Fatigue properties of Keronite® coatings on a magnesium alloy

Fatigue properties of Keronite® coatings on a magnesium alloy

Surface and Coatings Technology 182 (2004) 78–84 Fatigue properties of Keronite䉸 coatings on a magnesium alloy A.L. Yerokhina,*, A. Shatrovb, V. Sams...

2MB Sizes 0 Downloads 53 Views

Surface and Coatings Technology 182 (2004) 78–84

Fatigue properties of Keronite䉸 coatings on a magnesium alloy A.L. Yerokhina,*, A. Shatrovb, V. Samsonovb, P. Shashkovb, A. Leylanda, A. Matthewsa a

Department of Engineering Materials, The University of Sheffield, Mappin Str., Sheffield S1 3JD, UK b Keronite Ltd., Granta Park, Abington, Cambridge CB1 6GX, UK Received 2 April 2003; accepted in revised form 15 July 2003

Abstract In the paper, the feasibility of using the Keronite䉸 plasma electrolytic oxidation process to overcome the problem of fatigue performance reduction caused by anodising treatments in a Mg alloy is studied. Two types of coatings produced using different current regimes, and having two thicknesses of ;7 and ;15 mm, were tested using a rotating bending fatigue tester. SEM, XRD and optical microscopy techniques were used to evaluate possible fracture mechanisms involved in the initiation and propagation of the fatigue cracks. The results of the investigation demonstrate that Keronite䉸 coatings may cause no more than a 10% reduction in the endurance limit of the Mg alloy, which is substantially lower than the effect from conventional anodising. A probable cause of that reduction seems to be distortion of the metal subsurface layer rather than structural defects introduced by the oxide film. 䊚 2003 Elsevier B.V. All rights reserved. PACS: 62.20.Mk; 81.65.Mq; 89.20.qa Keywords: Plasma electrolytic oxidation; Fatigue; Oxide ceramic layer; Magnesium

1. Introduction During the last 5 years, magnesium alloys have made impressive inroads into areas where aluminium alloys were traditionally the materials of choice, e.g. applications in the aerospace and automotive industries, transportation, mobile communication devices and personal computers. The driving force behind this phenomenon lies not only in the improved affordability of commercial Mg alloys but also in the progress in magnesium surface science and technology. Being a particularly light material (dMgs1.74=103 kg my3) with a potentially very high strength-to-weight ratio (introducing the realistic prospect of weight savings of up to 50% against steel), magnesium exhibits great promise but also serious drawbacks—caused by its chemically active nature. The first of these is poor corrosion resistance, particularly for highly-alloyed grades exposed to galvanic corrosion; the *Corresponding author. Tel.: q44-1482-465072; fax: q44-1482466477. E-mail address: [email protected] (A.L. Yerokhin).

second is poor friction and anti-wear behaviour, particularly during dry sliding in oxidising environments, where galling can often occur. These drawbacks make a protective surface treatment an essential part of the manufacturing process for many magnesium components. A number of surface engineering techniques have been considered for the protection of Mg alloys w1–3x, with anodic oxidation being one of the most popular methods w1–7x. Anodising usually offers a relatively simple and cost-effective way of converting a metal surface into a thin layer of its own oxide, which is capable of protecting the parent metal from both corrosion and wear. However, a generic problem with anodising is the adverse effect on fatigue properties, which restricts applications for many anodised components w3x. Fatigue performance reduction in anodised metals is normally caused by several factors, including oxidationinduced surface tensile stress, structural defects in the oxide layer and substrate ‘age softening’, due to the heat associated with oxide film formation. In the case of magnesium, the combination of these factors appears

0257-8972/04/$ - see front matter 䊚 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0257-8972(03)00877-6

