Microstructure and Wear Behaviors of Plasma-Sprayed MoAlB Ceramic Coating

MoAlB ceramic coatings were prepared on a 316 steel surface by atmospheric plasma spraying with different arc power levels. The phase composition, microstructure and wear resistance of coatings against GCr15 and Si3N4 counterparts were studied. The MoAlB ceramic decomposed and was oxidized to form MoB and Al2O3 during plasma spraying. With the increase of the arc power, MoAlB experienced more decomposition, but the coatings became denser. When the arc power increased from 30 to 36 kW, the wear rates of coatings against GCr15 and Si3N4 balls reduced by 91% and 78%, respectively. The characterization of wear tracks shows that when against GCr15 counterparts, the main wear mechanisms are abrasive and adhesive wear, and when against Si3N4 counterparts, fatigue and abrasive wear are dominant. The refinement of wear resistance by increasing arc power can be attributed to the improvement of density and adhesive strength


Introduction
316 stainless steel (316L) has been broadly used in mechanical components for its excellent mechanical properties and cost effectiveness [1,2]. However, the wear failure of components, which leads to great economic losses, is the main mechanical failure mode during service. Therefore, how to ameliorate the wear resistance of mechanical parts is an urgent problem. Ceramic coatings have been extensively studied as wearresistant layers, such as WC-WCoB, WC-Cr 3 C 2 -Ni, NiCrBSi-Zr and YSZ coatings [3][4][5][6], which greatly ameliorated the friction performance of the metallic part. However, the wear resistance of these coatings is severely challenged by adverse industrial conditions. Therefore, researchers need to explore coatings with excellent tribological performance.
MoAlB ceramics are potential wear-resistant coatings for metal surfaces due to their outstanding mechanical properties [7][8][9] and wear performance, which have attracted much interest from researchers. For example, dense MoAlB ceramics prepared by highpressure sintering had excellent wear resistance, especially when the counterpart material was stainless steel [10,11]. This is due to the oxide film consisting of MoO 3 , B 2 O 3 and Fe 2 O 3 formed on the friction surface, which has a good lubrication and protection effect, contributing to reducing the wear. The worn surface was more prone to oxidation and formed oxide film at 600 • C, resulting in better wear resistance. Al matrix composites reinforced with MoAlB ceramic were prepared, and their tribological properties were investigated [12]. The results show that with the addition of MoAlB ceramic, the tribological properties improved greatly. Most notably, the wear rate of Al composites with 15 vol.% MoAlB decreased by 95% compared with pure Al. Tan et al. [13] also researched the friction performance of Mo-12Si-8.5B reinforced with MoAlB at room temperature (RT) (1000 • C).

Fabrication of Feedstock Powder and Coatings
The schematic of the preparation of MoAlB powder and coatings is exhibited below in Figure 1. In this study, commercially available molybdenum (Mo, <2 µm, 99.5%, Rhawn, Shanghai, China), aluminum (Al, <30 µm, 99%, Rhawn, Shanghai, China) and boride (B, <20 µm, 99%, Rhawn, Shanghai, China) powders were used as raw materials. The Mo, Al and B powders were mixed in zirconia jars with a molar ratio of 1:1.5:1.3 [29] using a planetary ball mill (YXQM-2L, Changsha Miqi Instrument Equipment Co., Ltd., Changsha, China). Ball milling lasted 10 h, the rotation rate was 300 rpm and alcohol was used as the dispersing agent. The micrographs of raw powders were characterized by a scanning electron microscope (SEM, S-3400N, Hitachi Incorporation, Tokyo, Japan) and are shown in Figure 2. B powder was not characterized because of its strong oxidizability.
