Materials and Design 30 (2009) 3731–3737

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier .com/locate /matdes

Comparative study of mechanical and tribological properties of aluminacoatings formed on aluminium in various conditions

W. Bensalah a,*, K. Elleuch b, M. Feki a, M. DePetris-Wery c, H.F. Ayedi a

a Unité de recherche de Chimie Industrielle et Matériaux (URCIM), ENIS, Département de Genie Des Materiaux, B.P.W. 1173-3038, Sfax, Tunisiab Laboratoire des Systèmes Electromécaniques (LASEM), ENIS, B.P.W. 1173-3038, Sfax, Tunisiac IUT Mesures Physiques d’Orsay – Université Paris XI, Plateau du Moulon, 91400 Orsay, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 16 December 2008Accepted 10 February 2009Available online 15 February 2009

Keywords:C. Alumina coatingE. Vickers microhardnessI. AbrasionI. Failure

0261-3069/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.matdes.2009.02.005

* Corresponding author. Tel.: +216 74 274 088; faxE-mail address: walbensalah@yahoo.fr (W. Bensala

Influence of oxalic acid addition to sulphuric acid bath on the mechanical properties of the oxide layersformed on aluminium has been examined. For this purpose two Doehlert experimental designs withthree variables (temperature, current density, sulphuric acid concentration) and four variables (oxalicacid concentration, temperature, current density, sulphuric acid concentration) were realized. Fourresponses were studied namely: growth rate (Ve), Vickers microhardness (D), weight loss after abrasion(Wa) and deflection at failure (Df) of the anodic oxide layer. A comparative study based on surfaceresponses was achieved. Compared with sulphuric acid bath, it was found that the addition of oxalic acidpermits high growth rates, high abrasion resistance and high microhardness but less ductile layers. Theobserved mechanical properties of the oxide layers can be related to their morphology revealed by SEMobservations and their chemical composition determined by GDOES.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

As weight-saving materials, aluminium and aluminium alloysare becoming increasingly important for both technical and eco-nomical considerations [1,2]. However, poor friction property andwear resistance restricted their applications in the industrial fieldsrequiring sliding contact. In order to improve mechanical proper-ties of aluminium, anodizing has been commonly used [1,2].

Anodizing, a surface treatment which originated in the 1930s, isan electrochemical process, consists on converting aluminium intoits oxide by appropriate selection of the electrolyte and the anod-izing conditions, such as current density, voltage and temperature[3,4]. By adjusting the conditions used in the anodizing process,oxide layers can be produced with almost any desired properties,from thin films used in decorative applications to the extremelyhard, corrosion resistant oxides used in engineering applications[5–9]. In the last few decades, the effect of anodizing conditionsand the composition of single acid electrolytes, i.e. solution of sul-phuric acid, chromic acid, phosphoric acid or oxalic acid, on theproperties of the anodic layer such as corrosion resistance, microh-ardness and abrasion resistance were investigated [5–8]. To im-prove the properties of the anodic layer and/or to find analternative of chromic acid anodizing process, mixed acid electro-

ll rights reserved.

: +216 74 275 595.h).

lytes such as oxalic acid–chromic acid, nitric acid–sulphuric acidand boric acid–sulphuric acid were implemented [9–11].

In previous works [12,13], we have optimized some of mechan-ical properties of the anodic oxide layer elaborated on aluminiumin sulphuric acid and oxalic acid–sulphuric acid baths using twoDoehlert experimental designs. The electrolyte composition, tem-perature and current density were retained as variables to conducteach study.

The objective of this paper is to investigate the effect of oxalicacid addition to a sulphuric acid bath, on the thickness and themechanical properties of the elaborated oxide layers, namely:Vickers microhardness (D), weight loss after abrasion (Wa) anddeflection at failure (Df). Comparisons were made using the previ-ously established models obtained from Doehlert experimental de-sign applied to each of the anodizing process [12,13]. Moreover,the morphology and the composition of the anodic oxide layerwere examined by scanning electron microscopy (SEM), opticalmicroscopy and glow-discharge optical emission spectroscopy(GDOES).