A.L. Yerokhin et al. / Surface and Coatings Technology 182 (2004) 78–84

to be particularly disadvantageous, since magnesia has both high specific heat of formation and a substantial lattice misfit with the metal. Although the residual surface stress could be reduced by adjustment of the film structure and chemical composition and the substrate age softening can be reduced by shortening the treatment time w4–8x, no adequate conventional anodising techniques have so far been found to minimise the risk of premature fatigue failure for magnesium alloys. An opportunity exists with the development of the plasma electrolytic oxidation (PEO) technique for magnesium alloys (which employs a plasma discharge at the oxide–electrolyte interface, usually referred to as a set of ‘sparks’ or ‘micro-arcs’) to modify the layer phase composition and structure and thus diminish the film strain w9–13x. The mechanism of oxide layer formation during PEO represents a complex combination of conventional anodic oxide film growth with plasma enhanced surface oxidation in ‘micro-arc’ discharge regions, leading to fusing and recrystallisation of the oxide film w9x. Originally, DC or amplitude-modulated AC were used in the PEO process, which allowed coating growth rates of only 1–2 mm miny1 w9,13x. However, an improved PEO method, utilising a pulsed bipolar current, has recently been developed and made commercially available by Keronite Ltd, which allows coating growth rates of up to 10 mm miny1 w14x. A substantial reduction in the thermal impact on the substrate can, therefore, be achieved. In this paper, a comparative study is carried out to investigate the fatigue properties of a magnesium alloy oxidised using both the original and the improved Keronite䉸 processes. 2. Experimental Fatigue test samples (type 2, ASTM E466-96 w15x) with ends diameter of 10 mm, minimum gauge length diameter of 6 mm and total length of 100 mm were made from a magnesium alloy rod (2% Al, 1% Zn, 0.2% Mn, Mg—balance). The samples were shaped by turning and grinding to the required dimensions, then polished to achieve a surface roughness of approximately 0.1 mm Ra. The Keronite䉸 oxidation treatment was performed under the conditions presented in Table 1 w13,14x. The original process (batch 1,2) was conducted using amplitude-modulated AC mode at mains frequency (50 Hz). The waveform of the applied voltage (current) was fixed during the process, determining the relation between the duration and amplitude of positive and negative cycles w13x. The improved Keronite䉸 process (batch 3) utilises a higher frequency (;103 Hz) bipolar current mode with independent control of duration and amplitude for both

79

positive and negative electrical pulses w14x. The amplitude of current pulses can, therefore, be increased whilst keeping the mean average current density in the same working range (i.e. 2–10 A dmy2). This, along with a short pulse lead time (;10y6 s), allows an increasing degree of ionisation in the plasma discharge. Higher frequency current pulses enable the creation of sequences of shorter, yet more energetic, microdischarge events, ensuring a better balance between ‘oxidising’ and ‘fusingyrecrystallising’ aspects of the coating formation process. Thus, the high frequency bipolar system is believed to allow three to five times enhancement of the coating deposition rate combined with substantial improvement in the surface layer quality, with less porosity and roughness as well as different phase composition w14x. After the oxidation treatment, the samples were thoroughly rinsed in water and dried using a hot air gun. Additionally, a batch of samples were oxidised to a depth of approximately 15 mm and then the oxide layer was carefully removed by polishing under a hydrocarbon coolant to restore the initial surface roughness, without disturbing the structure of the metal sub-surface layer possibly affected by the treatment. The coating microstructure and morphology were characterised by optical and scanning electron microscopy (Cambridge Stereoscan 200), observing the structure both parallel to the substrate and also in fractured and polished cross-sections. Phase composition was studied by XRD analysis using a Siemens D5000 diffractometer (Cu Ka radiation) operated in Seeman– Bohlin geometry with a 58 grazing angle. Knoop microhardness was evaluated using a Mitutoyo MVK G1 hardness tester at 25-g load with 15-s dwell time. Fatigue tests were performed according to ASTM E468-90 w16x at ambient atmospheric conditions using a rotating bending fatigue tester with a 125 mm cantilever length and operated at 1425 rpm. The base for the tests was chosen at the level of 106 cycles, though some of the samples were tested for (1.3–2)=106 cycles. Principle fatigue characteristics, such as endurance limit (S) and fatigue life (N), were derived from linear approximations applied to the corresponding parts of the ¨ fatigue curve plotted within standard Wohler coordinates (i.e. stress vs. log no. of cycles) w16x. 3. Results and discussion The magnesium alloy used for the experiments possessed a normally distributed and equiaxed granular microstructure (Fig. 1a), containing less than 1% porosity which was probably introduced during the manufacturing process. Pores of 1–3 mm in diameter are quite a common feature in the structure of magnesium alloys

80

Batch

1 2 3 a b

PEO regime

Layer

Current mode

Current density (A dmy2)

Electrolyte

Temperature (8C)

Treatment time (min)

Thickness (mm)

Composition

Hardness, HK25 (MPa)

Amplitude-modulated ACa Amplitude-modulated ACa Pulsed bipolarb

2–10 2–10 2–10

Low concentration, alkaline Low concentration, alkaline Low concentration, alkaline

15–40 15–40 15–30

3.5 7 4

7 15 15

MgO, MgAl2O4 MgO, MgAl2O4 MgO, Al2O3

400–650 400–650 400–650

As described in Ref. w13x. As described in Ref. w14x.

