Microstructure and mechanical performance of pulsed current gas tungsten arc surface engineered composite coatings on Mg alloy reinforced by SiCp

Materials Science and Engineering A 490 (2008) 208–220

Microstructure and mechanical performance of pulsed current gas tungsten arc surface engineered composite coatings on Mg alloy reinforced by SiCp Shuyan Zhang a , Fusong Jiang a,∗ , Wenbin Ding b a

Key Laboratory for Ultrafine Materials of Ministry of Education, School of Materials Science and Engineering, East China University of Science and Technology, 130 Meilong Road, Shanghai 200237, PR China b China Offshore Oil Engineering Co. Ltd., 1078 Danjiang Road, Tianjin 300451, PR China Received 20 July 2007; received in revised form 4 January 2008; accepted 7 January 2008

Abstract A new pulsed current GTA surface-modified process was used to fabricate composite layer on the surface of Mg alloy AZ31. Current pulsing enhances fluid flow, reduces temperature gradients and causes a continual change in the weld pool size and shape, so that it is responsible for refining the solidification structure in the composite layer. The observed grain refinement was shown to result in an appreciable increase in composite layer bend strength. Composite layers with lower scan speed have higher bend strengths and they also seem to have “good” metallurgical bond with the substrate thus showing better mechanical behavior than the other higher scan speeds used in this present study. The wear rate of the composite layer decreases linearly with increase in SiCp volume fraction and the wear resistance of composite layer varies inversely with square of the reinforcement size. Composite layers with higher H/E have smaller accumulative strain, smaller accumulative strain energy, and thus better wear resistance. The wear mechanism was oxidation at low-applied load levels and adhesion/delamination at high-applied load levels. © 2008 Elsevier B.V. All rights reserved. Keywords: Mechanical properties; Gas tungsten arc; Surface coatings; Magnesium alloy; Silicon carbide particles

1. Introduction Magnesium alloys are gaining increasing importance as structural materials for applications where weight reduction is critical owing to their low densities (1.75–1.85 g/cm3 ) and high stiffness-to-weight ratio, and these applications include automotive, industrial, materials handling, and aerospace equipment where there is an obvious need for lightweight materials [1,2]. However, magnesium alloys have not been used for high-performance applications due to their poor mechanical performance in engineering applications. Therefore, magnesium matrix composites are expected in high-performance applications due to their low density and enhanced mechanical performance [3–7]. But the actual applications of magnesium matrix composites have been limited because of their high production cost and complex process [8], and in some instances, it ∗

Corresponding author at: Key Laboratory for Ultrafine Materials of Ministry of Education, School of Materials Science and Engineering, East China University of Science and Technology, P.O. Box 258, 130 Meilong Road, Shanghai 200237, PR China. Tel.: +86 21 64252022; fax: +86 21 64250624. E-mail address: [email protected] (F. Jiang). 0921-5093/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2008.01.033

is unnecessary for the whole components being made by magnesium matrix composites. So it is important to produce the partial and selective reinforcement on the surface of magnesium alloy components, e.g. improving the mechanical properties of magnesium alloy by means of surface engineering which without causing significant adverse effects on the properties of the base metal. At present, wide variety of coating techniques (e.g. CVD, PVD, laser beam composite surfacing, electron beam surface modification, magnetron sputtering and plasma spraying) have been used to deposit metal or ceramic on different substrates [9–11]. However, each of the coating techniques are limited by several main factors, such as weak interfaces between coating and substrate, need for vacuum chamber, extremely slow deposition, inconvenient operating and costly manufacturing procedure. Pulsed current gas tungsten arc (PC-GTA) surface engineering is suitable technique for improving the mechanical, tribological and chemical properties of metal surfaces. In the PC-GTA process, the arc interacts primarily with the substrate and the ceramic particles are simultaneously injected into the melt pool produced by the absorbed heat of arc. It is indicated that short processing time, flexibility in operation, economy

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

Fig. 1. Schematic illustration of the PC-GTA preparation process (reinforced with SiCp).

in time, low energy and material consumption and processing precision are the important advantages of PC-GTA surface engineering over the conventional processes. Metal matrix composite (MMC) surface layers with interesting properties and “very good” connection to the metal substrate may be prepared by a selection of suitable combination of metallic substrate and ceramic particles [12–14]. Performances of these surfacemodified composite coatings depend on their ability to withstand load and stress variations acting upon them when they are put into service. In the current study, the performance of ceramic (SiCp)reinforced coatings on Mg alloy AZ31 was determined using the four-point bend test. Fracture strength will be calculated from the load-deflection measurements carried out during the bend test. The surface-modified composite layer was subjected to drysliding wear tests on a pin-on-disc test apparatus. This study also examines the effect of SiCp volume fraction and particle size on the dry-sliding wear behavior. Performance of these coatings, therefore, manufactured by PC-GTA techniques depend on the particle size, shape and distribution of the reinforcing material as well as the processing parameters such as scan current, scan speed, pulsed current frequency and powder feed rate. Hence, studies regarding the effect of these parameters on the mechanical properties and on the microstructure of the resulting surfacemodified composite layer is also carried out in this present investigation. This study is aimed at optimizing the processing parameters to manufacture coatings with a desired performance. 2. Experimental procedures A new pulsed current GTA surface-modified process was used to fabricate the various particle size and volume fraction composite layer. Fig. 1 is the PC-GTA surface-modified experimental set-up schematic diagram. In the present study, rectangular plates of magnesium alloy AZ31 (rolled condition; 200 mm long, 200 mm wide and 10 mm thick; chemical composition (wt.%,): Al, 2.9; Zn, 0.8; Mn, 0.1; Si, 0.05; Cu, 0.0005; Fe, 0.0025; balance Mg, were used as substrates in the PC-GTA processing experiments. The AZ31 plate surface was polished by