ing is formed during the continuous deposition of splats [16,17]. Compared to other coating techniques, APS technology has high deposition efficiency. Therefore, APS is an appropriate technique for coating MoAlB powder. The performance of plasma-sprayed coatings is related to the size of the feedstock, the pretreatment of the substrate and spraying parameters. The effect of the size of the feedstock, arc power, spraying distance and the flow rate of secondary gas on the microstructure, porosity and bonding strength of Ni/Al composite coatings was investigated by Zhang et al. [18]. The increase of power can enhance the density and bonding strength of the coating but results in phase transformation. The pretreatment of the substrate surface before plasma spraying is to increase the roughness and reduce the stress by sandblasting and preheating, respectively, which can enhance the bonding strength of the coating and the substrate. Otherwise, the coating is prone to cracking and chipping [19]. Moreover, previous studies have confirmed that MoAlB ceramic incongruently melts into MoB and Al2O3 at temperatures above 1435 °C [20][21][22]. MoB and Al2O3 show high hardness and excellent wear resistance [23][24][25][26], which is expected to further improve the wear resistance. Therefore, we expect that the MoAlB coating prepared by APS will improve the durability of various tribological pairs, such as the friction surfaces of vehicle components, shafts, sleeves and tools [27,28].
In this study, APS was used to fabricate MoAlB ceramic coatings on 316L substrates, and the optimum processing parameters were identified. The microstructure and wear behavior of MoAlB coatings against GCr15 and Si3N4 counterparts were investigated. Furthermore, the wear mechanisms were discussed in great detail.

Fabrication of Feedstock Powder and Coatings
The schematic of the preparation of MoAlB powder and coatings is exhibited below in Figure 1. In this study, commercially available molybdenum (Mo, <2 μm, 99.5%, Rhawn, Shanghai, China), aluminum (Al, <30 μm, 99%, Rhawn, Shanghai, China) and boride (B, <20 μm, 99%, Rhawn, Shanghai, China) powders were used as raw materials. The Mo, Al and B powders were mixed in zirconia jars with a molar ratio of 1:1.5:1.3 [29] using a planetary ball mill (YXQM-2L, Changsha Miqi Instrument Equipment Co., Ltd., Changsha, China). Ball milling lasted 10 h, the rotation rate was 300 rpm and alcohol was used as the dispersing agent. The micrographs of raw powders were characterized by a scanning electron microscope (SEM, S-3400N, Hitachi Incorporation, Tokyo, Japan) and are shown in Figure 2. B powder was not characterized because of its strong oxidizability.    The mixture was dried at 100 °C for 12 h in a vacuum oven. powder was obtained and then heated in an Al2O3 crucible und [30][31][32] for 6 h using an atmosphere furnace (AF1400, Kunshan Ltd., Suzhou, China) to form MoAlB ceramic. The block ceramics at a speed of 300 rpm for 1 h into MoAlB powder. In order to imp the MoAlB powder was granulated using an aqueous solution of (PVA) then dried in a vacuum oven at 100 °C for 5 h. Finally, th size of 35-75 μm was sieved out using 200 and 400 mesh sieves f The MoAlB coatings were deposited on 316L substrates (diam = 5 mm) using APS (XM-80SK, Shanghai Xiuma Spraying machin China). The substrates were sand-blasted and preheated before high mechanical bonding between the coatings and the substrate 500, 550 and 600 A were selected to study the effect of arc power o were labeled as C-500, C-550 and C-600, respectively. Table 1 sum parameters.  The mixture was dried at 100 • C for 12 h in a vacuum oven. After drying, the mixed powder was obtained and then heated in an Al 2 O 3 crucible under flowing Ar at 1100 • C [30][31][32] for 6 h using an atmosphere furnace (AF1400, Kunshan Aikexun Machinery Co., Ltd., Suzhou, China) to form MoAlB ceramic. The block ceramics were crushed and milled at a speed of 300 rpm for 1 h into MoAlB powder. In order to improve the flowability [33], the MoAlB powder was granulated using an aqueous solution of 5 wt.% polyvinyl alcohol (PVA) then dried in a vacuum oven at 100 • C for 5 h. Finally, the powder with a particle size of 35-75 µm was sieved out using 200 and 400 mesh sieves for APS.
The MoAlB coatings were deposited on 316L substrates (diameter = 25 mm, thickness = 5 mm) using APS (XM-80SK, Shanghai Xiuma Spraying machinery company, Shanghai, China). The substrates were sand-blasted and preheated before spraying to realize the high mechanical bonding between the coatings and the substrates [34][35][36]. The currents of 500, 550 and 600 A were selected to study the effect of arc power on coatings [17,37], which were labeled as C-500, C-550 and C-600, respectively. Table 1 summarizes the deposition parameters.