2. Experimental

2.1. Materials and procedures

Parrallelipedic AA1050 samples 100 � 25 � 3 mm3 were usedas the substrate for anodic conversion treatment. Prior to anodiz-ing, samples were mechanically polished to P1000 grade paper

3732 W. Bensalah et al. / Materials and Design 30 (2009) 3731–3737

followed by (i) chemical polishing in a 15:85 (v/v) mixture of con-centrated HNO3 and H3PO4 at 85 �C for 2 min, (ii) etching in 1 MNaOH solution at room temperature for 1 min and (iii) chemicalpickling in 30% (v/v) HNO3 solution at room temperature for 30 s.Water rinsing was used after each step. Afterwards, samples wereanodized in vigorously stirred acid solutions (sulphuric acid andoxalic–sulphuric acid bath) maintained within ±0.1 �C of the settemperature for 90 min then washed in deionised water and dried.The used cathodes were also aluminium sheets. Sulphuric, nitricand phosphoric acids are analytical grade chemicals.

2.2. Testing methods

In order to characterize the anodic oxide layer, four tests wereconducted. The oxide layer thickness was measured using ELCOM-ETER 355 Top Thickness Gauge. The Vickers microhardness wascarried out using DELTALAB HVS-1000 tester (200 g load for 15 s).

Abrasion tests were conducted using a pin-on-disc machine.Anodized samples with dimensions of 20 � 20 � 3 mm3 werebrought into contact with 320 grit SiC paper, fixed on a rotatingdisc with a constant speed of 20 rpm. The applied normal loadwas 5 N and the test duration was 1 min.

Deflection at failure of the anodic oxide films on aluminium wasmeasured by performing three point flexure tests on parrallelipe-dic samples 100 � 25 � 3 mm3 at room temperature. A universalmachine [Lloyd instruments LR 50KN] was used for this purpose.Loading speed was fixed at 2 mm min�1 and the calibrated dis-tance was 50 mm.

The morphology of the oxide layer was studied from the topside of the layer using a Scanning Electron Microscope SEM (JeolJSM-6400F and Philips XL30).

The morphology of worn oxide layer surfaces was studied usinga LEICA optical microscope.

The distribution of species in the anodic oxide layer was deter-mined by depth profiling using a Jobin Yvon GD Profiler instrumentequipped with a 4 mm diameter anode and operating at pressureof 800 Pa and a power of 600 W in an argon atmosphere. The rele-vant wave-lengths (nm) were as follows: Al, 396.15; O, 130.22, S,181.73 and C, 156.14. The sputtering layer was 6 lm thick.

Table 1Doehlert experimental design in coded variables (k = 4 and N0 = 4).

2.3. Methodology and design of experiments

The Doehlert experimental design [14] was performed to studythe effect of the anodizing conditions on the performance of the re-tained baths and on the aluminium oxide layer properties. As cur-rently used in experimental design, natural variables Uj weretransformed into coded variables Xj according to the followingrelation [15–17]:

Xj ¼Uj � Ujð0Þ

DUj

where Uj(0) is the value of Uj at the centre of the study domain andDUj is the variation step. Doehlert design requires N = k2 + k + N0

experiments, where k is the number of the factors and N0 the num-ber of centre runs. Experiments at the centre are required to con-duct statistical tests. For example, Doehlert experimental designin coded variables corresponding to 4 variables is given in Table1. The delimited zones correspond to the three and four variablesexperimental designs. Experiment 1 is the central run for each ofthe experimental design. Replicates at the central level of the vari-ables (experiments 22–25) were conducted in order to estimate thepure error variance.

A full quadratic model, including interaction terms, was as-sumed to describe the relationship between each response Yi andexperimental variables Xj:

Y ¼ b0 þXk

j¼1

bjXj þXk

j¼1

Xk

h¼jþ1

bjhXjXh þXk

j¼1

bjjX2j

where b0 is the constant of the model, bj the first degree coefficients,bjh the cross-products coefficients and bjj the quadratic coefficients.

It is to mention that NEMROD W software [18] was used fordata calculation and treatment.

3. Results and discussion

3.1. Variables, study domains and model expressions

In order to compare the performances of sulphuric acid andoxalic acid–sulphuric acid anodizing processes, a Doehlert experi-mental design was applied for each of them. Four responses wereretained:

– growth rate of the anodic oxide layer, Ve, noted Y1 (lm min�1),– Vickers microhardness of the anodic oxide layer, D, noted Y2

(HV),– weight loss by abrasion of the anodic oxide layer, Wa, noted Y3

(mg),– deflection at failure of the anodic oxide layer Fr, noted Y4 (mm).

