ion concentration, and phosphating time [3,4]. As a result, thisprompts further questions regarding the experimental conditions,such as bath temperature and pH, amenable for the synthesis of stablelow-valence Mn–PO4 coatings on Mg alloys and a study of their corro-sion resistance.

In this study, we apply an Mn2+–PO4 conversion coating ontodie-cast Mg alloy AZ91D with the view of achieving significantimprovement in corrosion resistance and exploiting a simple andlow cost strategy. The systematic study is aimed at developments to-wards a practical replacement for DOW chromate coatings. Followingpreliminary trials, the coating process was simplified to involve onedipping-step in an acidic Mn–PO4 solution, eliminating conventionalpretreatment steps [5,33,35], which is important for upscaling. Theeffect of the essential coating parameters over a certain range, i.e.,pH (2.0 to 6.0) and temperature (ambient to 80 °C), on the corrosionresistance of the resultant coatings was investigated, with a viewto understanding the nature of the Mn–PO4 conversion coating onAZ91D, and in producing more uniform protective films.

2. Experimental section

2.1. Mn–PO4 conversion coating preparation

ASTMMg alloy AZ91D, from theMg alloy supplier HNKWE (China)was used in this study. The AZ91D contained 9.1 wt.% Al, 0.8 wt.% Zn,0.24 wt.% Mn, 0.031 wt.% Si, 0.0023 wt.% Fe, 0.015 wt.% Cu, and0.0005 wt.% Ni, balance Mg, as measured by inductively coupledplasma-atomic emission spectroscopy (ICP-AES, Spectrometer Services,Coburg, VIC, Australia). A Toshiba 250 ton clamping force cold chamberhigh pressure die casting (HPDC) machine was used to cast test platesof 70×60×2 mm in size. Smaller test samples with dimensions of20 mm×20 mm×2 mmwere cut from these plates and were groundto 1200 grit finish and used as substratematerial. Prior to the one-stepcoating treatment, AZ91D samples were ultrasonicated in acetoneat room temperature (RT) for 15 min and then in absolute ethanolfor 10 min and rinsed with deionised water thereafter. The coatingsolution contained 0.01 M manganese nitrate (Mn(NO3)2), 0.01 Mammonium dihydrogen phosphate (NH4H2PO4). Nitric acid (HNO3)or ammonia (NH3·H2O) was utilised to adjust pH down or up (2.0–6.0). All chemicals used in this work were analytical grade fromSigma-Aldrich. The coating process was thus conducted by immersingpre-cleaned AZ91D specimens into the coating solution varying fromRT to 80 °C for 5 min. Following this, treated samples were dried inair and kept in a desiccator for further characterisation.

2.2. Thermodynamic calculations

The software MEDUSA (which is an acronym for Make EquilibriumDiagrams Using Sophisticated Algorithms) was used to make a firstorder calculation of the complexes and phases in a given chemical sys-tem via equilibrium formation constants.MEDUSA nominally calculatesat room temperature, however manual alteration of all parameters ispossible at the users' discretion using the programme edit function.

Fig. 3b illustrates how the predominance of Mg2+ in the presence ofphosphate (10 mM) and Mn2+ (10 mM) varies with pH and concen-tration of Mg2+ ions, while Fig. 3c presents how the predominance ofphosphate in the presence of Mg2+ (500 mM) and Mn2+ (10 mM)varies with pH and concentration of phosphate ions.

2.3. Microstructural analysis of Mn–PO4 conversion coatings

Surface morphology of coated AZ91D alloys was observed usingscanning electronmicroscopy (SEM, FEI Nova Nano) fitted with energydispersive X-ray spectroscopy (EDXS). Structure and phase composi-tion of the AZ91D surfaces before and after coating were identifiedby X-ray diffraction (XRD, Philips PW1140) using Cu-Kα radiation(λ=1.5418 Å, 40 kV, 25 mA) and a scanning speed of 0.01°/min at a2θ range of 20–50°. Surface chemistry was analysed by X-ray photo-electron spectroscopy (XPS, Thermo K-alpha, UK) with a hemispheri-cal analyser and the core level XPS spectra for Mn2p, Mg2p, Al2p,Zn2p, P2p, O1s and C1s were recorded. The measured binding energyvalues were calibrated by the C1s (hydrocarbon C–C, C–H) of 285 eV.The photoelectrons were generated by Al-Kα (1486.6 eV) primary ra-diation (20 kV, 15 mA).

