Corrosion and Thermal Fatigue Behaviors of TiC/Ni Composite Coating by Self-Propagating High-Temperature Synthesis in Molten Aluminum Alloy

TiC/Ni composite coatings on H13 steel plates were fabricated in situ by self-propagating high-temperature synthesis and combined with a pseudo-heat isostatic press. The microstructure of the coating was characterized by X-ray diffraction and scanning electron microscopy. The microhardness, corrosion, and thermal fatigue behaviors of the coating were investigated by a microhardness test, immersion test, and thermal fatigue test, respectively. The results showed that the in situ coating consisted of TiC and Ni binder phases. Spheroidal TiC particles were enveloped by a nearly continuous Ni binder phase. Coating showed good metallurgical bonding in the interface. The corrosive mechanism of the coating surface in molten aluminum alloy involves the Ni binder phase being etched by aluminum to form AlNi3 and the oxidization of the TiC-reinforced phase. The corrosive mechanism that occurred at the front of the corrosion involves the Ni binder phase of the coating being etched by aluminum to form AlNi3, while the TiC skeleton still maintains the original organizational structure. Hot fatigue cracks began at the defective tips of the coating and propagated in the TiC-reinforced phase. The crack is a trans-granular fracture, which is the result of brittle rupture.


Introduction
Die casting dies are exposed to high-temperature molten metals, high pressures and flow velocities, as well as large thermal gradients, which result in the initiation of a variety of mechanochemical failure mechanisms. The main failure modes of die casting dies involve heat cracking or thermal cycling inducing fatigue cracking, erosive wear by molten metal, or corrosion from cast metal during die filling or soldering [1][2][3][4]. Research has indicated that surface treatment and coating could be an effective way to protect die surfaces from thermal fatigue and extend die life by reducing the damage at contact surfaces [5]. Several surface modification methods, such as physical vapor deposition (PVD), chemical vapor deposition (CVD), ion implantation, thermal spray, self-propagating high-temperature synthesis (SHS), as well as laser and plasma cladding, have been adopted to produce coatings of carbide, nitride, boride, and silicide for enhancing the service life of dies [1,[5][6][7][8][9]. Among these, SHS has many remarkable advantages, such as high efficiency and purity of products, low energy and cost requirements, no high-temperature furnace process, non-polluting traits, and a relatively simple process [10,11]. At present, SHS is usually used to produce compounds or composite materials. The starting powders were made from commercial powders of Ti (99.9% purity, Shanghai Naiou Nano Technology Co., Ltd., Shanghai, China) with an average particle size of~50 µm; carbon black (99.9% purity, Tianjin Tianyi Century Chemical Products Technology Development Co., Ltd., Tianjin, China) with an average particle size of~0.2 µm; Ni (99.9% purity, Shanghai Naiou Nano Technology Co., Ltd.) with an average particle size of~10 µm; and Mo (99.95% purity, Shanghai Naiou Nano Technology Co., Ltd.) with an average particle size of~10 µm.