209

abrasive paper to clear the oxide film, and using acetone to clean the organic substance on the surface. After drying, the specimens were assembled on a water-chilled copper rectangular plate. PCGTA surface modification was carried out with an auto-pulsed square-wave alternating current inert gas tungsten arc-welding machine by melting of AZ31 substrate and simultaneous feeding of SiCp powders (the average particle size was from 20 ␮m, 40 ␮m to 60 ␮m, respectively; purity was >98.0 wt.% SiC and <2.0 wt.% silicon). During the process, using Ar gas (purity was 99.999%, flow rate was maintained constantly at 10 L/min) to flow over the PC-GTA processing region in order to provide a relatively inert environment. Scanning speed was maintained at a certain value to control the substrate–GTA interaction time and the area coverage. To achieve microstructural and compositional homogeneity of the PC-GTA modified surface, a 20–30% overlap between the successive scanning beads was followed. Sufficient time was allowed for each scanning bead to reach room temperature before the subsequent PC-GTA operation was resumed to treat the adjacent scanning bead. The angle between the argon arc torch and the specimen was maintained at 60◦ . The electrode was 2% ThO2 tungsten, its diameter was 2.0 mm and the nozzle diameter of the argon arc torch was 10 mm. The arc length was maintained at 2 mm and the arc voltage was at 26 V constantly. The processing parameters were PC-GTA scanning current (I), scanning speed (v), current pulsing frequencies (f) and SiCp powder feed rate. In this study, the applied scanning current was 100–200 A, the chosen scanning speed was 150–300 mm/min, current pulsing frequencies was applied 3–12 Hz and SiCp powder feed rate was maintained constant at 5, 10 and 15 mg/s, respectively. Macrography of PC-GTA surface-modified composite layer given in Fig. 2 indicates that under the optimum processing parameters, there are no obvious welding defects such as porosities, arc craters, slag, etc, on the surface of PC-GTA surface-modified composite layer. The characterization techniques such as SEM, hardness, bend test and sliding wear test were used to measure the properties of the PC-GTA surface-modified composite layer. The samples for metallographic studies were polished on 800–2000-grif emery paper and etched in nital (volume concentration 4%) for about 30 s to reveal the microstructure evolved during PC-GTA processing. The specimens used in the four-point bend test consist of straight beams of rectangular cross-section (with dimensions of 60 mm long, 10 mm wide and the thickness ratio of composite layer/substrate was 1:1) cut from each block processed under different conditions. Since the thickness of the composite layer (hc ) under all the conditions is 0.9–1.5 mm, the selected dimensions of the specimen ensure that the final test sample consisted of both coating and substrate, so that it ensured enough volume of the composite layer in the material system to provide a clear signal for the load corresponding to initiation of the crack within the composite layer. The blocks were cut by means of wire-cutting in order to have better control on the dimensions. A universal material testing machine was used to perform the four-point bend test at a cross-head displacement rate of

210

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

Fig. 2. Macrography of PC-GTA surface-modified composite layer with SiCp on Mg alloy AZ31, scanned by I = 150 A, v = 200 mm/min, f = 8 Hz and powder feed rate of 15 mg/s.

1 mm/min. The schematic set-up for the four-point bend test is as shown in Fig. 3. The length for inner and outer spans was 20 and 40 mm, respectively, and the samples processed under different conditions were tested in a configuration to place the composite layer in tension. Load and displacement during the test were recorded by an in-built computer data-acquisition system. From the load–displacement plots, the initiation of the first crack can be found by observing the point on the plot where deviation from linearity occurs. At the room temperature (20 ◦ C), the wear resistance of as-received AZ31 surface and PC-GTA surface-modified composite layer was measured by a pin-on-disc wear testing machine against hardened quenching steel disc (GCr15, the average hardness was 720 HV; the inner diameter and external diameter was 20 and 25 mm, respectively). The applied load was 49, 98 and 147 N, respectively, and 100 rpm wheel speed for a wear period of 2.0 h under dry-sliding conditions. Each wear experiment was repeated at least three times. The weight loss during the wear tests was calculated from the weight differences of the pins before and after the tests to the accuracy nearest 0.1 mg. 3. Results and discussion 3.1. Microstructure in PC-GTA surface-modified composite layer 3.1.1. Grain refinement Fig. 4(a) shows the metallographic micrograph of the cross-section near the fusion line of the pulsed current GTA

Fig. 3. Schematic set-up for the four-point bend test.

surface-modified composite layer of AZ31 with SiCp. A “black line” was placed in Fig. 4(a) to emphasize the interface between the composite layer and the AZ31 alloy. Fig. 4(b) shows the metallographic micrograph of scanning bead of the GTA surface-modified composite layer, where predominantly fine equiaxed grains of primary magnesium embedded with SiCp. From Fig. 4(a) and (b), it can be seen that PCGTA surface-modified process has produced a crack/defect-free microstructure and notable grain refinement. It is need to be mentioned that pulse frequency plays an important role in generating the equiaxed grains. Among the investigated pulse frequencies between 3 and 12 Hz, pulsing frequencies of 6–8 Hz are found to be obviously effective in obtaining grain refinement. The pulsing of the welding current has several effects on the solidifying weld pool. It directly affects the temperature distribution, and the periodic variations of energy input into the weld pool cause thermal fluctuations, the nature of which depends on the pulsing conditions. The thermal cycles during pulsed current welding have been determined using analytical models of the heat flow conditions [15]. It has been shown that the thermal cycles oscillate around a curve for constant power conditions, with each pulse contributing to a temperature excursion [16]. The amplitude of these oscillations was found to increase with increasing ratio of the peak to base currents, and to decrease with rising pulse frequency. One consequence of the thermal fluctuations is the periodic interruption in the solidification process. As the pulsed current decays, the solid–liquid interface advances in towards the arc and increasingly becomes vulnerable to any disturbance in the arc form [17]. As the current increases again in the next pulse, growth is arrested and remelting of the growing dendrites can also occur. Current pulsing also results in additional fluid motion, which enhances the convective forces already existing in the weld pool. The periodic variations in the arc current result in similar changes in the arc forces impinging on the weld puddle which are proportional to the square of the welding current [18]. A further consequence of the induced flow patterns in the weld pool is that the thermal gradients in front of the solid–liquid interface are lowered [19]. The magnitude of this effect depends on the pulsing variables. On account of the temperature fluctuations inherent in pulsed welding, there is a continual change in the weld pool size and

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

211

Fig. 4. The metallographic micrograph of the cross-section (a) near the fusion line and (b) the weld bead of the pulsing current GTA surface melting layer of AZ31 with SiC, welded by I = 150 A, v = 200 mm/min, f = 8 Hz and powder feed rate of 15 mg/s.