2.4. Corrosion evaluation

For electrochemical tests, a flat-cell (PAR, K0235) containing300 mL of 0.1 M NaCl electrolyte was used with an exposed workingelectrode area of 1 cm2, a saturated calomel (SCE) reference electrodeand Pt-mesh counter electrode. Electrochemical experiments wereperformed on a BioLogic® VMP-3Z potentiostat/galvanostat usingEC-Lab 10.2 software. Potentiodynamic polarisation experimentswere carried out at a sweep rate of 1 mV/s after 10 min of open circuit(OCP) conditioning. The polarisation curves were used to determinecorrosion current density, icorr, via a Tafel-type fit using EC-Lab soft-ware. As a general rule, fits were executed by selecting a portion ofthe curve that commenced>50 mV from corrosion potential; Ecorrand icorr were subsequently estimated from the value where the fitintercepted the potential value of the true Ecorr. More generally,polarisation testing was also able to visually reveal comparative infor-mation related to the kinetics of anodic and cathodic reactions betweenalloys and was in contrast to salt spray testing. Electrochemical im-pedance spectroscopy (EIS) was conducted over a frequency range of100 kHz to 10 mHz, with a sinusoidal amplitude of 10 mV and fivepoints per decade after 10 min of open circuit (OCP) conditioning. TheEIS data was analysed using EC-Lab 10.2 software. At least three sepa-rate scanswere performed for each data point to ensure reproducibility.The long-term corrosion resistance of various treated AZ91Dwas exam-ined by salt spray testing up to 48 h according to the ASTM B117 stan-dard. The samples were placed at a tilting angle of 30° in a chambercontaining 5 wt.% NaCl fog at 35 °C.

3. Results and discussion

3.1. OCP evolution during formation of Mn–PO4 basedconversion coatings

Observation of OCP evolution for AZ91D in the Mn–PO4 bath asa function of immersion time, pH and temperature, is presented inFig. 1, illustrating the development of the conversion coatings [7,10,36].The OCP curves were evidently affected by bath temperature (Fig. 1a).When a constant pH value of 4.0was applied, all theOCP curves obtainedat elevated temperatures (40–80 °C) initially demonstrate an abruptpotential shift in the more noble direction (first 30 s), revealing theinstant protective coating formed on the substrate upon exposure tothe bath [7,10]. A rapid drop in OCP at the initial stage can be observedexclusively on the case treated at RT, demonstrating removal of thespontaneously air-formed looseMgO/MgOH/MgCO3 film and subsequent

Fig. 1. Evolution of OCP for AZ91D immersed in anMn–PO4 bath at various temperatures(a) and pH (b). The effect of the growth temperature and pH on the coating evolution isevident.

149X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

exposure of the AZ91Dmetallic substrate [7,10]. No such sudden falls inOCP being observed on other samples can be attributed to the elevatedtemperatures (40–80 °C), which stimulated the coating formation andmoderated the dissolution time duration of the oxide layer. When thedynamic equilibrium of film dissolution–precipitation was established,OCP approached a plateau. The value of the plateau potential is a func-tion of bath temperature (Fig. 1a). Higher bath temperature produced acoating with more positive OCP and correspondingly, higher corrosionresistance [7,10].

It is evident that the coating evolution, i.e., dissolution–precipitation,was determined by bath pH value (Fig. 1b). When temperature waskept at 80 °C, the high bath-acidity (pH 2.0) led to a continuous poten-tial shift to the more noble region, which indicates ongoing precip-tation on substrate [10,36]. On the contrary, the potential reached aplateau after immersed for 300 s in themildly acidic bath (pH 4.0), in-dicating the establishment of an equilibrium between dissolution andprecipitation [10,36]. A peak potential can be seen in the approximatelyneutral bath (pH 6.0) and followed by a drop, which is likely due to thelocal dissolution of the existing coating [7]. In this study, the optimalimmersion time of 5 min was chosen for all cases according to theOCP graph. The coating formation process and their corrosion resis-tance will be discussed in following sections.

3.2. Microstructural characterisation of the Mn–PO4 conversion coatings

Phosphatation is an endothermic reaction [37], thus, coating pro-cedures conducted at low temperature do not provide sufficient energy

to satisfactorily generate phosphates. Under such circumstances,the phosphating rate is either slow or does not take place at all. Conse-quently, long immersion times, up to a few days, may be required togive rise to a complete phosphate film—with extremely thin coatingsbeing undesirable for corrosion resistance. In contrast, high processingtemperature can offer enough activation energy and eventually accel-erate the phosphating rate. A thick phosphate film can be achieved in arelatively short time and render corrosion resistance to Mg substrates.