Coating Procedure
The composite coating has three layers. Table 2 shows the composition and weight of each layer. Ni is used as a binder phase to improve the density and strength of the coating. The transition layer can improve the bonding strength between the substrate and coating, while Mo can improve the wettability of TiC by Ni [23]. The reactant mixture of each layer was mixed in a ball grinding mill for 8 h and dried in a vacuum drying oven for 24 h at a temperature of 90 • C, which is followed by compaction in a stainless steel die to form a compact sample. The green density of each compact was maintained at about 55% of theoretical density. A schematic illustration of the fabrication of composite coating by SHS/PHIP is shown in Figure 1. Three powder compacts were stacked on the substrate in turn. Loose powder particles of 20Ni-64Ti-16C were wrapped around the substrate and powder compacts acted as a pressure-transmitting medium. This created a pseudo-heat isostatic press state on the compacts and compensated for the heat loss of the Ti-C-Ni reaction system. The reactor was purged and filled with argon before ignition. The combustion system was preheated to a temperature of 300 • C and then ignited using a tungsten filament. After the completion of combustion 2 s later, the reaction system was pressed for 20 s at a compaction pressure of 15 MPa. wettability of TiC by Ni [23]. The reactant mixture of each layer was mixed in a ball grinding mill for 8 h and dried in a vacuum drying oven for 24 h at a temperature of 90 °C, which is followed by compaction in a stainless steel die to form a compact sample. The green density of each compact was maintained at about 55% of theoretical density. A schematic illustration of the fabrication of composite coating by SHS/PHIP is shown in Figure 1. Three powder compacts were stacked on the substrate in turn. Loose powder particles of 20Ni-64Ti-16C were wrapped around the substrate and powder compacts acted as a pressuretransmitting medium. This created a pseudo-heat isostatic press state on the compacts and compensated for the heat loss of the Ti-C-Ni reaction system. The reactor was purged and filled with argon before ignition. The combustion system was preheated to a temperature of 300 °C and then ignited using a tungsten filament. After the completion of combustion 2 s later, the reaction system was pressed for 20 s at a compaction pressure of 15 MPa.

Characterizations
The microstructural analysis of coating was conducted using a scanning electron microscope (SEM, NOVA 400 NanoSEM, FEI, Hillsboro, OR, USA) attached with an energy dispersive spectrometer (EDS). A phase analysis of coating was carried out by X-ray diffraction (XRD, X'Pert PRO MPD, PANalytical, Almelo, the Netherlands) using Cu-Kα radiation (λ = 1.54056 Å) in a 2θ range of 10°-90° with a scan speed of 1.2°/min. Surface microhardness and cross-sectional microhardness of the coating were measured by the Vickers hardness tester (HV-1000A, Shen Youda Industrial (Shanghai) Co. Ltd., Shanghai, China) using an indenting loading of 50 g and time of 10 s, as well as 500 g and 10 s, respectively.

Immersion Test
The coated sample was immersed in a clay crucible filled with molten aluminum alloy (ZLD301) for 2 h. The chemical composition of ZLD301 is provided in Table 3. The temperature of the melt was maintained at 700 °C, which is slightly above the melting temperature of pure aluminum. This is generally the temperature of the melted aluminum at the time of its injection into the die cast cavity. After the immersion test, the coated sample was taken out and cooled in air before being divided into sections for analysis.

Characterizations
The microstructural analysis of coating was conducted using a scanning electron microscope (SEM, NOVA 400 NanoSEM, FEI, Hillsboro, OR, USA) attached with an energy dispersive spectrometer (EDS). A phase analysis of coating was carried out by X-ray diffraction (XRD, X'Pert PRO MPD, PANalytical, Almelo, the Netherlands) using Cu-Kα radiation (λ = 1.54056 Å) in a 2θ range of 10 • -90 • with a scan speed of 1.2 • /min. Surface microhardness and cross-sectional microhardness of the coating were measured by the Vickers hardness tester (HV-1000A, Shen Youda Industrial (Shanghai) Co. Ltd., Shanghai, China) using an indenting loading of 50 g and time of 10 s, as well as 500 g and 10 s, respectively.

Immersion Test
The coated sample was immersed in a clay crucible filled with molten aluminum alloy (ZLD301) for 2 h. The chemical composition of ZLD301 is provided in Table 3. The temperature of the melt was maintained at 700 • C, which is slightly above the melting temperature of pure aluminum. This is generally the temperature of the melted aluminum at the time of its injection into the die cast cavity. After the immersion test, the coated sample was taken out and cooled in air before being divided into sections for analysis.