shape. Thus, the direction of the maximum thermal gradient at the weld pool boundary also changes with time. As a result, instead of a few favourably oriented grains growing over long distances, newer grains become favourably oriented with respect to the instantaneous direction of maximum thermal gradient [19]. It is reasonable to expect that all these factors might significantly influence the solidification process. The grain refinement observed in pulsed welding in previous investigations has often been attributed to a mechanism involving dendrite fragmentation [20]. The cyclic temperature variations that occur at the solidification front due to the effect of the pulsed current can cause remelting and breaking-off of the growing dendrites. This is aided by the mechanical action of the weld pool turbulence in bringing the dendrite fragments ahead of the solid–liquid interface. These fragments then become sites for heterogeneous growth which eventually block the columnar growth process. In this study, pulsing frequency of 6–8 Hz was found to produce optimum results. At very low frequencies, the effect of

Fig. 5. TEM micrograph of the typical interface between SiCp and AZ31 matrix formed by GTA surface-modified AZ31 with SiCp, scanned by I = 150 A, v = 200 mm/min, f = 8Hz and powder feed rate of 15 mg/s.

succeeding pulses on a solidifying bead is only minimal. It might also be expected that the thermal and mechanical disturbances would be less intense under these conditions. On the other hand, at very high frequencies, the amplitudes of the vibrations induced in the weld pool and of the temperature oscillations [21] are considerably reduced. Thus, there exists an optimum frequency at which the greatest effect is produced. In the current investigation, the frequency of 6–8 Hz resulted in maximum grain refinement. It is likely that the enhanced fluid flow in pulsed welding results in more shallow thermal gradients and hence in a reduced cooling rate. 3.1.2. Interface of SiCp/AZ31 It is need to be mentioned that the interface between the ceramic particle and matrix plays a crucial role in determining the properties of the PC-GTA surface-modified composite layer. The transmission electron micrograph in Fig. 5 clearly shows the formation of a defect-free and adherent SiCp/AZ31 matrix interface formed in PC-GTA surface-modified composite layer. The SEM micrograph and the EDS line scans in Fig. 6 show the distribution of magnesium, aluminum, zinc, silicon and carbon across the SiCp and matrix in the PC-

Fig. 6. The SEM micrograph of SiCp and AZ31 matrix, and EDS line scans of the distribution of Mg, Al, Zn, Si and C across SiC particle and matrix in the PCGTA surface melting layer with SiCp (scanned by I = 150 A, v = 200 mm/min, f = 8 Hz and powder feed rate of 15 mg/s).

212

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

interface formed in the PC-GTA surface-modified composite layer.

3.1.3. Distribution and volume fraction of SiCp Fig. 7(a)–(c) show the scanning electron micrograph of the cross-section of PC-GTA surface-modified composite layer on AZ31 with SiCp (particle size was 40 ␮m), it indicates that under the same scanning speed and scanning current, the volume fraction of SiCp scanned with a higher powder feed rate of SiCp (Fig. 7(c) with 15 mg/s, approximately 12% vol.) was more than that of with a lower powder feed rate of SiCp (Fig. 7(b) with 10 mg/s, approximately 8% vol. and Fig. 7(a) with 5 mg/s, approximately 4% vol.). Similar results of SiCp volume fraction with particle size of 20 ␮m and 60 ␮m were obtained under the same process parameters. From the morphology of the surfacemodified composite layer, volume fraction and distribution of the SiCp were found to vary with powder feed rate of SiCp. Fig. 7 shows that under the proper GTA processing parameters, SiCp dispersed uniformly throughout the surface of magnesium alloy AZ31 matrix and there was no significant variation of distribution of SiCp with depth and width, and the agglomeration of SiCp is not obvious.

3.2. Mechanical performance of PC-GTA surface-modified composite layer

Fig. 7. The cross-section microstructure of PC-GTA surface-modified composite layer scanned with I = 150 A, v = 200 mm/min, f = 8 Hz, particle size of 40 ␮m and powder feed rate of (a) 5 mg/s, (b) 10 mg/s and (c) 15 mg/s.

GTA surface-modified composite layer. It can be seen that a homogeneous distribution of Mg, Al and Zn (except along the SiCp) throughout the matrix zone may be noted. Furthermore, the absence of Mg, Al, and Zn and the presence of only Si and C in the SiCp region confirmed that the particle was SiC with no melting or dissolution at the interface, i.e. there is no obvious segregation of Si and C atoms occurring at the interface, which indicates that apparently no reaction products were formed at the interface. It is concluded that magnesium wets SiCp well and it is a good “host” for the SiCp embedment and the formation of a defect-free and adherent SiCp/AZ31 matrix

3.2.1. Hardness The hardness values of PC-GTA surface-modified composite layer depend on the PC-GTA process parameters and the SiC particles powder concentration. At a certain scanning speed, Fig. 8(a) shows that the average hardness of the surface composite layer increases with increase in GTA scanning current. Because, the higher scanning current could more adequately melt the surface metal than the lower scanning current, so that it cause homogeneous distribution of SiC particles in the molten pool. However, too high GTA scanning current leads to decreased area fraction of SiC particles in the GTA surfacemodified composite layer and hence, hardness decreases. On the other hand, Fig. 8(b) indicates that average hardness of the PC-GTA surface-modified layer decreases with too high scanning speed, because higher scanning speed decreased particles inputting as a result of lowering of interaction time between arc and AZ31 matrix. Fig. 8(c) shows the hardness profile with scanning bead depth for PC-GTA surface modification of AZ31 with SiCp scanned with different PC-GTA processing parameters. Fig. 8(c) indicates that the hardness of the PC-GTA surfacemodified composite layer has been improved significantly from 1.18 to 1.47 GPa, as compared to average 0.54 GPa of the asreceived AZ31.

