Sci Eng Compos Mater 2017; 24(2): 195–202

Aiguo Liu*, Da Li, Fanling Meng and Huanhuan Sun -matrix composite fabricated with gas arc melt injection and precoated with NiCrBSi alloy to increase the volume fraction of WC particles

DOI 10.1515/secm-2014-0221 investigation of Liyanage et al. [1] on plasma transferred Received July 12, 2014; accepted June 1, 2015; previously published arc welded Ni-WC overlays shows that the MMC overlays online September 5, 2015 are from two to five times more wear resistant than the matrix alloys without WC particles. Abstract: The volume fraction, dissolution, and segrega- Many welding and cladding methods have been used tion of WC particles in metal-matrix composites (MMCs) to produce WC particle-reinforced MMCs on low-cost are critical to their wear resistance. Low carbon steel substrates. However, there are still some problems to be substrates were precoated with NiCrBSi coatings and solved. The first problem is the degree of dissolution of processed with gas tungsten arc melt injection method to WC particles. In the process of welding or cladding, WC fabricate MMCs with high volume fraction of WC particles. particles are heated by a heat source and dissolve into The microstructures and wear resistance of the compos- the molten pool partially or even completely. Katsich and ites were investigated. The results showed that the volume Badisch [2] have accessed the effect of carbide degrada- fraction of WC particles increased with decreasing hopper tion in a WC/W C-reinforced Ni-based hard facing, and the height and was as high as 44% when hopper height was 2 results show significant carbide degradation with increas- 100 mm. The dissolution of WC particles was minimal. The ing welding current, resulting in a significant reduced content of the alloying elements decreased from the top to primary carbide content and carbide diameter. Reduced the bottom of the matrix. More WC particles dissolved in carbide content indicated a significant wear rate increase the overlapping area, where Fe W C carbide blocks could 3 3 under pure three-body abrasion conditions. Dissolution be found. The wear loss of the MMCs after 40 min was results in lowered wear resistance, as there are less WC 6.9 mg, which is 76 times less than that of the substrate particles that remained to provide wear protection. The after the 4 min test. contents of carbon and tungsten in the matrix increase and the toughness of the matrix decreases, so the risk of Keywords: gas tungsten arc melt injection; metal-matrix cracking increases. The extra carbon and tungsten precip- composite; NiCrBSi alloy coating; volume fraction; WC itate as M C-type carbides, which are less wear resistant particle. 6 than WC particles. The second problem is the volume fraction of WC particles in the matrix. Jankauskas et al. [3] have inves- 1 Introduction tigated the effect of WC grain size and content on low stress abrasive wear. They have found that the wear rate The wear resistance of metal-matrix composites (MMCs) of hard facing decreases with the increase in WC content, with remarkable content of hard phases with very and a factor of 9 has been achieved when WC content is high hardness, such as carbides, is well known. The 42–43 wt%. A further increase in WC content is generally considered difficult because of the problem of WC particle dissolution. *Corresponding author: Aiguo Liu, School of Materials Science and The third problem is the distribution of WC particles Engineering, Shenyang Ligong University, Shenyang 110159, in the matrix. Fernández et al. [4] have studied the tribo- P.R. China, e-mail: [email protected] logical improvement of NiCrBSi laser cladding coating Fanling Meng and Huanhuan Sun: School of Materials Science and Engineering, Shenyang Ligong University, Shenyang 110159, reinforced with different weight percentages of WC parti- P.R. China cles. They have found that WC particles tend to precipitate Da Li: Cixi Guanghua Industrial Co. Ltd., Cixi 315326, P.R. China at the bottom of the melt coating; hence, the percentage 196 A. Liu et al.: Gas tungsten arc melt injection of WC particles of carbides increases with the depth of the coating. The to lower the surface tension, to allow more WC particles to of WC particles is much higher than that of the be incorporated into the MMCs. The microstructures and matrix. They are prone to sink down at the bottom of wear resistance