Observation of surface morphology of bare and treated AZ91D canreveal the effect of operating parameters, i.e., pH and temperature,on the formation of the Mn–PO4 coating, as presented in Fig. 2.After removal of the surface layer containing die-spray and oxide bymechanical grinding, bare AZ91D presents a silvery metallic finishand a microstructural feature of abraded grooves. The inset EDXdata demonstrates the existence of Mg (~89.3 wt.%), Al (~9.9 wt.%)and Zn (~0.8 wt.%), which is in agreement with the ICP-AES results.Deposition of Mn–PO4 particles and subsequent coatings was signifi-cantly influenced by growth temperature when pH was maintainedat 4.0. A few newly deposited particles appeared on the substratetreated at RT for 5 min, while the density and size of the white roundparticles scattering over the surface was progressively increased whenbath temperature was elevated from 40 to 60 °C, which is attributedto the high reaction rate incurred by high temperature. The AZ91Dtreated at 70 °C displays a different surface morphology, where thinflakes and round clusters were distributed unevenly over surface.With higher treating temperature, the densely packed small particleshad much more energy to merge into large grains and produce a highcoverage. A compact but thin coating was produced at 80 °C, consistingof Mn (2.9 at.%), P (4.8 at.%) and O (39.0 at.%), in addition to Mg(46.7 at.%) and Al (6.1 at.%), as depicted by the inset EDX spectrum(Please refer to Table S1 in the Supporting information for all the EDXelemental analysis results). The high fraction of Mg and Al detected byEDX indicates that the conversion coating was thin (~1.5 μm, Figs. 3and S1) and the strong Mg and Al signals were mainly from the under-neath AZ91D. Some cracks were obvious on the sample treated at 80 °Cwhose surface exhibits a uniformly light-yellow colour, which is bene-ficial formany applications and has a less effect on the colour of the sub-sequent paints [7].

Bath pH was another critical factor that significantly influencedcoating morphology and thickness, particularly when coating growthwas conducted at 80 °C. Descending pH from 4.0 to 2.0 resulted in acoating with higher Mn (~4.4 at.%) and P (~7.3 at.%) content butevident network structure, similar to the one reported by Zhou et al.by immersing AZ31 in anMnHPO4 bath at 60 °C for 30 min. The thick-ness was also enhanced by the strong acidity to ~2.8 μm (Fig. S1).Correlating this with the OCP curve of the case of pH 2–80 °C, it canbe concluded that high acidity accelerated phosphating rate andthickened the coatings. The characteristic network structure of Mn–PO4

based conversion coatings can be correlated to hydrogen evolution pro-cess rather than dehydration effects [14]. Corrosivemediumwill readilypenetrate this thick barrier coating and approach to substrate throughthe defect sites in the network structure and eventually deteriorate cor-rosion resistance [3,4]. Thus, defects should be avoided for the sakeof corrosion resistance. Increasing pH from 4.0 to 6.0 at 80 °C, on thecontrary, induced a rough though thin (~0.8 μm, Fig. S1) coating witha lower Mn (1.1 at.%) and P (1.9 at.%) fraction. This can be attributedto the low acidity that inhibited the dissolution of Mg substrate andconsequently moderated the nuclei formation rate of Mn–PO4 coating.

Cross-sectional micrograph and corresponding XRD spectrum(Fig. 3a) of the conversion coated AZ91D at pH 4–80 °C illustratethat the coating system (~1.5 μm thick) consisted of an interlayerand a top film. Peaks detected by XRD were ascribed to substrateAZ91D, Mg(OH)2, and (Mg/Mn)3(PO4)2, respectively. The noticeabledetachment of the top layer from the intermediate film may be attrib-uted to the sample preparation procedure, including grinding andpolishing. Though not included in this study, the coating adhesion

Fig. 2. SEM micrographs of bare and various coated AZ91D samples revealing the effect of bath pH and temperature on the formation of Mn–PO4, and EDX (inset) of the bare andcoated AZ91D at pH 4–80 °C presenting the deposition of Mn and PO4 ions (scale bar 20 μm).

150 X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

strength and hardness will be further characterised in the future workto clarify this issue.