3.2.2. Bend test 3.2.2.1. Theoretical analysis of four-point bend test. The loading arrangement for the four-point bend test is as shown in Fig. 9. The maximum bending moment in a four-point bend test

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

213

Fig. 8. Variation of the average microhardness of the GTA surface-modified composite layer of SiC with (a) scanning current, (b) scanning speed and (c) microhardness profile along scanning bead depth with powder feed rate of 15 mg/s, pulse frequencies of 8 Hz and under different processing parameters.

where P, L, and l are the applied load, outer and inner loading span distances, respectively. The stress is maximum along the top and the bottom surfaces of the beam and is given as σ=

Fig. 9. Load arrangement for the four-point bend test.

is given by M=

P(L − l) 4

(1)

My It

(2)

where It is the moment of inertia for the beam and y represents the position of the neutral axis with respect to the top or the bottom surface of the beam. In present case, since the beam consists of the two layers, individual components of stress and moment of inertia for composite layer and the substrate have to be considered. Since the composite layer and the substrate are of different material having different Young’s moduli, there will be a shift in the neutral axis of the rectangular cross-section under bending load. Using the strength of materials approach, the actual crosssection can be transformed into an equivalent cross-section in terms of its two components. Fig. 10(a) shows the sketch of the cross-section of composite layer/substrate material, whereas Fig. 10(b) shows the transformed cross-section. The transformation depends only on the elastic modulus. Since in our case, the elastic modulus of composite layer is larger than the elastic modulus of the substrate, Es < Ec (the coating is composite in nature having a ceramic particulate reinforcement). In the transformed cross-section, the position of the neutral axis, y, and can

214

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

Fig. 10. (a) Original cross-section and (b) transformed cross-section (Ec /Es ) of the coating/substrate system.

be written with respect to the composite layer surface as follows: y=

h2 bc /2 + hs bs (hs /2 = hc ) Mc + Ms = c At h c bc + h s bs

(3)

where h and b are the height and width of the constituents, respectively, while the subscripts ‘c’ and ‘s’ stand for composite layer and substrate, respectively. All measurements and calculations are made with respect to this neutral axis. Since the neutral axis is not at the geometric center, a parallel axis theorem must be used to shift the moment of inertia, I of each constituent area to the neutral axis and, therefore, total moment of inertia It of the entire composite layer–substrate system is given by the following relationship:      hc 2 bc h3c 2 (Ii + Ai di ) = + h c bc y − It = 12 2 i=c,s    2  hc bs h3s (4) + h s bs y − h s − + 12 2 Thus, the maximum stress on the PC-GTA surface-modified composite layer surface is given as σ=

My P(L − l)y = It 4It

(5)

3.2.2.2. The effect of PC-GTA process parameters on bend strength. One of the objectives of the current investigation was to see if the observed grain refinement led to an improvement in PC-GTA surface-modified composite layer bend properties. For this purpose the unpulsed current welds, i.e. welds made with constant amplitude ac, were compared with those in which ac pulsed at optimum frequency was used. These results are listed in Table 1. The base material bend properties have also been included in this table for comparison. The bend strength data for each condition are an average of measurements made in three specimens. It can be seen from Table 1 that the unpulsed current GTA surface-modified composite layer exhibits a low tensile strength. The yield and ultimate bend strengths of pulsed current GTA surface-modified composite layer, on the other hand, are slightly higher than those of the as-received AZ31 Mg alloy, i.e. there is no degradation but reinforcing in bend strength of the PC-GTA surface-modified composite layer. It was probably due to the crack/defect-free microstructure with a notable grain refinement produced by PC-GTA surface modification process. It must be mentioned that the mechanical properties of the PC-GTA surface-modified composite layer are strongly influ-

enced by the nature of the reinforcement–matrix interface, which plays a significant role in the stress transfer and loadbearing characteristics of the composite layer. During the bend test, the hard SiC particles embedded in ductile AZ31 matrix constrained the deformation and resulted in local stress concentration in AZ31 matrix. As a result, at the Mg alloy AZ31 metal matrix–SiCp interface, the failure of fast fracture and cracking was promoted through the matrix of SiCp. This indicates that the interface between SiCp and AZ31 matrix in the PCGTA surface-modified composite layer does not considerably decrease the interfacial bond strength, i.e. the interfacial bond strength of the composite layer is not affected. On the other hand, Table 1 presents the variation in the bend strength values of the PC-GTA surface-modified composite layer with the change in the scan speed. It is observed that as the scan speed increases, the bend strength of the composite layer for individual alloys decreases. This variation in the bend strength of the composite layer can be well understood by considering the microscopic changes that occur with the change in the processing parameter. From Table 1, it can be inferred that, the crackfree and strongly adhered composite layer starts delaminating as the scan speed increases thus, indicating the development of a weak interface. In addition to the interfacial strength at the coating–substrate interface, the interfaces between the SiCp and matrix within the composite layer can provide potential sites for the crack initiation. Further propagation of these cracks under the applied loads can lead to lower strengths and ultimately fracture of the composite layers. The reduction in the bend strength of the composite layer with increasing scan speeds can be attributed to the reduced interaction time at higher processing speeds. At higher scan speed, there is lesser time of incidence for the sweeping arc and therefore, less energy density input for interaction between the substrate and the precursor [22]. Due to a shorter interaction time (less energy density input) there is no enough heat energy generated to cause the desirable melting and mixing of the SiCp into the substrate and this results into a poor interface between the SiCp and the matrix. Furthermore, weak bonding of the SiCp/AZ31 matrix decreases the overall strength of the composite layer and is proved to be detrimental for mechanical performance. 3.2.3. Effect of SiCp volume fraction and particle size on wear resistance At ambient temperature, Fig. 11 shows the variation of the wear rate (with the wear period of 2.0 h) of the PC-GTA surfacemodified composite layer against various volume fraction of

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

215

Table 1 Bend strength of the as-received AZ31 and PC-GTA surface-modified composite layer under different conditions (with I = 150 A, SiCp powder feed rate of 15 mg/s) Conditions

Yield bend strength (MPa)

Ultimate bend strength (MPa)

Elastic modulus (GPa)

As-received AZ31 f = 0 Hz, v = 200 mm/min f = 3 Hz, v = 200 mm/min f = 12 Hz, v = 200 mm/min f = 8 Hz, v = 200 mm/min f = 8 Hz, v = 150 mm/min f = 8Hz, v = 250 mm/min f = 8 Hz, v = 300 mm/min