of the MMCs were reported. the molten pool. The tendency of segregation of WC par- ticles is related to the time they spend in the molten pool, which is determined by the procedure and process para­ meters. With a certain procedure and fixed parameters, little 2 Materials and methods can be done to control the distribution of WC particles in the matrix. The segregation of WC particles decreases the The substrate was of Q235 low carbon steel (similar to ASTM wear resistance of the top layer of the MMC and increases A570 Gr.A; Anshan Steel Co., Ltd., Anshan, China). The the cracking tendency of the MMC/substrate interface. chemical composition and mechanical properties of the To solve the problem of dissolution and segregation substrate are shown in Table 1. The dimensions of the sub- of WC particles, Vreeling et al. [5] proposed a laser melt strate were 250 × 50 × 5 mm3. The NiCrBSi alloy powder used injection process. In the laser melt injection process, a for precoating was manufactured by VEHA Co., Ltd. (Keo- laser beam is used to melt the substrate to form a molten rekovsk, Russia), a Russian company that manufactures pool. With the movement of the laser beam, a tail is thermal spray equipment and materials. The morphology formed behind the molten pool, and the carbide parti- of the NiCrBSi powder particles is shown in Figure 1. The cles are injected into the tail of the molten pool, avoiding nominal diameters of the particles were 5–45 μm. The com- direct interaction with the laser beam. With this method, position of the NiCrBSi powder is shown in Table 2. The they have strengthened alloy with WC particles injected particles were cast and crushed WC-8% Co parti- [5] and aluminum alloy with SiC particles [6]. Zhao et al. cles (Harbin Welding Institute, Harbin, China). The size of [7] use a plasma arc to melt and inject WC particles into the WC-8% Co particles was 350–700 μm. low carbon steel substrates. Other carbide particles, such The substrate was degreased, dried, and grit blasted. as Cr3C2-NiCr [8] and SiC [9], have also been successfully The NiCrBSi powder was flame sprayed onto the substrate. incorporated into overlays with little dissolution by the The coating was 1 mm thick. melt and injection processes. The schematic diagram of the GTAMI system is shown However, the volume fraction of the incorporated in Figure 2. The coated substrate was fixed on a movable carbide particles is still limited. It is found that only the platform. The moving velocity of the platform was adjust- particles with velocity higher than a critical velocity vc can able. A gas tungsten arc torch was set above the platform overcome the surface tension and enter the molten pool with an angle α to the normal of the platform. The sub- [6]. To increase the volume fraction of the carbide particles strate with coating was heated by the arc, and a molten in the MMCs, the velocity of the injected particles must be pool formed under the arc. The platform moved horizon- increased or the surface tension of the molten pool must tally with a velocity of 4 mm/s. The molten pool left a tail be decreased. NiCrBSi alloys, which have low melting behind the arc. The injection nozzle was fixed together point, low surface tension when melted, and excellent with the torch with an angle β to the normal of the plat- wear resistance, are often sprayed or cladded on indus- form. WC particles were stored in a hopper. The height trial components for wear protection [10]. If they are pre- of the hopper was adjustable. WC particles were driven coated on the substrate, the surface tension of the molten out of the hopper by a roller and were accelerated down pool will be greatly decreased. Therefore, the volume frac- through the pipe and out the injection nozzle by gravity. tion of WC particles in the MMCs can be increased. WC particles left the nozzle with a certain velocity, entered In this paper, WC particle-reinforced MMCs were fab- the tail of the molten pool, and were caught in the matrix ricated on Q235 low carbon steel substrates with the gas after the solidification of the molten pool. The parameters tungsten arc melt injection (GTAMI) process. NiCrBSi alloy of the GTAMI process are shown in Table 3. Single-pass was deposited on the substrate before melt and injection, specimens with different hopper heights and multipass

Table 1: Chemical composition and mechanical properties of Q235 steel.