To further investigate the chemical state of the coating elements,XPS survey scans, high-resolution elemental analysis, and depth pro-filing were conducted on AZ91D treated for 5 min at pH 4–80 °C asdepicted in Fig. 4. The survey scan displays the existence of Mg, O,P, and Mn at the surface. O atomic concentration reduces (62.8 to7.8 at.%) and the Mg content increases (26.9 to 61.9 at.%) along thedepth of the coating. Meanwhile, Mn and P signals were detectedmainly from the outer region of the coating. Mn and P concentrationsdescend gradually from 3.4 and 6.6 to 0.9 and 2.4 at.%, respectively.The high-resolution scans further identify the presence of innerMg(OH)2, and outer (Mg/Mn)3(PO4)2 films.

3.3. Formation mechanism of the double-layered coating systemon AZ91D

Based on the abovementioned results, a formation mechanismof the double-layered coating system is proposed and schematicallydepicted in Fig. 5. In this study, the evolution of the double-layeredcoating system on AZ91D includes three primary phases, i.e., sub-strate dissolution, intermediate Mg(OH)2 film deposition, and final(Mg/Mn)3(PO4)2 co-precipitation. Once Mg components were placedin the prepared acidic Mn–PO4 bath, two classic dissolution reactions(Eqs. (1) and (2)) took place, which locally generated massive OH−

and Mg2+ ions, raising the pH in the vicinity of the metal surface(Fig. 6) [5,10,27,28]. The excess of the depleted Mg2+ and OH−

ions, compared to the small fraction of Mn2+ and PO43− in the bath,

would precipitate immediately in the form of Mg(OH)2 on metal sur-face (Eq. (3)) as an intermediate film, as demonstrated in Figs. 3b and4, and schematically depicted in Fig. 5. Such Mg(OH)2 intermediatelayer, especially the one produced at high temperature, say 80 °C, isa corrosion inhibitor [11,38] and led OCP suddenly ascending tomore passive direction as displayed in Fig. 1.

MgOþ 2Hþ→Mg2þ þH2O ð1Þ

Mgþ 2H2O→Mg2þ þ 2OH− þ 2H2↑ ð2Þ

Mg2þ þ 2OH−→Mg OHð Þ2↓ ð3Þ

4Mn2þ þ 3H2PO−4 →MnHPO4 þMn3 PO4ð Þ2 þ 5Hþ ð4Þ

As Zhou et al. proposed [9], in a Mn(H2PO4)2 solution, there isa balance between all ingredients at a certain temperature and pH,as described in Eq. (4). The continuous consumption of H+ ionsduring H2 evolution decomposed the hydrolyzing equilibrium forthe Mn(H2PO4)2 (Eq. (5)) that was forced to move towards right di-rection, resulting in insoluble Mn3PO4 conversion coating formation

Fig. 3. Cross-sectional micrograph and XRD spectrum of the AZ91D treated at pH 4–80 °C in 0.01 M Mn(NO3)2–NH4H2PO4 solution, which suggests the existence of a Mg(OH)2–(Mg/Mn)3(PO4)2 double layer system (a) and thermo-equilibrium predominance area diagram for the Mn2+, Mg2+ and PO4

3− ions calculated using the MEDUSA software package(b and c). It can be seen that Mg(OH)2 (exists as intermediate layer) is the dominate deposition when bath pH is high, and then Mg3(PO4)2 and Mn3(PO4)2 precipitate orderly alongwith pH decrease (outer layer).

151X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

on AZ91D surface. Thus, after immediate formation of an Mg(OH)2intermediate layer, insoluble Mn3(PO4)2 nuclei were producedon top of the Mg(OH)2 film and expanded to cover the entire surfaceafter 5 min. Due to the existence of Mg2+ ions, co-precipitation of(Mg/Mn)3(PO4)2 was available, as revealed in Fig. 3c. Similarco-precipitation of (Fe/Mn)3(PO4)2 has been noted on steel treatedin Mn–PO4 bath [6]. As the phosphating process proceeded, AZ91Dsubstrate dissolution was moderated by limited protection and theamount of Mg2+ ions in the solution gradually decreased, resulting ina declined concentration profile of Mg2+ and ascending concentrationprofile of Mn2+ in the phosphate layer. The XPS depth profile in Fig. 4demonstrates that the outer layer of theMn–PO4 phosphate depositionincludes ions from the coating bath while inner layer contained higherfraction of the ions from the AZ91D substrate [24]. It should be pointedout that Zhou et al. reported that Mn–PO4 nuclei directly grew on sub-strate rather than on an Mg(OH)2 intermediate layer, which may be

ascribed to the etching treatment prior to immersion and highMn con-centration (0.10–0.25 M) applied in the bath [9].