Table 2Study domains.

Variables Number of levels Centre Uj(0) Variation step DUj

Sulphuric acid anodizing processU1 (�C) 5 14 11U2 (A dm�2) 7 2 1U3 (g L�1) 3 160 40

Oxalic sulphuric acid anodizing processU1 (g L�1) 5 10 8U2 (�C) 7 16.5 13.5U3 (A dm�2) 7 2 1U4 (g L�1) 3 160 40

Fig. 3. Vickers microhardness of the anodic oxide layer formed in oxalic–sulphuricacid bath versus T, J: (a) Csul = 160 g L�1 and Cox = 6 g L�1; (b) Csul = 160 g L�1 andCox = 10 g L�1; (c) Csul = 160 g L�1 and Cox = 15 g L�1.

W. Bensalah et al. / Materials and Design 30 (2009) 3731–3737 3733

Three variables Uj were selected for the sulphuric acid anodiz-ing process: (i) U1: the anodizing temperature, T (�C), (ii) U2: thecurrent density, J (A dm�2) and iii) U3: the sulphuric acid concen-tration, Csul (g L�1). The study domain (number of levels, centreand variation step) is given in Table 2. The obtained models werevalidated using ANOVA analysis. Their expressions were the fol-lowing [12]:

Y1 ¼ 0:527� 0:064X1 þ 0:279X2 þ 0:013X21 � 0:004X2

2

þ 0:022X23 � 0:006X1X2 þ 0:014X1X3 þ 0:001X2X3

Y2 ¼ 478:2� 135:6X1 þ 41:8X2 þ 12:2X3 � 190:3X21 � 248:9X2

2

þ 22:9X23 � 75:1X1X2 þ 258:6X1X3 � 192:9X2X3

Y3 ¼ 25:35þ 15:98X1 þ 0:43X2 þ 2:94X3 þ 8:05X21 � 0:35X2

2

� 3:55X23 þ 3:46X1X2 þ 1:71X1X3 � 3:46X2X3

Y4 ¼ 6:32� 0:86X1 þ 1:55X2 þ 0:41X3 þ 0:42X21 � 0:31X2

2

� 0:17X23 þ 0:12X1X2 þ 0:27X1X3 � 0:08X2X3

Four variables Uj were retained for the oxalic–sulphuric acid anod-izing process: (i) U1: the oxalic acid concentration, Cox (g L�1), (ii)U2: the anodizing temperature, T (�C), (iii) U3: the current density,J (A dm�2) and (iv) U4: the sulphuric acid concentration, Csul

(g L�1). The corresponding study domain is given in Table 2. Theexpressions of the validated models were the following [13]:

Fig. 1. Growth rate of the anodic oxide layer versus T, J: (a) Csul = 160 g L�1 andCox = 10 g L�1; (b) Csul = 160 g L�1 and Cox = 0 g L�1.

Fig. 2. SEM of the anodic oxide layer obtained under: (a) T = 15 �C, J = 2 A dm�2 and Csul = 160 g L�1; (b) Cox = 10 g L�1, T = 15 �C, J = 2 A dm�2 and Csul = 160 g L�1.

Fig. 4. Vickers microhardness of the anodic oxide layer versus T, J: (a)Csul = 160 g L�1 and Cox = 10 g L�1; (b) Csul = 160 g L�1 and Cox = 0 g L�1.

Fig. 5. Weight loss by abrasion of the anodic oxide layer formed in oxalic–sulphuricacid bath versus T, J: (a) Csul = 160 g L�1 and Cox = 6 g L�1; (b) Csul = 160 g L�1 andCox = 10 g L�1; (c) Csul = 160 g L�1 and Cox = 15 g L�1.

Fig. 6. Weight loss by abrasion of the anodic oxide layer versus T, J: (a)Csul = 160 g L�1 and Cox = 10 g L�1; (b)Csul = 160 g L�1 and Cox = 0 g L�1.

Fig. 7. Optical microscopy of worn surfaces: (a) non treated aluminium; anodic oxide layeCsul = 160 g L�1.