Chemical composition (wt%) Mechanical properties

C Si Mn P S Fe σs (MPa) σb (MPa) Elongation (%) 0.14–0.22 0.30 0.3–0.65 ≤ 0.045 ≤ 0.05 Bal. 235 375–500 26 A. Liu et al.: Gas tungsten arc melt injection of WC particles 197

(5 g FeCl3, 25 ml HCl, 25 ml ethanol). The microstructures of the MMCs were investigated with a Hitachi model S-570 scanning electron microscopy (SEM). The compositions of different phases were analyzed with a Tracor North- ern model TN-5502 energy-dispersive X-ray spectroscopy (EDS). The phases in the MMCs were analyzed with a model D/max-rB X-ray diffraction (XRD) analyzer (Rigaku, Japan). The wear resistance of the MMCs was tested with a M-200 wear tester (Jinan Yihua Testing Equipment Co., Ltd., Jinan, China). The specimen was sectioned Figure 1: Morphology of NiCrBSi powder particles. to 20 × 6 × 5 mm3 and ground to obtain a flat surface for

testing. The counterpart was a Al2O3 ring with a diameter Table 2: Composition of NiCrBSi alloy (wt%). of 500 mm. The rotation rate of the ring was 200 rpm. The load on the ring was 490 N. The wear loss of the specimen Cr Fe Si B Ni was weighed every 20 min in the test with a CHI604C elec- 14.0–15.0 3.0–3.5 2.5–3.0 1.8–2.2 Bal. tric balance with precision of 0.1 mg (Shanghai Chenhua Co., Ltd., Shanghai, China). The substrate was also tested for comparison. The wear loss of the substrate was weighed every minute.

3 Results and discussion

Figure 3 shows the cross-sections of specimens prepared with different hopper height H. Both the precoated NiCrBSi alloy and the substrate were melted and mixed together, forming the matrix of the MMCs. Many irregular-shaped WC particles were found embedded in the matrix. The edges of WC particles were sharp. The distribution of WC particles in the matrix of the specimen with H = 100 mm (Figure 3A) Figure 2: Schematic diagram of the GTAMI system. was more uniform than those with larger H (Figure 3B–D). When H = 400 mm, almost all WC particles were in the Table 3: Parameters of the GTAMI process. lower part of the MMCs. All MMCs were crack free. The volume fraction of WC particles shown in Figure 3 Melting current (A) 60 was estimated with Image-Pro Plus and is shown in Flow rate of shielding gas (l/min) 7–8 Figure 4. Figure 4 shows that the volume fraction of WC Velocity of the substrate (mm/s) 4 particles increased with the decrease in H in the tested Feeding rate of WC particles (mg/s) 280 Height of hopper H (mm) 100 range. Compared with the MMCs produced with plasma 200 arc melt injection process [11], the volume fraction of WC 300 particles of the MMCs produced with the GTAMI process 400 and precoated NiCrBSi alloy was greatly increased. Diameter of the injection nozzle (mm) 3.5 The increase in the volume fraction of WC particles Tilt angle of the gas tungsten arc torch α (°) -15 Tilt angle of the injection nozzle β (°) 15 in the MMCs was caused by precoating the substrate with NiCrBSi alloy. WC particles were driven out of the hopper by the roller at height H and were accelerated down specimens were prepared. The overlap of the multipass through the pipe by gravity. The velocities of these parti- specimen was 3 mm. cles were within a certain range when they arrived at the The specimens were sectioned, mounted, ground, surface of the molten pool. The mean velocity and distri- polished, and etched. The etchant was FeCl3 solution bution of velocities of the particles depended on H and 198 A. Liu et al.: Gas tungsten arc melt injection of WC particles

A B

C D

Figure 3: Cross-sections of the specimens prepared with different hopper heights: (A) H = 100 mm, (B) H = 200 mm, (C) H = 300 mm, and (D) H = 400 mm.

the injection system. When a certain WC particle arrived at the surface of the molten pool, whether it could enter the molten pool depended on its velocity. Only when its velocity was higher than a critical velocity v could it over-