Tribological Testing
The tribological tests were conducted under dry sliding conditions using ball-on-disk friction and a wear tester (HT-1000, Zhongke Kaihua Science and Technology Development Ltd., Lanzhou, China) at RT and ambient humidity. During the wear test, the coated specimen was fixed on a rotating tray, and the abrasive ball was pressed against the coating surface by applied load, as illustrated in Figure 3. The Si 3 N 4 ceramic balls and GCr15 steel balls with a diameter of 6 mm were employed as the counterpart balls [10,11,13]. Before each test, the coating was polished and then cleaned in anhydrous ethanol. The applied force, rotation diameter and sliding speed were 5 N, 6 mm and 0.17 m/s, respectively. The friction time of each parameter was 15 min and conducted three times to guarantee the accuracy of the results.

Characterizations
The phase composition of the feedstock powders and the three spected using X-ray diffraction (XRD, D8 Advance, Bruker Incorporati many). The XRD was performed at the range of 10°-80° with a speed crostructure of the powder, coatings and worn surfaces were examined with an energy-dispersive X-ray spectrometer (EDS, AZtecOne, Hit Tokyo, Japan). The porosities of the coatings were measured by ima the Image J software (Image J, v1.8.0, National Institutes of Health, Be The cross-section area of the wear tracks was examined by the con microscope (keyencevk-x200, Keyence Co., Ltd., Osaka, Japan). The w ings was calculated from W = V/(F × S), where V is the wear volume (m load (N) and S is the total sliding distance (m) [38].

Characterizations
The phase composition of the feedstock powders and the three coatings were inspected using X-ray diffraction (XRD, D8 Advance, Bruker Incorporation, Karlsruhe, Germany). The XRD was performed at the range of 10-80 • with a speed of 5 • /min. The microstructure of the powder, coatings and worn surfaces were examined by SEM combined with an energy-dispersive X-ray spectrometer (EDS, AZtecOne, Hitachi Incorporation, Tokyo, Japan). The porosities of the coatings were measured by imaging method using the Image J software (Image J, v1.8.0, National Institutes of Health, Bethesda, MD, USA).
The cross-section area of the wear tracks was examined by the confocal laser scanning microscope (keyencevk-x200, Keyence Co., Ltd., Osaka, Japan). The wear rate of the coatings was calculated from W = V/(F × S), where V is the wear volume (mm 3 ), F is the applied load (N) and S is the total sliding distance (m) [38].

Characterizations
The phase composition of the feedstock powders and the thre spected using X-ray diffraction (XRD, D8 Advance, Bruker Incorporat many). The XRD was performed at the range of 10°-80° with a speed crostructure of the powder, coatings and worn surfaces were examine with an energy-dispersive X-ray spectrometer (EDS, AZtecOne, Hit Tokyo, Japan). The porosities of the coatings were measured by ima the Image J software (Image J, v1.8.0, National Institutes of Health, Be The cross-section area of the wear tracks was examined by the con microscope (keyencevk-x200, Keyence Co., Ltd., Osaka, Japan). The w ings was calculated from W = V/(F × S), where V is the wear volume (m load (N) and S is the total sliding distance (m) [38].     Figure 6 displays the XRD patterns of the MoAlB coatings deposited with differen powers. The patterns reveal that MoAlB, MoB, Mo and Al2O3 phases were present in the three coatings, which indicates that the MoAlB ceramic partially decomposed and trans formed into MoB and Al2O3 due to the higher temperature at the center position of the plasma than that at the edge [21,22]. However, the peak intensity of MoB in C-600 was significantly higher than those of C-500 and C-550, indicating that with the increase o power, the decomposition of MoAlB accelerated fast. The Al2O3 phase was composed o stable α-Al2O3, metastable γ-Al2O3 and amorphous Al2O3 in the three coatings. This is in good agreement with the results reported by Misra et al. [40,41]. The content of γ-Al2O3 in the coatings rose with the arc power, which indicates that more MoAlB decomposed and suffered a rapid solidification during the deposition process [42]. Surface morphologies of the deposited MoAlB ceramic coatings with different spray ing parameters are shown in Figure 7. The surfaces of C-500 and C-600 were mainly com posed of unmolten or semimolten particles, with little completely molten powder, and the powder spreading was poor. Therefore, there were a large number of holes and pores  Figure 6 displays the XRD patterns of the MoAlB coatings deposited with different powers. The patterns reveal that MoAlB, MoB, Mo and Al 2 O 3 phases were present in the three coatings, which indicates that the MoAlB ceramic partially decomposed and transformed into MoB and Al 2 O 3 due to the higher temperature at the center position of the plasma than that at the edge [21,22]. However, the peak intensity of MoB in C-600 was significantly higher than those of C-500 and C-550, indicating that with the increase of power, the decomposition of MoAlB accelerated fast. The Al 2 O 3 phase was composed of stable α-Al 2 O 3 , metastable γ-Al 2 O 3 and amorphous Al 2 O 3 in the three coatings. This is in good agreement with the results reported by Misra et al. [40,41]. The content of γ-Al 2 O 3 in the coatings rose with the arc power, which indicates that more MoAlB decomposed and suffered a rapid solidification during the deposition process [42].  