Thermal Fatigue Test
A schematic illustration of the thermal fatigue test apparatus is shown in Figure 2 [24,25]. It enables controlled thermal fatigue testing at conditions similar to aluminum alloy die casting. The test sample was immersed in a water-based lubricant at about 25 • C for 7 s, which prevented the aluminum from sticking to the test sample. After that, the sample was moved through air at 25 • C for 15 s and placed into the bath of the molten aluminum alloy ZLD301 at about 700 • C. After 15 s of immersion in this bath, the sample was air-cooled for 14 s before being placed into a bath of water-based lubricant. The total cycle duration is 51 s. The test sample is not subjected to pressure or aluminum flow, unlike the die during die casting. The movements of the test sample during the test were achieved by two-screw sliding tables controlled by a two-axis numerical control system. The test sample underwent 600 cycles.

Thermal Fatigue Test
A schematic illustration of the thermal fatigue test apparatus is shown in Figure 2 [24,25]. It enables controlled thermal fatigue testing at conditions similar to aluminum alloy die casting. The test sample was immersed in a water-based lubricant at about 25 °C for 7 s, which prevented the aluminum from sticking to the test sample. After that, the sample was moved through air at 25 °C for 15 s and placed into the bath of the molten aluminum alloy ZLD301 at about 700 °C. After 15 s of immersion in this bath, the sample was air-cooled for 14 s before being placed into a bath of water-based lubricant. The total cycle duration is 51 s. The test sample is not subjected to pressure or aluminum flow, unlike the die during die casting. The movements of the test sample during the test were achieved by two-screw sliding tables controlled by a two-axis numerical control system. The test sample underwent 600 cycles.

Phase Analysis
The X-ray diffraction pattern of the original coating is shown in Figure 3. As shown in Figure 3, the original coating only consisted of TiC and Ni peaks. TiC and Ni are thermodynamically stable phases, although no Ni-Ti or Ni-Ti-C compounds were detected using X-ray diffraction analysis. Dunmead et al. [26], Zhang et al. [27], and Han et al. [28] found that the final products were TiC and unreacted Ni for the Ti-C-Ni system. Wong et al. [29] observed the Ni-Ti compounds (Ni3Ti or NiTi) using time-resolved X-ray diffraction on a Ti-C-25Ni reactant mixture. Xiao et al. [11] studied the mechanism of SHS of Ti-C-Ni system by means of a combustion front quenching method. They found that the formation of Ni3Ti phase was attributed to the incomplete reaction due to the usage of coarser Ni and Ti powders. In addition to the above-mentioned factors, C particle size [30], Ni content, and preheating temperature will also affect the degree to which the SHS reaction will be completed. Therefore, the reaction of SHS in this work was complete because of the usage of finer starting powders.

Phase Analysis
The X-ray diffraction pattern of the original coating is shown in Figure 3. As shown in Figure 3, the original coating only consisted of TiC and Ni peaks. TiC and Ni are thermodynamically stable phases, although no Ni-Ti or Ni-Ti-C compounds were detected using X-ray diffraction analysis. Dunmead et al. [26], Zhang et al. [27], and Han et al. [28] found that the final products were TiC and unreacted Ni for the Ti-C-Ni system. Wong et al. [29] observed the Ni-Ti compounds (Ni 3 Ti or NiTi) using time-resolved X-ray diffraction on a Ti-C-25Ni reactant mixture. Xiao et al. [11] studied the mechanism of SHS of Ti-C-Ni system by means of a combustion front quenching method. They found that the formation of Ni 3 Ti phase was attributed to the incomplete reaction due to the usage of coarser Ni and Ti powders. In addition to the above-mentioned factors, C particle size [30], Ni content, and preheating temperature will also affect the degree to which the SHS reaction will be completed. Therefore, the reaction of SHS in this work was complete because of the usage of finer starting powders.  Figure 4 shows the scanning electron micrographs of the surface layer and interlayer. The resulting microstructures for the TiC/Ni coating consisted of a spheroidal TiC phase (dark) embedded in a nearly continuous Ni binder phase (white). The very fine Ni network around the TiC particles should produce a considerable increase in toughness compared with the pure TiC [31]. Moreover, as shown in Figure 4, the TiC size of surface layer was larger than that of interlayer. The grain growth of TiC is an exponential function of the combustion temperature [28]. The interlayer with 40 wt % Ni has a low combustion temperature and long diffusion path for TiC, which reduces the driving force for TiC grain growth and prevents the sintering between TiC grains to form larger grains.