) c come the surface tension to enter the molten pool and was embedded in the matrix after the solidification of the

molten pool. The critical velocity vc is related to the radius and density of the carbide particles and the surface ten- olume fraction (% V sions between phases. It is determined by Vreeling et al. [6] as

3 v =+()σσ−σ (1) cl2σρR vlppv lv Hopper height (mm) where R is the radius of the particle (m), ρ is the density 3 Figure 4: Influence of hopper height H on the volume fraction of WC of the particle (kg/m ), σlv is the surface tension between particles in the MMCs. the liquid and the vapor (N/m), σlp is the surface tension A. Liu et al.: Gas tungsten arc melt injection of WC particles 199

between the liquid and the particle (N/m), and σpv is the A surface tension between the particle and the vapor (N/m). From Equation (1), it was easy to know that the decrease in σlv and σlp would result in a lower critical veloc- ity vc, which meant that the velocities of more WC parti- cles would exceed vc and would enter the molten pool. NiCrBSi is a self-fluxing alloy powder used for cladding and hard facing. The surface tension of its melt was much lower than that of the liquid steel. Therefore, precoating the Q235 substrate with NiCrBSi alloy powder resulted in a low surface tension of the molten pool, and more WC particles entered the molten pool. When the hopper was moved higher, the distance from the tip of the nozzle to the substrate increased. WC particles leaving the nozzle scattered to a larger area, and less WC particles arrived at the molten pool. There- fore, the volume fraction of WC particles in the MMCs B decreased with increasing hopper distance (Figure 4). The mean velocity of WC particles increased when the hopper was moved higher, so WC particles entered deeper into the molten pool. With the increase in hopper height H, the segregation of WC particles was achieved. When H = 400 mm, the segregation of WC particles became very evident, as shown in Figure 3D. Limited by the structure of the system, H could not be adjusted to < 100 mm. Results with shorter H could not be obtained, but it was reason- able to guess a shallower distribution of WC particles in the MMCs. Figure 5 shows the XRD patterns of the single-pass MMCs. Besides WC particles, two other phases, Fe-Ni solid solution and Fe3W3C carbide, were found in the MMCs. The precoated NiCrBSi alloy was melted and mixed with the substrate, forming the Fe-Ni solid solution phase. This C was the matrix of the MMCs. The XRD patterns showed the Intensity (cps)

2θ (°) Figure 6: Microstructures of the single-pass specimen in the vicin- ity of a WC particle at the (A) top, (B) middle, and (C) bottom parts of Figure 5: XRD patterns of the single-pass MMCs. the MMCs. 200 A. Liu et al.: Gas tungsten arc melt injection of WC particles

Table 4: EDS results of locations shown in Figure 6. were mixed with the iron from the substrate to form the matrix ( could not be detected with the EDS analysis). Location W Fe Ni Si Cr The NiCrBSi coating was on the top of the substrate, so the A1 11.1 67.1 15.7 2.1 4.0 contents of , chromium, and silicon decreased with A2 7.5 75.0 12.1 1.6 3.8 distance from the surface of the MMCs. The distribution A3 2.8 81.3 11.3 1.2 3.4 of tungsten was the same as that of other elements, but B1 2.3 71.3 24.1 0.9 1.4 the mechanism was different. When a WC particle entered B2 1.0 75.8 21.4 0.6 1.2 B3 0.8 79.4 17.7 0.4 1.7 the molten pool, it was heated by the liquid metal and C1 33.1 41.5 15.7 6.5 3.2 began to dissolve. The dissolution process continued until C2 31.2 42.8 17.2 6.0 2.8 the particle was caught by the solidifying matrix metal. C3 34.5 41.2 16.3 4.2 3.8 The particle found at the bottom part must get through the top part of the molten pool. This meant that the dis- solution of the particles at the bottom part of the matrix