3734 W. Bensalah et al. / Materials and Design 30 (2009) 3731–3737

Y1 ¼ 0:546þ 0:002X1 � 0:044X2 þ 0:303X3 � 0:008X4

þ 0:006X1X3 � 0:039X2X3 � 0:021X1X4 � 0:001X2X4

þ 0:001X3X4 þ 0:009X21 � 0:011X2

2 þ 0:007X23 � 0:052X2

4

Y2 ¼ 458:4þ 19:2X1 � 130:8X2 � 14:7X3 � 7:7X4 þ 72:7X1X2

� 11:6X1X3 � 16:2X2X3 � 25:5X1X4 þ 35X2X4

þ 103:5X3X4 þ 14:6X21 � 120:1X2

2 � 30:7X23 � 25:2X2

4

Y3 ¼ 21:4� 3:2X1 þ 1:847X2 þ 0:939X3 þ 1:739X4 � 9:238X1X2

þ 2:654X1X3 þ 8:131X2X3 þ 2:056X1X4 � 1:004X2X4

þ 1:42X3X4 þ 1:1X21 þ 1:1X2

2 þ 5:39X23 � 8:349X2

4

Y4 ¼ 3:6þ 0:270X1 þ 0:191X2 þ 1:347X3 þ 0:032X4 þ 0:173X1X2

� 0:551X1X3 � 0:271X2X3 � 0:427X1X4 � 0:283X2X4

þ 0:052X3X4 þ 0:55X21 � 0:217X2

2 þ 0:367X23 � 0:34X2

4

3.2. Comparison of the anodizing processes

The established models for the retained anodizing processeswill be used to undertake comparisons via response surfaces.

3.2.1. Growth rate of the anodic oxide layerThe effect of oxalic acid addition to sulphuric acid bath on the

growth rate can provide rapid information on the anodic layerstructure modifications. Hence these modifications can be relatedto and help to understand possible effect on the mechanical prop-erties of the oxide layer.

Fig. 1 shows surface responses of growth rates corresponding tosulphuric acid (Fig. 1b) and oxalic/sulphuric acid (Fig. 1b) anodiz-ing processes. As can be seen, (i) oxalic acid addition enlarges thetemperature domain and (ii) favours the growth rate of the anodicoxide layer. For both anodizing processes, high values of thegrowth rate are obtained with high current densities and low elec-trolyte temperatures. In fact, the addition of oxalic acid (weak acid)to sulphuric acid (strong acid) decreases the aggressiveness of theelectrolyte towards the oxide and favours, thus, the oxide layergrowth [10]. On the other hand, the increase of current density

rs, (b) Cox = 0 g L�1,(c) Cox = 6 g L�1, (d) Cox = 15 g L�1 for T = 16.5 �C, J = 2 A dm�2 and

W. Bensalah et al. / Materials and Design 30 (2009) 3731–3737 3735

make the dissolution of the oxide in the bottom of pores more pro-nounced and lead to higher growth rates [9].

SEM photographs, on top surface, of anodic layers formed in sul-phuric acid and oxalic acid–sulphuric acid baths are shown inFig. 2.

Both anodic layers present a porous structure. In Fig. 2a thestructure is more open with higher pore diameters (black spots)and lower wall thicknesses. It could be noted that at some loca-tions the oxide walls between neighboring pores had completelybeen dissolved, inducing a merging of the involved pore mouths.Fig. 2b shows oxide nuclei (white dots) [19] and small pores sepa-rated by thick pore walls. From these observations, the anodic layerformed with the addition of oxalic acid appears more compact.Therefore, the organic acid addition forms thicker and less porousanodic oxide layers.

3.2.2. Vickers microhardness of the anodic oxide layerThree oxalic acid concentrations, i.e. 6 g L�1, 10 g L�1 and

15 g L�1, were selected to investigate the effect of current densityand electrolyte temperature on the microhardness evolution(Fig. 3). Surface responses indicate that the amount of added oxalicacid changes significantly the microhardness of the oxide layer.High concentrations of the added acid lead to high Vickers microh-ardness of the layer for temperatures superior than 10 �C.

On the other hand, the trend indicates that increasing the cur-rent density in the selected range does not change significantlythe microhardness of the formed oxide layers. It is to mention thata tendency to decrease at the highest current densities seems tooccur.