163.2 171.6 178.9 180.3 189.7 181.5 164.4 139.4

406.4 411.2 424.1 429.6 442.3 426.7 409.5 389.1

41.2 42.4 43.1 43.6 45.3 41.9 41.3 37.5

SiCp (particle size of 20 ␮m) with the different applied load. This result indicates that wear rate of the composite layer increased with increasing applied load. The percentage improvement in wear resistance of composite layer with respect to the AZ31 matrix could be calculated from the measured wear rate values using following relation: WRc,m =

Wm − W c × 100% Wm

(6)

where WRc,m is the percentage improvement in wear resistance of PC-GTA surface-modified composite layer with respect to the AZ31 matrix, Wm is the wear rate of the AZ31 matrix and Wc is the wear rate of PC-GTA surface-modified composite layer. As shown in Fig. 11, wear rate of AZ31 matrix is higher than those of the PC-GTA surface-modified composite layer tested under the similar conditions. Wear rates of AZ31 matrix are 36.7, 72.1 and 154.5 mg at the loads of 49, 98 and 147 N, respectively. The increase in wear rate of AZ31matrix is respectively 96.5% and 114.3% with increasing applied load from 49 to 98 and 98 to 147 N. As shown in Fig. 11, the addition of 12 vol.% SiCp to the surface of AZ31 matrix reduced the wear rate to 6.2, 11.8 and 21.4 mg under the loads of 49, 98 and 147 N, respectively. The increase of the wear rate of 12 vol.% SiCp reinforced composite layer are respectively, 90.3% and 81.3%, with increasing load from 49 to 98 and 98 to 147 N, which are lower than the increase in wear rate of AZ31 matrix. The percentage improvement in wear resistance (WRc,m ) of 4, 8 vol.%

Fig. 11. Wear rate of the PC-GTA surface-modified composite layer against various volume fraction of SiCp (particle size of 20 ␮m) with the different applied load and the wear period of 2.0 h, at ambient temperature (20 ◦ C).

and 12 vol.% SiCp (particle size of 20 ␮m) reinforced composite layer is respectively, 32.1%, 57.8% and 86.1%, with the applied load of 98 N. It is noted that irrespective of applied load the wear rate of composite layer decreases linearly with volume fraction of SiCp. This clearly demonstrates that the wear resistance of composite layer is proportional to the volume fraction of SiCp. Fig. 12 indicates the effect of various SiC particle size (volume fraction of 12%) on the wear rate of the PC-GTA surface-modified composite layer with the different applied load. Fig. 12 shows that the improvement in wear resistance is a strong function of SiC particle size or granularity. The addition of 12 vol.% (particle size of 40 ␮m) SiCp to the surface of AZ31 matrix reduced the wear rate to 27.6, 55.9 and 119.3 mg under the loads of 49, 98 and 147 N, respectively. Under the applied load of 98 N, it is noted from Fig. 12 that the percentage improvement in wear resistance (WRc,m ) of 12 vol.% composite layer with SiC particle size of 20, 40 and 60 ␮m is approximately 83.6%, 22.4% and 10.2%, respectively, as compared to the as-received AZ31 matrix. Fig. 12 suggests that under the same wear applied load, due to reduction in SiC particle size by a factor of 1/2, there is an improvement in wear resistance (WRc,m ) by approximately four times. In the case of the PC-GTA surface-modified composite layer, it is expected that the SiCp will carry the major portion of the load applied on the surface of composite layer and the wear of the PC-GTA surface-modified composite layer is controlled

Fig. 12. Effect of various SiC particle size (volume fraction of 12%) on the wear rate of the PC-GTA surface-modified composite layer with the different applied load and the wear period of 2.0 h, at ambient temperature (20 ◦ C).

216

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

be given by the following equation: N=

π(D2 − d 2 )V AV = cG2 4cG2

(7)

where G is the average granularity or particle size of the reinforcement, c is a constant and it depends on the shape of the SiCp. Assuming that the total applied load (P) is being carried primarily by the SiCp and is shared equally by each of the SiCp, so the average effective load transferred to each of the SiCp will be calculated by the following equation: p= Fig. 13. Wear loss of specimens with various H/E value (particle size of 20 ␮m, applied load of 98 N, wear period of 2.0 h, at ambient temperature).

by the wear of the SiCp. The number of SiCp in wear area A (A = π(D2 − d2 )/4, d and D are the inner diameter and external diameter of the steel disc, respectively) will vary with the SiC particle size and SiCp volume fraction. If V presents the volume fraction of SiCp, and SiCp dispersed random and uniformly throughout the surface of PC-GTA surface-modified composite layer and there was no significant variation of distribution of SiCp with depth and width (as shown in Fig. 7), then the area covered by the SiCp contacted with steel disc could be calculated as AV. Thus, the number of SiCp (N) in wear area A could

P 4cG2 P = N π(D2 − d 2 )V

(8)

where p is the average effective applied load transferred to each of the SiCp. Eq. (3) qualitatively indicates that the average effective applied load on the SiCp varies inversely with the volume fraction of the SiCp and increases proportionately with square of the SiCp average granularity or particle size. If the wear rate of PC-GTA surface-modified composite layer is proportional to the average effective applied load on each of the SiCp, the relative wear resistance (RWR) [23] of one composite layer with respect to the other one as a function of SiC particles size and volume fraction could be calculated by the following equation:   V1 G 2 2 RWR1,2 = (9) V2 G1

Fig. 14. Worn surfaces of the PC-GTA surface-modified composite layer scanned with I = 150 A, v = 200 mm/min, f = 8 Hz, SiC particle size of 20 ␮m and powder feed rate of 15 mg/s, with the applied load of (a) 49 N, (b) 98 N, (c) 147 N and (d) worn surface of as-received AZ31 Mg alloy with the applied load of 147 N and wear period of 2.0 h, at ambient temperature (20 ◦ C).