Fe3W3C phase in the MMCs, which meant dissolution of had a contribution to the tungsten content at the top part, WC particles. However, SEM images showed that the dis- but the dissolution of the particles at the top part of the solution of WC particles was not serious. Most of the WC matrix had no contribution to the tungsten content at the particles retained their original shapes (Figure 3). Only bottom part. Therefore, the tungsten content decreased some surface of WC particles dissolved into the molten with distance from the surface of the MMCs. This kind of pool and precipitated as Fe3W3C carbides. distribution of the alloying elements was beneficial to the Figure 6 shows the microstructures of the single- mechanical property of the MMCs. The ratio of phase C pass specimen. Besides WC particles, three other kinds in Figure 6A–C to the matrix (phase A+phase B) was cal- of microstructures were found in the matrix, marked as culated with Image-Pro Plus as 25.5%, 22.1%, and 14.7%, phases A–C, respectively. Phase B looked a little darker respectively. This trend was the same as the distribution of than phase A in the SEM image. Phase C was a white her- elemental tungsten. ringbone structure. Both phases A and C were located at The microstructures of the multipass specimen are the grain boundary of phase B. shown in Figure 7. Figure 7A is an overview of the over- The EDS results of these three microstructures are lapping area of the first and second passes. To investigate shown in Table 4. The EDS analysis results showed that the influence of the second pass, the boundary between both phases A and B were Fe-Ni solid solutions. Although the first and second passed must be determined. The local there were more alloying elements (tungsten, silicon, enlargement of area 1 in Figure 7A is shown in Figure 7B. and chromium) in phase A (comparing the amount of the The white belt consisted of herringbone structure. A alloying elements at A1 and B1, A2 and B2, and A3 and columnar crystal was found growing from the white belt B3, as shown in Table 4), there were less alloying ele- towards the second pass. This suggested that the right ments in phase B, and its melting point must be higher. edge of the white belt was the boundary between the first In the solidifying process, phase B solidified first. With and second passes, as indicated by the dashed line in less chromium content, its corrosion resistance was worse Figure 7A and B. The white belt was located at the heat- than phase A. Therefore, it looked darker in the SEM image affected zone of the second pass, and its shape fit the after etching. There were more alloying elements in phase fusion boundary of the second pass. This meant that the A, and it solidified after phase B in the solidifying process herringbone structure was formed in the melting process at the grain boundary of phase B and looked a little whiter of the second pass. in the SEM image. After the solidification of the Fe-Ni solid The dashed line in Figure 7A crossed the particle solution, more alloying elements were left in the remain- marked P. This suggested that particle P was injected into ing liquid metal, and the liquid metal solidified as eutectic the molten pool of the first pass in the melt and injec- herringbone structure. tion processes of the first pass, and its top right corner For the same kind of Fe-Ni solid solution, such as was reheated by the arc in the melt and injection pro- phase A, the contents of nickel, tungsten, silicon, and cesses of the second pass. The local enlargement of area chromium decreased from the top part to the bottom part 2 in Figure 7A is shown in Figure 7C. For comparison, the of the matrix (comparing the amount of the alloying ele- local enlargement of the bottom left corner of particle P, ments at A1, A2, and A3, as shown in Table 4). The pre- area 3 in Figure 7A, is shown in Figure 7D. The bottom left coated NiCrBSi coating and the substrate were melted and corner of particle P was not reheated by the arc, and the mixed when heated. Nickel, chromium, boron, and silicon microstructures in its vicinity were the same as the ones A. Liu et al.: Gas tungsten arc melt injection of WC particles 201

A B

C D

Figure 7: Microstructures of the multipass specimen: (A) overview of the overlapping area of the first and second passes, (B) local enlarge- ment of area 1 in (A), (C) local enlargement of area 2 in (A), and (D) local enlargement of area 3 in (A).