Bonding Performance and Microhardness
The microstructure of the interface between interlayer and substrate is shown in Figure 5. From Figure 5, the transition layer disappeared and the interface was not flat because of high preheating temperature, which resulted in high Ti-C-Ni reaction temperature and sufficient element diffusion. Figure 6 shows the line scanning spectra of the interface between substrate and interlayer. As seen from Figure 6, the main elements for diffusion were Ti and Fe, while the diffusion distance was longer than 15 μm. The curve of Fe changed gently in the interface, while Ti showed a dramatic change. The non-flat interface and element diffusion in the interface showed good metallurgical bonding.
The surface microhardness of the original coating was 1159.06 HV0.05, which was about two times the substrate microhardness of 439 HV0.2. Table 4 lists the cross-sectional microhardness of the original coating. From Table 4, the cross-sectional microhardness of the original coating gradually decreased from the coating surface to the substrate. There was little difference between the surface Intensity (cps)  Figure 4 shows the scanning electron micrographs of the surface layer and interlayer. The resulting microstructures for the TiC/Ni coating consisted of a spheroidal TiC phase (dark) embedded in a nearly continuous Ni binder phase (white). The very fine Ni network around the TiC particles should produce a considerable increase in toughness compared with the pure TiC [31]. Moreover, as shown in Figure 4, the TiC size of surface layer was larger than that of interlayer. The grain growth of TiC is an exponential function of the combustion temperature [28]. The interlayer with 40 wt % Ni has a low combustion temperature and long diffusion path for TiC, which reduces the driving force for TiC grain growth and prevents the sintering between TiC grains to form larger grains.  Figure 4 shows the scanning electron micrographs of the surface layer and interlayer. The resulting microstructures for the TiC/Ni coating consisted of a spheroidal TiC phase (dark) embedded in a nearly continuous Ni binder phase (white). The very fine Ni network around the TiC particles should produce a considerable increase in toughness compared with the pure TiC [31]. Moreover, as shown in Figure 4, the TiC size of surface layer was larger than that of interlayer. The grain growth of TiC is an exponential function of the combustion temperature [28]. The interlayer with 40 wt % Ni has a low combustion temperature and long diffusion path for TiC, which reduces the driving force for TiC grain growth and prevents the sintering between TiC grains to form larger grains.

Bonding Performance and Microhardness
The microstructure of the interface between interlayer and substrate is shown in Figure 5. From Figure 5, the transition layer disappeared and the interface was not flat because of high preheating temperature, which resulted in high Ti-C-Ni reaction temperature and sufficient element diffusion. Figure 6 shows the line scanning spectra of the interface between substrate and interlayer. As seen from Figure 6, the main elements for diffusion were Ti and Fe, while the diffusion distance was longer than 15 μm. The curve of Fe changed gently in the interface, while Ti showed a dramatic change. The non-flat interface and element diffusion in the interface showed good metallurgical bonding.
The surface microhardness of the original coating was 1159.06 HV0.05, which was about two times the substrate microhardness of 439 HV0.2. Table 4 lists the cross-sectional microhardness of the original coating. From Table 4, the cross-sectional microhardness of the original coating gradually