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

217

Table 2 EDS analysis of the regions presented in Fig. 14 Region

Element Mg K

Al K

Zn K

OK

Wt.% At.%

34.602 46.062

2.153 2.237

0.562 0.245

19.586 22.412

Wt.% At.%

31.080 42.949

1.927 1.896

0.429 0.221

Wt.% At.%

36.918 45.434

1.956 2.014

Wt.% At.%

38.989 46.972

Wt.% At.%

Si K

CK

Fe K

Cr K

Mn K

Total%

6.021 7.925

5.313 7.033

30.536 13.614

0.631 0.213

0.596 0.259

A 100 100

21.734 24.613

7.156 9.137

4.124 6.203

32.412 14.535

0.723 0.245

0.415 0.201

B 100 100

0.834 0.486

24.134 26.971

7.315 10.364

4.732 5.395

23.245 8.901

0.429 0.207

0.437 0.228

C 100 100

2.876 3.634

0.491 0.218

26.736 29.421

4.697 6.057

3.127 4.875

21.589 8.236

0.983 0.353

0.512 0.234

D 100 100

38.037 47.420

2.654 3.205

1.314 0.728

12.819 15.649

8.256 11.504

7.135 8.979

28.431 11.902

0.591 0.219

0.763 0.394

E 100 100

Wt.% At.%

44.597 55.553

2.914 3.765

0.873 0.324

9.649 11.897

6.582 9.107

4.231 5.945

30.117 12.934

0.413 0.186

0.624 0.289

F 100 100

Wt.% At.%

22.914 40.657

2.175 3.264

0.293 0.117

10.586 13.937

8.941 14.094

6.853 8.103

46.753 19.104

0.871 0.453

0.614 0.271

G 100 100

Wt.% At.%

30.704 49.719

2.917 3.806

0.495 0.253

7.577 9.129

7.976 12.064

5.412 6.231

43.676 18.307

0.532 0.194

0.711 0.297

H 100 100

Wt.% At.%

48.731 51.509

2.765 3.542

0.773 0.297

6.941 8.657

18.712 25.324

6.354 7.937

14.917 5.362

0.231 0.134

0.576 0.238

100 100

Wt.% At.%

51.235 53.072

1.834 2.419

0.751 0.263

5.537 6.937

15.367 21.804

7.726 9.125

16.645 5.904

0.543 0.291

0.362 0.185

J 100 100

Wt.% At.%

55.493 56.287

2.245 2.936

0.816 0.417

8.134 9.931

13.286 18.934

5.397 6.783

13.583 4.235

0.411 0.183

0.635 0.294

K 100 100

Wt.% At.%

57.772 59.295

2.055 2.383

0.637 0.352

7.649 8.241

14.418 19.153

6.153 7.217

10.414 2.986

0.395 0.164

0.507 0.209

L 100 100

I

where RWR1,2 is the RWR of composite layer 1 with respect to that of composite layer 2; V1 and V2 are the volume fraction of SiCp in composite layer 1 and composite layer 2, respectively; and G1 and G2 are the granularity or size of SiCp in composite layer 1 and composite layer 2, respectively. Eq. (9) indicates that RWR of composite layer with respect to the other one is a linear function of relative volume fraction of SiCp and a function of inverse-square of SiC particle size provided the type of SiCp is same in all the cases. On the other hand, Eq. (9) indicates that if the volume fraction of the SiCp is kept constant and SiC particle size varies, the wear resistance is a function of inverse square of the SiC particle size. This is exactly noted in the present study in

Figs. 11 and 12. These results suggested that the improvement in wear resistance of PC-GTA surface-modified composite layer could be proportional to volume fraction of SiCp and inversely proportional to the square of SiCp average granularity or particle size. 3.2.4. Effects of the ratio of hardness to elastic modulus on the wear behavior The ratio of hardness to elastic modulus, H/E, could be proposed as one of the key parameters controlling wear [24]. A higher value of H/E means an improved resistance to scratching in the contact area during dry sliding [24–27]. In

218

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

the present study, the hardness and elastic modulus of PCGTA surface-modified composite layer were simultaneously enhanced by incorporating the hard SiC particles and grain refinement, which result the H/E value increase. For example, as shown in Fig. 13, the values of H/E for as-received AZ31 matrix, S1 (I = 150 A, v = 150 mm/min, f = 8Hz), I2 (I = 150 A, v = 250 mm/min, f = 8 Hz) and S3 (I = 150 A, v = 200 mm/min, f = 8 Hz) are 0.013, 0.028, 0.029, and 0.032, corresponding wear loss (the wear period of 2.0 h, at ambient temperature) were 72.1, 28.2, 24.3, and 11.8 mg, respectively, which is in good agreement with the gradually enhanced wear resistance of the PC-GTA surface-modified composite layer. This observation is consistent with the observations by Leyland and Matthews [28–30] although their conclusion was based on wear couples of different materials at interfaces. It is known that friction force consists of two parts: adhesion force and plowing force [31]. Since the plowing term is the force needed to deform materials in the direction of sliding, the plowing force is expected to be smaller for sinking-in than that for piling-up. Thus, the plowing contribution to friction is expected to decrease with increasing H/E, as seen in the present set of experiments. In general, wear of tribological coating can be categorized into two types: (1) wear caused by gradual removal of coating materials or (2) by cracking and delamination [32]. The present set of experiments unambiguously demonstrates the H/E effect on wear. The effect of H/E on wear can be rationalized by assuming that gradual material removal or gradual wear is caused by plastic deformation and not by elastic deformation or fracture. It follows that, first, for materials with high H/E, the “plasticity index” [33] which equals E/H multiplied by a factor determined by surface morphology, is small. Consequently, the deformation under contact is more likely to be elastic for high H/E materials [28]. Second, the ratio of reversible work, We , to total work, Wtot , under conical and pyramidal indentation has recently been shown to be proportional to H/E, i.e. We /Wtot ∝H/E [34–36]. Consequently, a larger fraction of the work is consumed in plastic deformation when H/E is smaller when the degree of deformation is the same. Third, the unrecoverable strain, measured by the ratio of final indentation depth, hf , to maximum indentation depth, hmax , is found to be related to H/E through a relationship that hf /hmax ∝Wtot − We /Wtot [34–36]. Thus, larger plastic strain is expected when contacting a material with smaller H/E. Assuming similar relationships hold for multiasperity contact during sliding, materials with higher H/E are expected to have smaller accumulative strain, smaller accumulative strain energy, and thus better wear resistance. 3.2.5. Wear mechanism of PC-GTA surface-modified composite layer The worn surfaces of the 12 vol.% SiCp (particle size of 20 ␮m) reinforced PC-GTA surface-modified composite layer after 2.0 h wear period and under the applied loads of 49, 98 and 147 N are given in Fig. 14(a)–(c). Fig. 14(a) shows the worn surface of composite layer worn under 49 N applied loads. It can be seen that no excessive damage is on the composite layer worn surface, and smooth and crater region in the worn surface of 12 vol.% SiCp reinforced composite layer worn under 49 N