As observed, increasing oxalic acid concentration leads to a lit-tle increase in layer thickness which can also generates an increaseof sulfate anions composition along the deeper pore walls [20]. Allthese facts seem to enhance the anodic oxide layer microhardness.

In addition, higher driving force accelerates the reactions at thesubstrate/oxide which can causes cracks in this region. Thesecracks are considered most likely ascribed to the internal stressgenerated by the growth of the oxide [21]. At the oxide/electrolyteinterface many defects can also occur (open porosities, local burn-ing, roughening, etc.) [22–24]. All these defects are expected to de-crease the microhardness of anodic oxide layers.

Fig. 4 shows surface responses of Vickers microhardness ob-tained by sulphuric acid and oxalic/sulphuric acid anodizing pro-

Fig. 8. Deflection at failure of the anodic oxide layer formed in oxalic–sulphuricacid bath versus T, J: (a) Csul = 160 g L�1 and Cox = 6 g L�1; (b) Csul = 160 g L�1 andCox = 10 g L�1; (c) Csul = 160 g L�1 and Cox = 15 g L�1.

cesses. As can be seen, oxalic acid addition enlarges thetemperature domain and enhances significantly the response over-all the working domain. For both of the anodizing processes, only,at low electrolyte temperatures, oxides with high microhardnesswere formed.

3.2.3. Weight loss by abrasion of the anodic oxide layerAs proceeded with the Vickers microhardness, the same oxalic

acid concentrations were selected to investigate the influence ofcurrent density and temperature on the weight loss by abrasionof the elaborated anodic oxide layers (Fig. 5). The tendencies indi-cate that the amount of added oxalic acid has a strong positive ef-fect on the abrasion behavior. In fact, with high oxalic acidconcentrations low weight losses were recorded.

On the other hand, high abrasion resistance of the oxide layer isobtained at low temperatures and current densities. From Fig. 5two domains can be distinguished. For temperatures superior than10 �C, high oxalic acid concentrations are required to ensure highabrasion resistance. At temperatures inferior than 10 �C, only lowoxalic acid concentrations are sufficient to produce abrasion resis-tant layers.

Fig. 6 shows that oxalic acid addition to the sulphuric acid bathenhances the abrasion resistance of the anodic oxide layers fortemperatures up to 10 �C. The variation of the current density, inthe working domain, does not change significantly the response.

The morphology of the worn surface is strongly dependent onthe anodizing conditions. The change in morphology may be ob-served in Fig. 7. Four samples are shown after abrasive wear tests.As can be seen, non treated aluminium has experienced severeabrasive wear. Distinct parallel and continuous grooves with highquantity of plastic deformation were occurred (Fig. 7a). With theaddition of oxalic acid, grooves were reduced to fine scratches(Fig. 7b–d) and plastic deformations were observed. Surface crackswere sometimes noticed (Fig. 7b). From Fig. 7b–d, there is no dras-tic change in morphology of the worn surfaces was perceived withincreasing addition of oxalic acid. The main wear mechanism forthe treated specimens was dominant by abrasive wear. Apparently,the above characteristics of worn surfaces can be related to the sur-face microhardness of different specimens.

3.2.4. Deflection at failure of the anodic oxide layerThe obtained anodic oxide on aluminium is hard and usually

accompanied by the risk of brittle failure, especially under sur-face-concentrated loads from static or cyclic contacts. In this para-graph, the influence of oxalic acid added to sulphuric acid bath on

Fig. 9. Deflection at failure of the anodic oxide layer versus T, J: (a) Csul = 160 g L�1

and Cox = 10 g L�1; (b) Csul = 160 g L�1 and Cox = 0 g L�1.

3736 W. Bensalah et al. / Materials and Design 30 (2009) 3731–3737

deformation and fracture of these films will be discussed. The ef-fect of current density and electrolyte temperature on the deflec-tion at failure of anodic layers was examined through threeoxalic acid concentrations: 6 g L�1, 10 g L�1 and 15 g L�1. The ten-dencies (Fig. 8) indicate that the amount of added oxalic acid hasa negative effect on the deflection at failure. In fact, with high oxa-lic acid concentrations more brittle oxide layers were formed.

According to Fig. 9, oxalic acid addition leads to less ductile ano-dic oxide layers and this throughout the study domain.