in the single-pass specimen. Although the top right corner Table 5: EDS results of locations shown in Figure 7C. of particle P was reheated by the arc, more WC particles dissolved into the liquid metal. The EDS results of dif- Location W Fe Ni Si Cr ferent microstructures in Figure 7C are shown in Table 5. A4 15.6 59.8 20.1 1.8 2.7 Comparing Table 5 with Table 4, the tungsten content B4 12.8 59.3 22.2 2.9 2.8 at point A4 was much higher than that at points A1, A2, C4 14.6 57.7 22.4 2.6 2.7 D4 60.2 29.3 9.4 0.4 0.7 and A3, and it was also higher for point B4. This was the result of more WC dissolution. The difference of the tung- sten content between points A4 and B4 was less than that provided the matrix with more tungsten and carbon, and between points A1 and B1. The same was true for other ele- when the tungsten content was high, Fe3W3C was more ments. The remelting provided more time for the elements likely to precipitate other than the herringbone structure. to distribute uniformly. A herringbone structure was not The wear loss of the specimen produced with fully developed, so the alloying element contents at point H = 100 mm and the substrate versus time are shown C4 were very close to those at point A4. Around the WC in Figure 8. The wear test of the substrate could not be particle, some carbide blocks with high tungsten content carried on after 4 min because of severe wear. The wear appeared. Judged by its appearance and tungsten content, loss of the substrate after 4 min was 524.9 mg. The wear it was believed to be Fe3W3C. The remelting of WC particles loss of the MMCs after 40 min test was only 6.9 mg. It could 202 A. Liu et al.: Gas tungsten arc melt injection of WC particles

as herringbone structure. The contents of the alloying ele- ments in the matrix decreased from the top to the bottom of the matrix. WC particles in the overlapping area of the successive passes were reheated by the following melting

and dissolved more. Fe3W3C carbide blocks could be found in the overlapping area. More herringbone structure appeared in the previous pass, at the heat-affected zone of ear loss (mg )

W the following pass. The wear loss of the MMCs after 40 min was 6.9 mg, 76 times less than that of the substrate after the 4 min test.

Acknowledgments: The research work was supported by Time (min) Open Project Fund of Key Disciplines of Liaoning Province (Project Number 4771004kfx05). Figure 8: Wear loss of the specimen produced with H = 100 mm and the substrate versus time.

References be concluded that the wear resistance of the substrate was greatly improved with WC particle-reinforced MMCs pro- [1] Liyanage T, Fisher G, Gerlich AP. Wear 2012, 274–275, 345–354. duced by the GTAMI process. [2] Katsich C, Badisch E. Surf. Coat. Technol. 2011, 206, 1062–1068. [3] Jankauskas V, Antonov M, Varnauskas V, Skirkus R, ­Goljandin D. Wear 2015, 328–329, 378–390. 4 Conclusions [4] Fernández MR, García A, Cuetos JM, González R, Noriega A, Cadenas M. Wear 2015, 324–325, 80–89. MMCs were fabricated using the GTAMI process, with WC [5] Vreeling JA, Ocelík V, De Hosson JTM. Acta Mater. 2002, 50, 4913–4924. particles injected into low carbon steel Q235 substrate [6] Vreeling JA, Ocelík V, Pei YT, Agterveld DTLV, De Hosson JTM. precoated with NiCrBSi alloy. The volume fraction of WC Acta Mater. 2000, 48, 4225–4233. particles in the MMCs was found to be much higher than [7] Zhao M, Liu A, Guo M, Liu D, Wang Z, Wang C. Surf. Coat. Tech- that in the MMCs produced with plasma arc injection nol. 2006, 201, 1655–1659. process, without precoated coatings. The volume fraction [8] Liu A, Guo M, Hu H, Li Z. Scr. Mater. 2008, 59, 231–234. [9] Guo M, Liu A, Zhao M, Hu H, Wang Z. Surf. Coat. Technol. 2008, increased with decreasing hopper height and was as high 202, 4041–4046. as 44% when the hopper height was 100 mm. The dissolu- [10] Bergant Z, Grum J. J. Therm. Spray Technol. 2009, 18, 380–391. tion of WC particles was very little, and WC particles kept [11] Liu A, Guo M, Zhao M, Wang C. Surf. Coat. Technol. 2007, 201, their sharp edges. The dissolved WC particles precipitated 7978–7982.