applied load can be found. The EDS analysis (shown in Table 2) indicates that the crater region (regions A and B in Fig. 14(a) was found to have more Fe content than in the smooth surface, that because wear debris and particles having high Fe content were pushed into craters along the sliding direction and mixed to form the mechanically mixed layer. The EDS analysis results also means that SiCp scratch heavily the counterface of the steel disc and the hard SiCp bear the applied load. At the middle stage of formation of mechanically mixed layer, these Fe particles flowed plastically and mixed with composite debris [37]. On the other hand, a thin discontinues mechanically mixed layer (regions C and D in Fig. 14(a)) is seen on the smooth surface which has lower Fe content than crater region (regions A and B in Fig. 14(a)). It is reported that the tribolayer had high stability within the low load [38]. The microgrooves and plowing can be observed in these regions (near the regions C and D in Fig. 14(a)). During wear test, the local contact can also induce introducing oxygen into tribosystem and causing the oxidation of Fe and Mg. The EDS analysis of the regions A–D in Fig. 14(a) also confirms that there is high oxygen content in the worn surface (shown in Table 2). Iron oxide was known to have low friction coefficient and produce lubricating effect and it results in a lower wear rate and slow increase of wear rate with applied load [39,40]. From the analysis mentioned above, we can confirm that the oxidation type of wear is dominant mechanism under the wear condition of 49 N applied load. Fig. 14(b) shows that the smooth and crater regions in the worn surface of 12 vol.% SiCp reinforced Mg alloy AZ31 PCGTA surface-modified composite layer worn under 98 N applied load. The EDS analysis (shown in Table 2) indicates that a thick discontinuous mechanically mixed layer (regions E and F in Fig. 14(b)) is seen on smooth surface which has lower Fe content than the crater region (regions G and H in Fig. 14(b)). But this mechanically mixed layer has higher Fe content than mechanically mixed layer generated under applied load of 49 N (regions C and D in Fig. 14(a)). It is reported that the thickness of mechanically mixed layer increases with increasing applied load and there were cracks and microvoids in Fe-rich sublayers [41,42]. Fig. 14(b) also shows that the worn surface of PC-GTA surface-modified composite layer worn under applied load of 98 N exhibits the cracks. These cracks in this region would give rise to delamination wear. So we can conclude that the delamination is dominant wear mechanism under the wear condition of 98 N applied load. Fig. 14(c) indicates that big breaking and no extruded lips are presented at the worn surface of 12 vol.% SiCp reinforced Mg alloy AZ31 PC-GTA surface-modified composite layer specimen worn under applied load of 147 N. EDS analysis (shown in Table 2) shows that worn surface (regions I and J in Fig. 14(c)) has higher Mg, Si content and lower Fe content than that of specimen worn under applied load of 98 N (regions G and H in Fig. 14(b)). A similar result is found that the relative smooth surface (regions K and L in Fig. 14(c)) also has higher Mg, Si and lower Fe content than that of specimen worn under applied load of 98 N (regions E and F in Fig. 14(b)). The EDS analysis result indicates that the SiCp fragmentation and massive AZ31 matrix materials removal from the composite layer took place,

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

which means that SiCp lost their abrasive effect against steel disc. For this reason, this specimen has the lowest Fe content on the worn surface. Therefore, the SiCp lost their load-bearing capability and the wear rate of PC-GTA surface-modified composite layer increased with increasing applied load. Although 12 vol.% SiCp reinforced composite layer exhibited better wear resistance than the unreinforced AZ31 matrix materials, a transition from delamination to adhesion is observed with increasing applied load from 98 to 147 N. So it can be concluded that the dominant wear mechanism is adhesion under the applied load of 147 N. Fig. 14(d) shows the worn surface of Mg alloy AZ31matrix worn under applied load of 147 N. The wide and deep grooves from where matrix materials have been removed as debris are seen on the worn surface of specimen. It can be seen as an extensive plastic deformation and the extruded lips on the worn surface of the Mg alloy AZ31 matrix specimen. There are also dimples or crater generated from delamination wear on the worn surface of specimen. It can be confirmed that the mechanism of the AZ31 matrix material is severe adhesive wear and partly delamination under the wear condition of 147 N applied load. 4. Conclusions SiCp can be deposited and forms high-performance composite layer on the surface of magnesium alloy by the PC-GTA surface modification processing. Under the optimum processing parameters (I = 150 A, v = 200 mm/ min, f = 8 Hz and powder feed rate of 15 mg/s), SiCp dispersed uniformly throughout the AZ31 surface composite layer, and PC-GTA surface-modified process produced a notable grain refinement. The strong, adherent and clean interface of SiCP/AZ31 matrix was achieved under the optimum processing parameters. The hardness values of PC-GTA surface-modified composite layer depend on the PC-GTA process parameters and the SiC particles powder concentration. The hardness of the PC-GTA surface-modified composite layer has been improved significantly from 1.18 to 1.47 GPa, as compared to average 0.54 GPa of the as-received AZ31. Too high GTA scanning current leads to decreased area fraction of SiC particles in the GTA surfacemodified composite layer and hence, hardness decreases. On the other hand, the hardness of the PC-GTA surface-modified layer decreases at higher scanning speeds. Under the optimum processing parameters, the bend yield and ultimate bend strength values of PC-GTA surface-modified composite layer are slightly higher than those of the as-received AZ31 Mg alloy. The inverse relationship between the bend strength of the composite layer and scan speed was observed. When the scan speed increases, the bend strength of the coating decreases. During the bend test, the hard SiC particles embedded in ductile AZ31 matrix constrained the deformation and resulted in local stress concentration in AZ31 matrix. As a result, at the AZ31 matrix–SiCp interface, the failure of fast fracture and cracking was promoted through the matrix of SiCp. This indicates that the interface between SiCp and AZ31 matrix in the PC-GTA surface-modified composite layer does not consid-