Fig. 10. Back side optical micrographs of anodized samples elaborated under: (a) T = 1Csul = 160 g L�1.

0

200

400

600

800

1000

0 100 200 300 400 500 600Abrasion time(s)

Inte

nsity

(arb

it. u

nit)

O*1000C*10AlS*40

(1)

o

Al

S

C

0

200

400

600

800

1000

0 100 200Abrasion

Inte

nsity

(arb

.uni

t)

o

S

a

b

Fig. 11. GDOES profile of anodized surface obtained under: (a) T = 14 �C, J = 2 A dm�

distribution of Al, O, C and S; (b-2) distribution of C.

When the anodized aluminium is subjected to flexural test, theoxide layer undergoes elastic deformation while the metal sub-strate undergoes elastic–plastic deformation. As the continuity be-tween the anodic oxide layer and the substrate must bemaintained at their interface before failure, this gives rise to inter-facial shear stress when the hard oxide layer, which shows littledeformation, inhibits deformation of the substrate. As a result ofthe interfacial shear stress, force is transferred from aluminiumto the oxide layer thus inducing the tensile stress in the coating.

5 �C, J = 2 A dm�2 and Csul = 160 g L�1; (b) Cox = 10 g L�1, T = 15 �C, J = 2 A dm�2 and

0

20

40

60

80

0 200 400 600 800Abrasion time (s)

Inte

nsity

(arb

it. u

nit)

)2(

C*40

300 400time (s)

O*1000AlS*40

Al

2, Csul = 160 g L�1; (b) Cox = 10 g L�1, T = 14 �C, J = 2 A dm�2, Csul = 160 g L�1: (b-1)

W. Bensalah et al. / Materials and Design 30 (2009) 3731–3737 3737

As the substrate deformation increases, this tensile stress willaccumulate and eventually reach the fracture strength of the coat-ing. Parallel long cracks were, then, observed on the surface of theoxide layer in a direction perpendicular to the tensile loading axis(Fig. 10). According to all the obtained results (not shown here), thepatterns of the fractured oxide layers were independent to the bathcomposition. In fact, it was reported that the deflection at failureand the average crack spacing between normal fracture lines weresignificantly affected by the thickness of the anodic oxide layer[25].

3.3. Chemical analysis of the anodic oxide layer

In order to inspect possible incorporation of sulfate and oxalateanions in porous oxide layer, chemical analysis of the oxide layerwas conducted using GDOES. Fig. 11 shows a depth profile of theoxide layer elaborated in sulphuric acid bath (Fig. 11a) and sulphu-ric acid–oxalic acid bath (Fig. 11b and c). The distribution of alu-minium, oxygen and sulphur were revealed clearly through theoxide layer elaborated in sulphuric acid bath. With the additionof oxalic acid to sulphuric bath, the GDOES of the oxide layer showsthe presence of carbon. The presence of sulphur and carbon can beexplained by the inward migration of SO2�

4 and C2O2�4 anions

through the pores of the layer [26,27]. A relatively narrow peak lo-cated at the inner interface metal/oxide suggests an accumulationof SO2�

4 ions (inner interface).

4. Conclusion

A comparative study of the sulphuric acid and oxalic acid–sul-phuric acid anodizing processes was conducted using responsesurfaces. A three variables (temperature, current density, sulphuricacid concentration) and four variables (oxalic acid concentration,temperature, current density, sulphuric acid concentration) Doehl-ert experimental designs were applied for each process. Four re-sponses were retained namely: growth rate (Ve), Vickersmicrohardness (D), weight loss after abrasion (Wa) and deflectionat failure (Df). It was found that the addition of oxalic acid to sul-phuric acid bath permits (i) high growth rates and more compactlayers, (ii) high abrasion resistance and microhardness and (iii) lessductile layers.

From the obtained results, abrasion resistance and microhard-ness were likely to be highly correlated and the mechanical prop-erties of oxide layers appear to be related with their morphologyrevealed by SEM observations and their chemical compositiondetermined by GDOES.

References

[1] Baumeister J, Banhart J, Weber M. Aluminium foams for transport industry.Mater Design 1997;18:217–20.