A.L. Yerokhin et al. / Surface and Coatings Technology 182 (2004) 78–84

Table 1 Characteristics of Keronite䉸 PEO processes for magnesium

A.L. Yerokhin et al. / Surface and Coatings Technology 182 (2004) 78–84

81

Fig. 1. Typical microstructure of the magnesium alloy used in the experiments (a) and typical morphology of PEO coating on a magnesium alloy (b).

and can be considered as pre-existing stress concentrators, which may affect the material fatigue properties regardless of the applied surface treatment. A typical surface morphology of a PEO coating on magnesium, as shown in Fig. 1b, reveals a non-uniform cellular structure of the oxide ceramic layer, created by the plasma microdischarges. The coating morphology features a sub-micrometer range porosity and a network of fine microcracks caused by relaxation of thermal and structural stresses in the surface layer during its oxidation. The microcrack network can also be considered as a possible source of pre-existing nuclei for fatigue cracks. Cross-sectional fracture micrographs of typical Keronite䉸 coatings on Mg (Fig. 2) illustrate the effect of current regime on the oxide layer morphology. The layer produced using the original Keronite process, which employed amplitude-modulated mains AC, possesses a coarser structure with larger pores and major fragments of brittle fracture (Fig. 2a). The main layer constituent thought to be responsible for the formation of such a brittle, glassy coating is MgAl2O4 (Fig. 3). Conversely, the layer produced using bipolar pulsing current contains a fine mixture of MgO (Periclase) and k-Al2O3 and appears to be dense and uniform, demon-

Fig. 2. SEM images of cross-sectional fractures of Keronite䉸 ceramic surface layers produced using the original process (a) and the improved process (b).

strating that the improved PEO technique offers wider possibilities for controlling the layer formation processes.

Fig. 3. Typical XRD patterns of oxide ceramic coatings on Mg alloys produced using the original (a) and the improved (b) Keronite䉸 processes.

82

A.L. Yerokhin et al. / Surface and Coatings Technology 182 (2004) 78–84

Fig. 4. Optical images of polished cross-sections of the Keronite䉸 coatings subjected to the fatigue testing. (a) Batch 1; (b) batch 2 and (c) batch 3. Insert to the figure (b) shows a strain accumulated in the metal at the coating–substrate interface.

Polished cross-sections of the coatings shown in Fig. 4a–c represent structures of the layers subjected to fatigue testing (batches 1, 2 and 3, respectively). It can be seen that the layer ;15-mm thick produced by the original Keronite䉸 technology is more uneven compared to that produced by the modified process. A treatmentaffected sub-layer is present in both cases at the metal– oxide interface (see insert in Fig. 4b), which could be characterised as a strain in the metal substrate rather than a thermally re-crystallised structure. The thickness of the treatment-affected sub-layer for these two coatings is estimated to be of the order of 10–12 and 5–7 mm, respectively, whereas the thickness of the sub-layer under the 7-mm-thick coatings cannot be reliably determined by optical microscopy. Results of coating fatigue tests are shown in Fig. 5, where the region of high-cycle fatigue failure and the unaffected zone are represented by linear fits with different slopes. The inflection point coordinates, which can be considered as the endurance limit and fatigue life, respectively, are collated in Table 2. From analysis of the test results, it follows that PEO causes a 3–10% reduction of the endurance limit; however, the transition to the non-fatigue region occurs at a much lower number of cycles, compared to the untreated Mg alloy. Although the experimental data are widely scattered (this is quite common for fatigue test results—which tend statistically to have a highly non-symmetrical and ‘skewed’ distribution) it can be noticed that the thickness increase of the coatings produced by the original Keronite䉸 technique leads to a moderate degradation of the fatigue performance (Fig. 5a, c and d, correspondingly) and application of the modified Keronite䉸 process reduces significantly this adverse effect (Fig. 5e). The fatigued substrate region of the samples where the oxide layer was removed after treatment is characterised by a steep SyN slope (Fig. 5b), which points to the dominating role of the treatment-affected sub-layer in the initiation and propagation of fatigue cracks. Thus, the strengthening effect from the oxide ceramic coating becomes very evident when the bending stress approaches the endurance limit. It seems that in this region the coating provides a more effective barrier to the development of fatigue cracks. The process of crack propagation through the sample can be traced in Fig. 6, where three following fracture stages are marked: (I) crack initiation; (II) crack propagation and (III) catastrophic failure. The fracture of the bare Mg alloy (Fig. 5a) features an extended stage of crack propagation; a ductile fracture mode is characteristic for this batch of samples. At the same time, the samples subjected to the oxidation treatment show a trend towards brittle fracture with a large ‘stage III’ region. This suggests that the fatigue cracks in oxidised samples are formed and propagated at a higher stress intensity factor, which is probably caused by stress concentrations at the structural

A.L. Yerokhin et al. / Surface and Coatings Technology 182 (2004) 78–84

83

defects in the oxide layer andyor the treatment affected zone.