3.4. Corrosion resistance of conversion coated AZ91D alloy

Immersing Mg coupons into acidic phosphate baths resulted in acrystalline product that drastically reduces the electrical conductivityof the surface and plays as a barrier to isolate corrosive electrolytefrom the substrate. Since corrosion relies on the flow of electronsbetween anodes and cathodes that exist on a heterogeneous surfacelike AZ91D, decrease in the electrical conductivity, and separationbetween metal and electrolyte will significantly retard the corrosionprocess [39]. In this study, corrosion resistance of various AZ91Dspecimens was evaluated via potentiodynamic polarisation curvesand EIS in 0.1 M NaCl, and salt spray according to the ASTM B117standard.

Fig. 4. XPS analysis of the Mg–PO4 conversion coating, survey, high resolution of O1s, Mg2p, Mn2p, and P2p, and depth profile of the coating formed after 5 min of immersion atpH 4–80 °C, revealing the existence of elements of O, Mg, P, and Mn, and their distribution over the coating system.

152 X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

Because analysis of potentiodynamic curves is not always in agree-ment with the ‘long-term’ corrosion rate measured by salt spray [7],the analysis of various AZ91D samples was used exclusively herein todemonstrate the instant anodic and cathodic reaction mechanisms, aspresented in Fig. 7a (variable temperature) and b (variable pH). Withrespect to the effect of temperature, at a constant pH (4.0), it can be

Fig. 5. Schematic presentation of the forma

seen in Fig. 7a that the cathodic reactions were tremendously inhibitedby the conversion coating treatment, which has also been reported byother researchers [30,40], though the reason is still unclear. It may beattributed to the formation of intermediate insoluble Mg(OH)2 whichacts as an inhibitor of oxygen and consequent cathodic kinetics, similarto that of chromate coating on Al alloys as Kendig and Buchheit

tion of the Mn–PO4 coating on AZ91D.

Fig. 6. The pH change of the coated AZ91 alloy during the immersion process at variousconditions.

153X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

presented [41]. It is evident that in the anodic branch even above Ecorr,the corrosion current density still remained almost the same indicatingthe passive nature of these layers [1,42]. A breakdown potential pointand passivation region can be seen in all the anodic branches, revealingthat the anodic dissolution reaction was also retarded [28]. In contrast,corrosion current density (icorr), an indicator of corrosion resistance,was drastically regulated by bath temperature, same as the OCP results(Fig. 1a). Specifically, icorr of AZ91D was reduced by an order of magni-tude by the coating formed at 80 °C, as illustrated in Fig. 7a and Table 1.In terms of the effect of pH, both higher (pH 2.0) and lower acidity(pH 6.0) were a detrimental factor on the corrosion resistance whentreated at 80 °C. It is obvious that the corrosion current density in theanodic branch rose steeply immediately above the Ecorr revealing thepit formation on the AZ91D treated at pH 2.0. The least corrosion resis-tance of the samples treated at pH 2.0 may be attributed to the net-work structure. The existence of incomplete double-layered coatingsystem incurred in the bath with pH 6.0 could not fulfil a satisfactoryinhibiting role in corrosion either.

The corrosion resistance of MnPO4 coated AZ91D was furtherexamined using EIS (Fig. 7c and d). An example of a calculatedcurve is presented to illustrate the goodness of fit (Fig. S2). EIS canprovide instantaneous data on the impedance of a surface subject topolarisation, which is directly proportional to the corrosion resistance(i.e., inversely proportional to corrosion rate) [43,44]. The Nyquistplots of bare AZ91D coupon exhibit a capacitive loop at high frequen-cies, a capacitive loop at intermediate frequencies, and an inductivearc at low frequencies. The capacitive loop at high and mediumfrequency can be ascribed to the characteristic of an electric doublelayer and loose naturally formed oxide film on the substrate, respec-tively [27]. Meanwhile, the presence of the inductive arc suggests theoccurrence of relaxation of absorbed species, which is generally relat-ed to the exposure of the AZ91D and the subsequent release of Mg2+

ions and Mg(OH)2 [45]. Only one capacitive loop can be observed inthe EIS plots of all the coated AZ91D, indicating the passive natureof the coatings. The increase in diameter of the capacitive loop corre-lates to the increase in the surface film corrosion resistance [46].The EIS data were simulated using the equivalent circuit presentedin Fig. 7f (EC-Lab package, version 10.2). The parameters Rs (the solu-tion resistance), Q (constant phase element representing a non-idealcapacitance related to the coating), Rc (the coating resistance), Qdl