[2] King F. Aluminium and its alloys. UK: Ellis Horwood Limited; 1987.[3] Wernick S, Pinner R, Sheasby P. The surface treatment of aluminium and its

alloys. 5th ed. UK: Finishing Publication Ltd.; 1987.[4] Young L. Anodic oxide films. London: Academic Press; 1961.[5] Mezlini S, Elleuch K, Kapsa Ph. The effect of sulphuric anodisation of

aluminium alloys on contact problems. Surf Coat Technol 2007;201:7855–64.[6] Lopez V, Otero E, Bautista A, Gonzalez JA. Sealing of anodic films obtained in

oxalic acid baths. Surf Coat Technol 2000;124:76–84.[7] Jagminas A, Bigeliene D, Mikulskas I, Tomasiunas RJ. Growth peculiarities of

aluminium anodic oxide at high voltages in diluted phosphoric acid. CrystGrowth 2001;233:591–8.

[8] Lunder O, Walmsley JC, Mack P, Nisancioglu K. Formation and characterisationof a chromate conversion coating on AA6060 aluminium. Corros Sci2005;124:1604–24.

[9] Shih H-H, Tzou S-L. Study of anodic oxidation of aluminum in mixed acid usinga pulsed current. Surf Coat Technol 2000;124:278–85.

[10] Moutarlier V, Gigandet MP, Pagetti J, Normand B. Influence of oxalic acidaddition to chromic acid on the anodising of Al 2024 alloy. Surf Coat Technol2004;182:117–23.

[11] Domingues L, Fernandes JCS, Belo MDC, Ferreira MGS, Rosa LG. Anodizing ofAl2024-T3 in a modified sulphuric acid/boric acid bath for aeronauticalapplications. Corros Sci 2003;45:149–60.

[12] Bensalah W, Elleuch K, Feki M, Wery M, Gigandet MP, Ayedi HF. Optimizationof mechanical and chemical properties of sulphuric anodized aluminium usingstatistical experimental methods. Mater Chem Phys 2008;108:296–305.

[13] Bensalah W, Elleuch K, Feki M, Wery M, Ayedi HF. Optimization of anodic layerproperties on aluminium in mixed oxalic/sulphuric acid bath using statisticalexperimental methods. Surf Coat Technol 2007;201:7855–64.

[14] Doehlert DH. Appl Stat 1970;19:231.[15] Goupy J. Plans d’expériences pour surfaces de réponses. Paris: DUNOD; 1999.[16] Khuri AI, Cornell JA. Response surfaces design and analyses. New York: M.

Dekker; 1996.[17] Mathieu D, Phan-Tan-Luu R. Plans d’Expériences: application à

l’entreprise. Paris: Technip; 1995.[18] Mathieu D, Noney J, Phan-Tan-Luu R. NEMROD-W software. Marseille: LPRAI;

2002.[19] Sullivan JPO, Wood GC. The morphology and mechanism of formation of

porous anodic films on aluminium. Proc Roy Soc Lond 1970;317:511–43.[20] Rasmussen J. New insights into the microhardness of anodized aluminium.

Met Finish 2001;99:46–51.[21] Li X, Nie X, Wang L, Northwood DO. Corrosion protection properties of anodic

oxide coatings on an Al–Si alloy. Surf Coat Technol 2005;200:1994–2000.[22] Aerts T, De Graeve I, Terryn H. Study of initiation and development of local

burning phenomena during anodizing of aluminium under controlledconvection. Electrochim Acta 2001;54:270–9.

[23] Aurongzeb D. Interface evolution during electrochemical oxidation–dissolution. Appl Surf Sci 2005;252:872–7.

[24] Liu W, Zuo Y, Tang Y, Zhao X. The cracking behaviour of anodic films on castaluminium alloy after heating in the temperature range up to 300 �C. Surf CoatTechnol 2008;202:4181–8.

[25] Choo YH, Devereux OF. Mechanical failure of anodic films on aluminium andtantalum. J Electrochem Soc 1976;123(12):1868–75.

[26] Shimizu K, Habazaki H, Skeldon P, Thompson GE, Wood GC. Migration ofsulphate ions in anodic alumina. Electrochim Acta 2000;45:1805–9.

[27] Shimizu K, Habazaki H, Skeldon P, Thompson GE, Wood GC. Migration ofoxalate ions in anodic alumina. Electrochim Acta 2001;46:4379–82.