Figure 6 displays the XRD patterns of the MoAlB coatings deposited with differen powers. The patterns reveal that MoAlB, MoB, Mo and Al2O3 phases were present in the three coatings, which indicates that the MoAlB ceramic partially decomposed and trans formed into MoB and Al2O3 due to the higher temperature at the center position of the plasma than that at the edge [21,22]. However, the peak intensity of MoB in C-600 was significantly higher than those of C-500 and C-550, indicating that with the increase o power, the decomposition of MoAlB accelerated fast. The Al2O3 phase was composed o stable α-Al2O3, metastable γ-Al2O3 and amorphous Al2O3 in the three coatings. This is in good agreement with the results reported by Misra et al. [40,41]. The content of γ-Al2O3 in the coatings rose with the arc power, which indicates that more MoAlB decomposed and suffered a rapid solidification during the deposition process [42]. Surface morphologies of the deposited MoAlB ceramic coatings with different spray ing parameters are shown in Figure 7. The surfaces of C-500 and C-600 were mainly com posed of unmolten or semimolten particles, with little completely molten powder, and the powder spreading was poor. Therefore, there were a large number of holes and pores formed in the coatings. The surface of C-600 showed full powder spreading with fewer Surface morphologies of the deposited MoAlB ceramic coatings with different spraying parameters are shown in Figure 7. The surfaces of C-500 and C-600 were mainly composed of unmolten or semimolten particles, with little completely molten powder, Coatings 2021, 11, 474 6 of 13 and the powder spreading was poor. Therefore, there were a large number of holes and pores formed in the coatings. The surface of C-600 showed full powder spreading with fewer semimolten particles, and the pores were smaller and fewer. These result in a denser coating with a stronger bonding strength [40,43,44].

Microstructure of the Plasma-Sprayed MoAlB Coatings
Coatings 2021, 11, x FOR PEER REVIEW semimolten particles, and the pores were smaller and fewer. These result in a den ing with a stronger bonding strength [40,43,44]. Cross-sectional images of the MoAlB coatings sprayed with different arc po shown in Figure 8. Coatings with a thickness of about 100 ± 8.3 μm were well bond substrates without evident concentrated cracks in the interfaces. Deep pores melted particles can be observed in C-500 (Figure 8a,b). The formation of pore account for the inadequate spreading during deposition and the spallation of spla course of polishing. Moreover, microcracks can be found in C-500 as a result of th stress generated during deposition due to the inconsistent melting state of parti number of pores in C-550 significantly reduced, but unmelted particles and mic were still present. C-600 shows a typical lamellar structure (Figure 8e,f) due to th deformation of the molten and partly molten powder during the deposition proc deformation of droplets and the bonding between splats improved with the powe is beneficial to the formation of a denser coating [45]. There were few pores in Ca porosity of 1.94% compared to 9.26% of C-500 and 6.42% of C-550. This corre well with the morphology of the surface of C-600 presented in Figure 7e,f. Cross-sectional images of the MoAlB coatings sprayed with different arc power are shown in Figure 8. Coatings with a thickness of about 100 ± 8.3 µm were well bonded with substrates without evident concentrated cracks in the interfaces. Deep pores and unmelted particles can be observed in C-500 (Figure 8a,b). The formation of pores might account for the inadequate spreading during deposition and the spallation of splats in the course of polishing. Moreover, microcracks can be found in C-500 as a result of the tensile stress generated during deposition due to the inconsistent melting state of particles. The number of pores in C-550 significantly reduced, but unmelted particles and microcracks were still present. C-600 shows a typical lamellar structure (Figure 8e,f) due to the heavy deformation of the molten and partly molten powder during the deposition process. The deformation of droplets and the bonding between splats improved with the power, which is beneficial to the formation of a denser coating [45]. There were few pores in C-600 with a porosity of 1.94% compared to 9.26% of C-500 and 6.42% of C-550. This corresponded well with the morphology of the surface of C-600 presented in Figure 7e,f.