(constant phase element representing a non-ideal capacitance relatedto the double layer), Rct (the charge transfer resistance) and W(representing the Warburg element which represents the diffusionof species) were calculated. This equivalent circuit was chosen dueto being the most simple representation of a filmed surface that can

accommodate local breakdown/defects [7,29,44]. Based on the fittedresults in Table 1, it is observed be found that the low frequencyimpedance (determined at 10 mHz) and the total resistance (Rt) ofthe coatings were increasing as a function of bath temperature,and pH 4.0 was favourable to give rise to the most protective film.Accordingly, the AZ91D coated at pH 4.0 and 80 °C had the largestpolarisation resistance, indicating that the coating system was effec-tive as a barrier to protect AZ91D alloy against corrosion in 0.1 MNaCl.

After 48 h salt spray, the area fraction corroded, which is simplythe ratio of corroded area versus the total area exposed to NaCl fog,was measured and listed in Table 1. Compared to bare AZ91D, thecorrosion area fraction was reduced by the proposed Mn–PO4 conver-sion coating treatment. The coatings obtained at pH 4-(RT~50 °C)offered a mild protection, suffering 5–10% corrosion after salt spraytest for 48 h. Improved protection is observed on coatings at pH 4(60–70 °C), where only 1–5% area corroded. The minimum corrosionarea fraction (less than 1%) was noted on the case of pH 4–80 °C. Theoptimal Mn–PO4 conversion coating thus can offer a corrosion protec-tion similar to that of the DOWconversion coating and other conversioncoatings with high protection efficiency [7,47]. Moreover, high temper-ature, though has less remarkable influence on the improvement of cor-rosion rate than that of pH, always favours the anticorrosion performanceof the phosphate coatings [30].

4. Conclusions

A series of Mn–PO4 containing conversion coatings were producedby a simple immersion method capable of improving the corrosionresistance of Mg alloy AZ91D. A coating formation mechanism wasproposed with the aid of thermodynamic equilibrium calculations.Upon exposure to the coating bath, the matrix Mg dissolved and re-leased, Mg2+, H2 gas and OH− ions, increasing the pH in the vicinityof solid–liquid interface. The pH increase facilitated the formation ofthin Mg(OH)2 intermediate layer on the substrate, which was thenverified by SEM and XPS analysis. Finally, a top coating in form of(Mg/Mn)3(PO4)2 was produced due to their decreasing solubilityalong with decreasing pH, as predicted from the thermodynamicequilibrium diagram and confirmed by XPS results. Calculations couldprovide a theoretical rationalisation to engineering protective coatingformation for AZ91D.

The effect of coating growth parameters, pH and temperature,on coating morphology and subsequent corrosion resistance wassystematically studied. It was found that pH was a key factor to deter-mining coating thickness and characteristics. A thicker (~2.8 μm)surface film obtained at high acidity (pH 2.0) was associated with anetwork of severe structural defects due to hydrogen evolution.Because cracks are detrimental to corrosion resistance, a mild acidity(pH 4.0) resulted in a more compact Mn–PO4 conversion coating withoptimal corrosion resistance. Higher alkalinity (pH 6.0), however,retarded the release of Mg2+ ions from the substrate, which led to acoating with low corrosion resistance. The operating temperaturealso had an influence on the coating formation, albeit only minor topH. Due to its exothermic characteristic, the phosphating reactionrate heavily depends on the energy offered by the heating process. Ingeneral, the higher temperature, the quicker the phosphating process.Overall, the instant growth of a dense and thin Mg(OH)2 intermediatelayer and a complete coverage by an (Mg/Mn)3(PO4)2 top layer atpH 4–80 °C exhibited the best corrosion resistance in corrosive envi-ronments, displaying a corrosion area fraction less than 1% after 48 hof salt spray testing. With respect to the salt spray evaluation, theproposed Mn–PO4 surface film outperformed chromate conversioncoatings.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.surfcoat.2012.12.005.