219

erably decrease the interfacial bond strength, i.e. the interfacial bond in the composite layer keeps strong. Wear rate of PC-GTA surface-modified composite layer increased with increasing applied load and decreased with increasing reinforcement particle volume fraction. The wear resistance increases linearly with increase in SiCp volume fraction and decreases with increase in SiCp particle size. The RWR is proportional to inverse-square of the size of SiCp. Although the hardness and elastic modulus of PC-GTA surface-modified composite layer were simultaneously enhanced by incorporating the hard SiC particles, the H/E value increases with increasing SiCp content. The values of H/E for as-received AZ31 matrix, S1 , S2 and S3 are 0.013, 0.028, 0.029 and 0.032, respectively, which is in good agreement with the gradually enhanced wear resistance of the composite layer. The wear mechanism of PCGTA surface-modified composite layer exhibited the transition from oxidation, delamination to adhesive under the wear applied load of 49, 98 and 147 N, respectively. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]

[19] [20] [21] [22] [23]

[24] [25] [26] [27]

A. Stern, A. Munitz, J. Mater. Sci. Lett. 18 (1999) 853–856. M. Marya, G.R. Edwards, S. Liu, Weld. J. 83 (2004) 203–212. Q.C. Jiang, X.L. Li, H.Y. Wang, Scripta Mater. 48 (2003) 713–717. J.H. Gu, X.N. Zhang, Y.F. Qiu, M.Y. Gu, Compos. Sci. Technol. 65 (2005) 1736–1742. H. Watanabe, T. Mukai, M. Mabuchi, K. Higashi, Acta Mater. 49 (2001) 2027–2037. H.Z. Ye, X.Y. Liu, J. Alloys Compd. 402 (2005) 162–169. Y.Q. Wang, M.Y. Zheng, K. Wu, Mater. Lett. 59 (2005) 1727–1731. S.Y. Chang, H. Tezuka, A. Kamio, Mater. Trans. 38 (1997) 526–535. E. Gariboldi, J. Mater. Process. Technol. 134 (2003) 287–295. M.H. Lee, I.Y. Bae, K.J. Kim, K.M. Moon, T. Oki, Surf. Coat. Technol. 169 (2003) 670–674. G. Garces, M.C. Cristina, M. Torralba, P. Adeva, J. Alloys. Compd. 309 (2000) 229–238. J.D. Ayers, T.R. Tucker, Thin Solid Films 73 (1980) 201–207. A.B. Kloosterman, B.J. Kooi, J.Th.M. De Hosson, Acta Mater. 46 (1998) 6205–6217. D. Pantelis, A. Tissandier, P. Manolatos, O. Ponthiaux, Mater. Sci. Technol. 11 (1995) 299–303. A. Grill, Metall. Trans. 12B (3) (1981) 187–192. P.R. Vishnu, W.B. Li, K.E. Easterling, Mater. Sci. Technol. 7 (7) (1991) 649–659. J.G. Garland, Met. Const. Br. Weld. J. 6 (4) (1974) 121–127. A.A. Gokhale, A.A. Tzavaras, H.D. Brody, G.M. Ecer, in: G.J. Abbaschian, S.A. David (Eds.), Grain Refinement in Castings and Welds, TMS-AIME, Warrendale, PA, 1983, p. 223. G.M. Reddy, A.A. Gokhale, K.P. Rao, J. Mater. Sci. 32 (1997) 4117–4126. T. Shinoda, Y. Ueno, I. Masumoto, Trans. JWS 4 (1990) 18–23. H. Yamamoto, S. Harada, T. Ueyama, S. Ogawa, F. Matsuda, K. Nakata, Weld. Int. 7 (6) (1993) 456–461. A. Agarwal, N.B. Dabotre, Int. J. Refract. Met. H. Mater. 17 (1999) 283–293. I.M. Hutchings, S. Wilson, A.T. Alpas, Comprehensive Composite Materials (Surface Modified Composite), vol. 3, Elsevier Science Ltd., UK, 2000, p. 501. W.Y. Ni, Y.T. Cheng, M.J. Lukitsch, A.M. Weiner, L.C. Lev, D.S. Grummon, Appl. Phys. Lett. 85 (18) (2004) 4028–4030. K.H.Z. Gahr, Tribol. Int. 31 (1998) 587–596. Y.J. He, A.J. Winnubst, D.J. Schipper, P.M.V. Bakker, A.J. Burggraaf, H. Verweij, Wear 184 (1995) 33–43. C.F. Hu, Y.C. Zhou, Y.W. Bao, D.T. Wan, J. Am. Ceram. Soc. 89 (11) (2006) 3456–3461.

220 [28] [29] [30] [31]

S. Zhang et al. / Materials Science and Engineering A 490 (2008) 208–220

T.L. Oberle, J. Met. 3 (1951) 438–446. A. Leyland, A. Matthews, Wear 246 (2000) 1–11. J. Jin, H. Wang, Acta Metall. Sin. 24 (1988) A66–A71. I.M. Hutchings, Tribology: Friction and Wear of Engineering Materials, CRC, Boca Raton/Arnold/London, 1992. [32] S. Hogmark, S. Jacobson, M. Larsson, Wear 246 (2000) 20–33. [33] J.A. Greenwood, J.B.P. Williamson, Proc. R. Soc. London Ser. A 295 (1966) 300–319. [34] Y.T. Cheng, Z. Li, C.M. Cheng, Philos. Mag. A 82 (2002) 1821–1829.

[35] Y.T. Cheng, C.M. Cheng, Appl. Phys. Lett. 73 (1998) 614–616. [36] W. Ni, Y.T. Cheng, C.M. Cheng, D.S. Grummon, J. Mater. Res. 19 (2004) 149–157. [37] D. Lu, M. Gu, Z. Shi, Tribol. Lett. 6 (1999) 57–61. [38] R.L. Deuis, C. Subramarian, Compos. Sci. Technol. 57 (1997) 415–435. [39] X.L. Li, K.N. Tandon, Wear 225 (1999) 640–648. [40] M. Gui, S.B. Kang, J.M. Lee, Mater. Sci. Eng. A 293 (2000) 146–156. [41] M. Gui, S.B. Kang, J.M. Lee, Wear 240 (2000) 186–198. [42] X.L. Li, K.N. Tandon, Wear 245 (2000) 148–161.