In order to observe the phase distribution in the coatings, the cross-section of C-600 was characterized by backscattered electron (BSE) imaging. BSE images of the cross-section of C-600 are presented in Figure 9  In order to observe the phase distribution in the coatings, the cross-section was characterized by backscattered electron (BSE) imaging. BSE images of the c tion of C-600 are presented in Figure 9, which reveal the presence of four distinc nent contrasts in the coatings. According to the average atomic number of MoA Mo and Al2O3 (20, 23.5, 42 and 10, respectively) and the XRD patterns ( Figure 6), inferred that the areas with decreasing brightness in the BSE images correspon MoB, MoAlB and Al2O3, respectively. The MoB and MoAlB phases showed a structure with Al2O3 and some pores distributed around it.    In order to observe the phase distribution in the coatings, the cross-section was characterized by backscattered electron (BSE) imaging. BSE images of the cr tion of C-600 are presented in Figure 9, which reveal the presence of four distinct nent contrasts in the coatings. According to the average atomic number of MoAl Mo and Al2O3 (20, 23.5, 42 and 10, respectively) and the XRD patterns ( Figure 6), inferred that the areas with decreasing brightness in the BSE images correspond MoB, MoAlB and Al2O3, respectively. The MoB and MoAlB phases showed a structure with Al2O3 and some pores distributed around it.  Figure 10a,b displays the variation of the friction coefficients of various coatin sliding time against GCr15 and Si3N4 balls, respectively. The friction coefficients coatings reached a relatively stable stage after a short period of running-in. Th friction coefficients against GCr15 varied (Figure 10a) in the range of 0.4-0.5. H the friction coefficient curve of C-600 was steadier than those of C-500 and C-55 pores and unevenly distributed unmolten and partly molten particles in the caused the fluctuation of the friction coefficient [46]. In contrast, C-600 is dense phases are uniformly distributed in lamellar structure [33], so the friction coefficie of C-600 was steadier. Figure 10b exhibits the friction coefficient curves of the thr ings against Si3N4 counterparts. The average friction coefficients decreased with power, and C-600 presented the lowest average friction coefficient of 0.46. The  Figure 10a,b displays the variation of the friction coefficients of various coatings with sliding time against GCr15 and Si 3 N 4 balls, respectively. The friction coefficients of both coatings reached a relatively stable stage after a short period of running-in. The stable friction coefficients against GCr15 varied (Figure 10a) in the range of 0.4-0.5. However, the friction coefficient curve of C-600 was steadier than those of C-500 and C-550. Large pores and unevenly distributed unmolten and partly molten particles in the coatings caused the fluctuation of the friction coefficient [46]. In contrast, C-600 is dense and the phases are uniformly distributed in lamellar structure [33], so the friction coefficient curve of C-600 was steadier. Figure 10b exhibits the friction coefficient curves of the three coatings against Si 3 N 4 counterparts. The average friction coefficients decreased with the arc power, and C-600 presented the lowest average friction coefficient of 0.46. The friction coefficient curve of C-500 continued to rise after the running-in period and eventually stabilized at about 0.7.   Figure 11 displays the profile curves of wear tracks of the three coatings against GCr15 and Si3N4 balls, respectively. It is obvious from Figure 11 that the depth of wear tracks gradually decreased with the arc power, and C-600 showed the lowest depth and width. The wear rates of the MoAlB coatings against GCr15 and Si3N4 are presented in Figure 12. With the increase of power, the wear rate shows a significant trend of decrease. The wear rates of C-600 against GCr15 and Si3N4 reach minimum values of 26.30 × 10 −5 mm 3 •N −1 m −1 and 42.89 × 10 −5 mm 3 •N −1 m −1 , respectively. These results reveal that the wear resistance of the MoAlB coatings increased with the arc power. It can be attributed to the dense coating and high hardness and wear resistance of the decomposition phases. Moreover, the tribological behaviors of C-500 and C-550 against Si3N4 counterparts were superior to those against GCr15 balls. However, the wear rate of C-600 against Si3N4 was higher than that against GCr15, which corresponds well with the results of previous studies [10,11].   