Bonding Performance and Microhardness
The microstructure of the interface between interlayer and substrate is shown in Figure 5. From Figure 5, the transition layer disappeared and the interface was not flat because of high preheating temperature, which resulted in high Ti-C-Ni reaction temperature and sufficient element diffusion. Figure 6 shows the line scanning spectra of the interface between substrate and interlayer. As seen from Figure 6, the main elements for diffusion were Ti and Fe, while the diffusion distance was longer than 15 µm. The curve of Fe changed gently in the interface, while Ti showed a dramatic change. The non-flat interface and element diffusion in the interface showed good metallurgical bonding.
The surface microhardness of the original coating was 1159.06 HV 0.05 , which was about two times the substrate microhardness of 439 HV 0.2 . Table 4 lists the cross-sectional microhardness of the original coating. From Table 4, the cross-sectional microhardness of the original coating gradually decreased from the coating surface to the substrate. There was little difference between the surface layer and the interlayer, while the interfacial microhardness was between the substrate and the interlayer. Figure 7 shows the interfacial indentation. As shown in Figure 7, the indentation had no crack and the indentation area was very small. The above description indicated that the interface between the coating and steel substrate had strong metallurgical bonding.
Coatings 2017, 7, 203 6 of 12 crack and the indentation area was very small. The above description indicated that the interface between the coating and steel substrate had strong metallurgical bonding.             . Interfacial indentation at a load of 500 g. Figure 7. Interfacial indentation at a load of 500 g. Figure 8 shows the XRD pattern of the original coating after immersion test. Compared with the XRD pattern of the original coating before immersion test, the new phases of TiO 2 and AlNi 3 appeared. As the nickel aluminum reaction is an intense exothermic reaction, Ni reacted preferentially with the Al of the coating surface to form AlNi 3 , which resulted in the Ni oxide peaks missing in the XRD pattern. TiO 2 was caused by the oxidation of TiC.  Figure 8 shows the XRD pattern of the original coating after immersion test. Compared with the XRD pattern of the original coating before immersion test, the new phases of TiO2 and AlNi3 appeared. As the nickel aluminum reaction is an intense exothermic reaction, Ni reacted preferentially with the Al of the coating surface to form AlNi3, which resulted in the Ni oxide peaks missing in the XRD pattern. TiO2 was caused by the oxidation of TiC.  Figure 9 shows the SEM and EDS results of the coating surface after immersion test. From Figure 9, a considerable number of white crystal particles (2 in Figure 9a), globular protuberances (3 in Figure 9a) and a few flat surfaces (1 in Figure 9a) appeared in the coating surface. According to the EDS results and XRD pattern (Figure 8), region 1 consisted of TiC and TiO2; region 2 was mainly AlNi3; and region 3 was mainly composed of TiC, TiO2, Al, and Mg, which were adhered to the surface of TiC.    Figure 9 shows the SEM and EDS results of the coating surface after immersion test. From Figure 9, a considerable number of white crystal particles (2 in Figure 9a), globular protuberances (3 in Figure 9a) and a few flat surfaces (1 in Figure 9a) appeared in the coating surface. According to the EDS results and XRD pattern (Figure 8), region 1 consisted of TiC and TiO 2 ; region 2 was mainly AlNi 3 ; and region 3 was mainly composed of TiC, TiO 2 , Al, and Mg, which were adhered to the surface of TiC. Figure 10 shows the line scanning spectra from the coating surface to the inner part in the cross-section after the immersion test. From Figure 10, the local area of coating surface was exfoliated and micro-cracks were found. A large amount of O and Al appeared in the coating. The line scanning spectral fluctuation of C was inconsistent with Ti as the C content was less than that of Ti. Aluminum and oxygen in the molten aluminum alloy diffused inwardly by pores; Al reacted with the Ni binder phase to form AlNi 3 ; and O reacted with TiC to form TiO 2 and C. Some C was further oxidized to produce an overflow of CO 2 , which resulted in a lower C content compared to that in Ti. Therefore, the corrosive mechanism of the coating surface in molten aluminum alloy involves the Ni binder phase being etched by aluminum to form AlNi 3 and the oxidization of the TiC-reinforced phase. Figure 9 shows the SEM and EDS results of the coating surface after immersion test. From Figure 9, a considerable number of white crystal particles (2 in Figure 9a), globular protuberances (3 in Figure 9a) and a few flat surfaces (1 in Figure 9a) appeared in the coating surface. According to the EDS results and XRD pattern (Figure 8), region 1 consisted of TiC and TiO2; region 2 was mainly AlNi3; and region 3 was mainly composed of TiC, TiO2, Al, and Mg, which were adhered to the surface of TiC.     