Fig. 7. Potentiodynamic polarisation curves of various AZ91D obtained in 0.1 M NaCl at a sweep rate of 1 mV/s (a and b), Nyquist plots of the various AZ91D obtained in 0.1 M NaClwith a frequency range from 100 kHz to 10 mHz (c and d), e) 3D plot of the corrosion current density icorr (z axis) against treating temperature (x axis) and pH (y axis), and f) theequivalent circuit used for analysis of the EIS data of various AZ91D. (Rs is solution resistance, Rc is the resistance and constant phase element of a film on the sample surface, and Rct,Q and Qdl are the charge transfer resistance of a coating, constant phase element of the charge transfer, charge transfer capacitance of a double-layer, respectively. W is Warburgelement accounting the diffusion of species).

Table 1The corrosion area fraction after 48 h of the salt spray test, corrosion current density (icorr) and film resistance (Rc, generated from the Nyquist plots in Fig. 7c and d using the equiv-alent circuit presented in Fig. 7f) of various AZ91D.

Coating parameters pH 4-RT pH 4–40 °C pH 4–50 °C pH 4–60 °C pH 4–70 °C pH 4–80 °C pH 2–80 °C pH 6–80 °C Bare AZ91D

Corrosion area fraction, % 5–10 5–10 5–10 1–5 1–5 b1 15–30 15–25 70–80icorr, 10−6 A cm−2 4.5±0.5 4.1±0.6 3.6±1.1 2.7±0.6 2.3±0.2 1.1±0.2 11.5±0.8 11.9±0.2 18.0±0.8Rc, ohm 2950±360 3470±170 4110±320 4410±170 4870±110 6110±80 7580±140 1790±180 3640±210

154 X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

155X.-B. Chen et al. / Surface & Coatings Technology 217 (2013) 147–155

Acknowledgement

The CAST CRC was established under, and is funded in part by theAustralian Commonwealth Government Cooperative Research CentreScheme. The authors acknowledge the facilities (SEM-EDX and XPS),and the scientific and technical assistance, of the AustralianMicroscopy&Microanalysis Research Facility at the RMITMicroscopy &Microanal-ysis Facility. The Australian Research Council (Centre of Excellence forDesign in Light Metals) and Victorian State Government for the estab-lishment of the Victorian Facility for Light Metals Surface are gratefullyacknowledged.

References

[1] W.-J. Xu, J.-L. Song, J. Sun, Y. Lu, Z.-Y. Yu, ACS Appl. Mater. Interfaces 3 (2011)4404.

[2] B.L. Mordike, T. Ebert, Mater. Sci. Eng., A 302 (2001) 37.[3] J.E. Gray, B. Luan, J. Alloys Compd. 336 (2002) 88.[4] X.-B. Chen, N. Birbilis, T.B. Abbott, Corrosion 67 (2011) 035005.[5] X.-B. Chen, T. Abbott, N. Birbilis, Corros. Sci. 55 (2012) 226.[6] C.-M. Wang, H.-C. Liau, W.-T. Tsai, Surf. Coat. Technol. 201 (2006) 2994.[7] Y.C. Yang, C.Y. Tsai, Y.H. Huang, C.S. Lin, J. Electrochem. Soc. 159 (2012) C226.[8] K.Y. Chiu, M.H. Wong, F.T. Cheng, H.C. Man, Surf. Coat. Technol. 202 (2007) 590.[9] W.Q. Zhou, D.Y. Shan, E.H. Han, W. Ke, Corros. Sci. 50 (2008) 329.

[10] Y.W. Song, D.Y. Shan, E.H. Han, Corros. Sci. 51 (2009) 62.[11] C.S. Lin, C.Y. Lee, W.C. Li, Y.S. Chen, G.N. Fang, J. Electrochem. Soc. 153 (2006) B90.[12] M. Mosiałek, G. Mordarski, P. Nowak, W. Simka, G. Nawrat, M. Hanke, R.P. Socha, J.

Michalska, Surf. Coat. Technol. 206 (2011) 51.[13] H. Huo, Y. Li, F. Wang, Corros. Sci. 46 (2004) 1467.[14] F. Zucchi, A. Frignani, V. Grassi, G. Trabanelli, C. Monticelli, Corros. Sci. 49 (2007)

4542.[15] H.H. Elsentriecy, K. Azumi, H. Konno, Electrochim. Acta 53 (2007) 1006.[16] C. Wang, S.L. Zhu, F. Jiang, F.H. Wang, Corros. Sci. 51 (2009) 2916.[17] C.S. Lin, S.K. Fang, J. Electrochem. Soc. 152 (2005) B54.[18] K. Brunelli, M. Dabalà, I. Calliari, M. Magrini, Corros. Sci. 47 (2005) 989.[19] K.H. Yang, M.D. Ger, W.H. Hwu, Y. Sung, Y.C. Liu, Mater. Chem. Phys. 101 (2007)

480.