Figure 11 displays the profile curves of wear tracks of the three coatings against GCr15 and Si 3 N 4 balls, respectively. It is obvious from Figure 11 that the depth of wear tracks gradually decreased with the arc power, and C-600 showed the lowest depth and width. The wear rates of the MoAlB coatings against GCr15 and Si 3 N 4 are presented in Figure 12. With the increase of power, the wear rate shows a significant trend of decrease. The wear rates of C-600 against GCr15 and Si 3 N 4 reach minimum values of 26.30 × 10 −5 mm 3 ·N −1 m −1 and 42.89 × 10 −5 mm 3 ·N −1 m −1 , respectively. These results reveal that the wear resistance of the MoAlB coatings increased with the arc power. It can be attributed to the dense coating and high hardness and wear resistance of the decomposition phases. Moreover, the tribological behaviors of C-500 and C-550 against Si 3 N 4 counterparts were superior to those against GCr15 balls. However, the wear rate of C-600 against Si 3 N 4 was higher than that against GCr15, which corresponds well with the results of previous studies [10,11].  Figure 11 displays the profile curves of wear tracks of the three coatings against GCr15 and Si3N4 balls, respectively. It is obvious from Figure 11 that the depth of wear tracks gradually decreased with the arc power, and C-600 showed the lowest depth and width. The wear rates of the MoAlB coatings against GCr15 and Si3N4 are presented in Figure 12. With the increase of power, the wear rate shows a significant trend of decrease. The wear rates of C-600 against GCr15 and Si3N4 reach minimum values of 26.30 × 10 −5 mm 3 •N −1 m −1 and 42.89 × 10 −5 mm 3 •N −1 m −1 , respectively. These results reveal that the wear resistance of the MoAlB coatings increased with the arc power. It can be attributed to the dense coating and high hardness and wear resistance of the decomposition phases. Moreover, the tribological behaviors of C-500 and C-550 against Si3N4 counterparts were superior to those against GCr15 balls. However, the wear rate of C-600 against Si3N4 was higher than that against GCr15, which corresponds well with the results of previous studies [10,11].   Surface images of the wear tracks of MoAlB coatings against GCr15 are sh Figure 13. A lot of debris and small areas of the smooth surface were found on t tracks of C-500 and C-550, as shown in Figure 13a,c. Moreover, the high conten which was transferred from the GCr15 counterparts to the coatings, was found worn surface (Table 2). It is reasonable to conclude that the wear mechanism of C-C-550 against GCr15 was mainly severe abrasive and adhesive wear. From the hig nification image (Figure 13b,d), it can be seen that there were more cracks and p the smooth area of C-500 than that of C-550. However, the worn surface of C-600 13e,f) displays larger areas of the smooth surface with some grooves, which is signi different from C-500 and C-550. Moreover, the smooth areas consisted of Motribo-oxidative production ( Table 2), which greatly improved the wear resistan 600. Therefore, the wear mechanisms were transformed from severe abrasive an sive wear to slight abrasive and adhesive wear as the power increased from 30 and to 36 kW. The differences of worn surfaces were possibly attributed to the differen ing strength between splats caused by the inconsistent melting and deformatio powder. Severe abrasive wear of C-500 and C-550 was mainly caused by the wea facial bonding strength between the splats and high porosity. The particles in the c are easily dislodged to form debris, which aggravated the wear [33]. When the ar improved, the deposited coatings were denser, and the bonding strength betwee cles was stronger, which could effectively protect the substrate from abrasion wea over, the content of MoB and Al2O3 rose with the arc power, and their high hardn wear resistance contribute a lot to the enhancement of the wear resistance of the coatings. Surface images of the wear tracks of MoAlB coatings against GCr15 are shown in Figure 13. A lot of debris and small areas of the smooth surface were found on the wear tracks of C-500 and C-550, as shown in Figure 13a,c. Moreover, the high content of Fe, which was transferred from the GCr15 counterparts to the coatings, was found on the worn surface ( Table 2). It is reasonable to conclude