Table 2 Effect of Keronite䉸 surface treatments on fatigue properties of magnesium

4. Conclusions

Batch

Layer thickness (mm)

Endurance limit (MPa)

Fatigue life (cycles)

As received Mg 1 2 3

– 7 15 15

85 81 77 83

6.5=105 2.3=105 2.8=105 1.3=105

The results of fatigue tests demonstrate that Keronite䉸 coatings may cause no more than a 10% reduction in endurance limit of the Mg alloy studied, which is substantially lower than the effect from conventional anodising. At the endurance limit, the transition to the non-fatigue region for the oxidised samples

As derived form linear fits of the fatigue diagrams shown in Fig. 4.

Fig. 5. Fatigue diagrams for the tested samples. (a) Uncoated Mg alloy; (b) Mg alloy after removing the oxide coating; (c) batch 1; (d) batch 2 and (e) batch 3. Blank dots indicate that the sample has not failed.

84

A.L. Yerokhin et al. / Surface and Coatings Technology 182 (2004) 78–84

metal subsurface layers induced during the oxidation process. Acknowledgments The coated samples examined in this work were provided by Keronite Ltd. Financial support from the UK Engineering and Physical Research Council is acknowledged with thanks. References

Fig. 6. Fatigue fracture appearances for the samples tested at 92 MPa. (a) Uncoated Mg alloy; (b) Mg alloy after removing the oxide coating; (c) batch 2 and (d) batch 3. Marks on the micrographs represent the following fracture regions: (I) crack initiation; (II) crack propagation and (III) catastrophic failure.

occurs substantially earlier than for the bare Mg alloy, due to the inhibition of the crack initiation by the oxide ceramic layer. The improved Keronite䉸 process provides dense and uniform oxide ceramic layers with a fine-grained microstructure, which is more favourable for components experiencing fatigue loading. A significant role in initiation and propagation of the fatigue cracks through the oxidised samples is played by strain distortion of the

w1x J.E. Gray, B. Luan, J. Alloys Compd. 336 (2002) 88. w2x E.F. Emley, Principles of Magnesium Technology, Pergamon Press, London, 1966. w3x J.E. Hillis, ASTM Handbook, Surface Engineering, vol. 5, ASTM Int, 1994, p. 819. w4x S. Ono, K. Asami, T. Osaka, N. Masuko, J. Electrochem. Soc. 143 (3) (1996) L62. w5x O. Khaselev, J. Yahalom, J. Electrochem. Soc. 145 (1) (1998) 190. w6x Y. Zhang, Ch. Yan, F. Wang, H. Lou, Ch. Cao, Surf. Coat. Technol. 161 (2002) 36. w7x US 5792335, T.F. Barton, Publ. 11.08.98. w8x P. Ross, J. MacCulloch, C. Clapp, R. Esdaile, SAE Technical Paper Series 1999-01-0925. w9x A.L. Yerokhin, X. Nie, A. Leyland, A. Matthews, S.J. Dowey, Surf. Coat. Technol. 122 (1999) 73. w10x US 5385662, H.-Kletke, B. Dora, P. Kurtze, Publ. 31.01.95. w11x O. Khaselev, D. Weiss, J. Yahalom, J. Electrochem. Soc. 146 (5) (1999) 1757. w12x O. Khaselev, D. Weiss, J. Yahalom, Corros. Sci. 43 (2001) 1295. w13x WO 99y31303, A.S. Shatrov, Publ. 24.06.99. w14x WO 01y30572, A.S. Shatrov, Publ. 03.05.01; WO 01y72384, A.S. Shatrov, Publ. 04.10.01; WO 02y22902, N.A. Belov, V.S Zolotarevsky, A.S. Shatrov, Publ. 21.03.02. w15x ASTM E466-96, Annual Book of ASTM Standards, Section 3, 03.01 (1996) 465. w16x ASTM E468-90, Annual Book of ASTM Standards, Section 3, 03.01 (1996) 475.