[20] A.S. Hamdy, I. Doench, H. Möhwal, Electrochim. Acta 56 (2011) 2493.[21] T. Ishizaki, Y. Masuda, N. Saito, Electrochem. Solid-State Lett. 12 (2009) D68.[22] N. Birbilis, P.C. Howlett, D.R. MacFarlane, M. Forsyth, Surf. Coat. Technol. 201

(2007) 4496.[23] P.C. Howlett, S. Gramet, J. Lin, J. Efthimiadis, X.-B. Chen, N. Birbilis, M. Forsyth,

Science China Chem. 55 (2012) 1598.[24] J. Flis, J. Mańkowski, T. Zakroczymski, T. Bell, Corros. Sci. 43 (2001) 1711.[25] Y. Totik, Surf. Coat. Technol. 200 (2006) 2711.[26] D. Hawke, D.L. Albright, Met. Finish. 93 (1995) 34.[27] Y.W. Song, D.Y. Shan, R.S. Chen, F. Zhang, E.H. Han, Surf. Coat. Technol. 203 (2009)

1107.[28] X.B. Chen, N. Birbilis, T.B. Abbott, Corros. Sci. 53 (2011) 2263.[29] R. Amini, A.A. Sarabi, Appl. Surf. Sci. 257 (2011) 7134.[30] Y.L. Cheng, H.L. Wu, Z.H. Chen, H.M. Wang, L.L. Li, Trans. Nonferrous Met. Soc.

China 16 (2006) 1086.[31] L.Y. Niu, G.Y. Li, Z.H. Jiang, L.P. Sun, D. Han, J.S. Lian, Trans. Nonferrous Met. Soc.

China 16 (2006) 567.[32] E.H. Han, W.Q. Zhou, D.Y. Shan, W. Ke, Mater. Sci. Forum 419–422 (2003) 879.[33] W.Q. Zhou, W. Tang, Q. Zhao, S.W. Wu, E.H. Han, Mater. Sci. Forum 686 (2011)

176.[34] X.-J. Cui, J.-X. Zhou, X.-Z. Lin, H. Luo, M. Gong, Chin. J. Nonferrous Met. 22 (2012)

15.[35] H.Y. Yang, X.B. Chen, X.W. Guo, G.H. Wu, W.J. Ding, N. Birbilis, Appl. Surf. Sci. 258

(2012) 5472.[36] S.-E. Bae, J.-H. Yoon, C.-W.J. Lee, J. Phys. Chem. C 1125 (2008) 1533.[37] W. Kibalczyc, A. Zielenkiewicz, J. Cryst. Growth 82 (1987) 733.[38] M. Santamaria, F. Di Quarto, S. Zanna, P. Marcus, Electrochim. Acta 53 (2007)

1314.[39] G. Song, Adv. Eng. Mater. 7 (2005) 563.[40] K.Z. Chong, T.S. Shih, Mater. Chem. Phys. 80 (2003) 191.[41] M.W. Kendig, R.G. Buchheit, Corrosion 59 (2003) 379.[42] T. Ishizaki, N. Saito, Langmuir 26 (2010) 9749.[43] D.D.MacDonald,M.C.H.McKubre, in: J.R.Macdonald (Ed.), Impedance Spectroscopy:

Emphasizing Solid Materials and Systems, Wiley, New York, 1987, p. 133.[44] N.T. Kirkland, N. Birbilis, M.P. Staiger, Acta Biomater. 8 (2012) 925.[45] N.T. Kirkland, G. Williams, N. Birbilis, Corros. Sci. 65 (2012) 5.[46] M. Forsyth, W.C. Neil, P.C. Howlett, D.R. Macfarlane, B.R.W. Hinton, N. Rocher, T.F.

Kemp, M.E. Smith, ACS Appl. Mater. Interfaces 1 (2009) 1045.[47] M.D. Ger, K.H. Yang, Y. Sung, W.H. Hwu, Y.C. Liu, in: U.S.P.a.T. Office (Ed.) United

States Patent and Trademark Office, US, 2004.