that the wear mechanism of C-500 and C-550 against GCr15 was mainly severe abrasive and adhesive wear. From the high-magnification image (Figure 13b,d), it can be seen that there were more cracks and pores on the smooth area of C-500 than that of C-550. However, the worn surface of C-600 (Figure 13e,f) displays larger areas of the smooth surface with some grooves, which is significantly different from C-500 and C-550. Moreover, the smooth areas consisted of Mo-Fe-Cr-O tribo-oxidative production ( Table 2), which greatly improved the wear resistance of C-600. Therefore, the wear mechanisms were transformed from severe abrasive and adhesive wear to slight abrasive and adhesive wear as the power increased from 30 and 33 kW to 36 kW. The differences of worn surfaces were possibly attributed to the different bonding strength between splats caused by the inconsistent melting and deformation of the powder. Severe abrasive wear of C-500 and C-550 was mainly caused by the weak interfacial bonding strength between the splats and high porosity. The particles in the coatings are easily dislodged to form debris, which aggravated the wear [33]. When the arc power improved, the deposited coatings were denser, and the bonding strength between particles was stronger, which could effectively protect the substrate from abrasion wear. Moreover, the content of MoB and Al 2 O 3 rose with the arc power, and their high hardness and wear resistance contribute a lot to the enhancement of the wear resistance of the MoAlB coatings.   The wear track morphologies of the three coatings against Si3N4 are significantly different from that against GCr15, as shown in Figure 14. The worn surfaces of the three coatings (Figure 14a,c,e) display delamination and fatigue cracks [47], which is the representative fatigue wear. Owing to the high hardness of Si3N4 balls, some wear debris and particles were compacted to a form sheet attached to the worn surface, which retards the progress of wear. Moreover, from the EDS analysis results, it can be seen that there was a small amount of Si element on the worn surfaces (Table 3). However, more debris was observed from the worn surface of C-500 and C-550 (Figure 14b,d) than that of C-600 (Figure 14f). It is reasonable to conclude that the wear mechanism of MoAlB coatings against Si3N4 was mainly abrasive and fatigue wear [14]. The wear track morphologies of the three coatings against Si 3 N 4 are significantly different from that against GCr15, as shown in Figure 14. The worn surfaces of the three coatings (Figure 14a,c,e) display delamination and fatigue cracks [47], which is the representative fatigue wear. Owing to the high hardness of Si 3 N 4 balls, some wear debris and particles were compacted to a form sheet attached to the worn surface, which retards the progress of wear. Moreover, from the EDS analysis results, it can be seen that there was a small amount of Si element on the worn surfaces (Table 3). However, more debris was observed from the worn surface of C-500 and C-550 (Figure 14b,d) than that of C-600 ( Figure 14f). It is reasonable to conclude that the wear mechanism of MoAlB coatings against Si 3 N 4 was mainly abrasive and fatigue wear [14].

Conclusions
MoAlB coatings were deposited with different arc powers by APS technique. The phase composition, microstructure and wear behavior of the coatings were studied. The following conclusions were obtained: 1.
The sprayed MoAlB coatings with different arc power levels consisted of MoAlB, MoB and Al 2 O 3 phases. The MoAlB powder partly decomposed and was oxidized to form MoB and Al 2 O 3 . With the increase of power, the decomposition of MoAlB increased, and the compactness of the coatings and the bonding strength of splats increased greatly.

2.
When MoAlB coatings slid against GCr15, the main wear mechanisms were changed from severe abrasive and adhesive wear to slight abrasive and adhesive wear as the arc power increased from 30 and 33 kW to 36 kW. When the arc power increased to 36 kW, the wear rate was the lowest (26.30 × 10 −5 mm 3 ·N −1 m −1 ).

3.
The main wear mechanism of MoAlB coatings against Si 3 N 4 was abrasive and fatigue wear. The friction coefficient decreased with the arc power, and C-600 showed the lowest wear rate and presented excellent wear resistance. Moreover, MoB and Al 2 O 3 with high hardness greatly improved the wear resistance of the MoAlB coatings.