Figure 10 shows the line scanning spectra from the coating surface to the inner part in the cross-section after the immersion test. From Figure 10, the local area of coating surface was exfoliated and micro-cracks were found. A large amount of O and Al appeared in the coating. The line scanning spectral fluctuation of C was inconsistent with Ti as the C content was less than that of Ti. Aluminum and oxygen in the molten aluminum alloy diffused inwardly by pores; Al reacted with the Ni binder phase to form AlNi3; and O reacted with TiC to form TiO2 and C. Some C was further oxidized to produce an overflow of CO2, which resulted in a lower C content compared to that in Ti. Therefore, the corrosive mechanism of the coating surface in molten aluminum alloy involves the Ni binder phase being etched by aluminum to form AlNi3 and the oxidization of the TiC-reinforced phase.  Figure 11 shows the line scanning spectra of the cross-section at the internal corrosive front after the immersion test. The non-eroded area still maintained the original organizational structure. The bright Ni phase was still visible in the erosion area, but the color of the Ni phase was lighter than that of the Ni phase in the non-eroded area. Although TiC grains could still be seen in the eroded area, the bright Ni phase disappeared almost completely. From the line scanning spectra, the Al content gradually decreased to zero from the eroded area to non-eroded area. The spectrum peaks and valleys of Al were nearly consistent with those of Ni, which indicated that the peak positions of Al and Ni had more AlNi3 produced by reaction of Al with the Ni binder phase. The spectrum peaks and valleys of Ti were nearly consistent with those of C and completely different to those of Al and Ni, which suggested that Ti and C peak positions were TiC instead of AlNi3. TiO2 generated before on the surface of TiC particles hindered the further diffusion of oxygen within TiC particles. At the same time, the oxygen content inside the coatings is relatively low, so TiC particles inside the coating were  Figure 11 shows the line scanning spectra of the cross-section at the internal corrosive front after the immersion test. The non-eroded area still maintained the original organizational structure. The bright Ni phase was still visible in the erosion area, but the color of the Ni phase was lighter than that of the Ni phase in the non-eroded area. Although TiC grains could still be seen in the eroded area, the bright Ni phase disappeared almost completely. From the line scanning spectra, the Al content gradually decreased to zero from the eroded area to non-eroded area. The spectrum peaks and valleys of Al were nearly consistent with those of Ni, which indicated that the peak positions of Al and Ni had more AlNi 3 produced by reaction of Al with the Ni binder phase. The spectrum peaks and valleys of Ti were nearly consistent with those of C and completely different to those of Al and Ni, which suggested that Ti and C peak positions were TiC instead of AlNi 3 . TiO 2 generated before on the surface of TiC particles hindered the further diffusion of oxygen within TiC particles. At the same time, the oxygen content inside the coatings is relatively low, so TiC particles inside the coating were barely oxidized. Therefore, the corrosive mechanism that occurred at the corrosion front involves the Ni binder phase of coating being etched by aluminum to form AlNi 3 , while the TiC skeleton still maintains the original organizational structure.  Figure 12 shows the cracks of coating surface after 600 thermal fatigue cycles. As shown in Figure 12, a severe crack could be observed on the surface of the original coating. Figure 13 shows the micrograph of the cross section of the original coating after 600 thermal fatigue cycles. From Figure 13, there was a main crack that was located from the surface to the substrate interface and was perpendicular to the interface. However, the crack did not extend to the H13 steel substrate or along the coating-substrate interface due to the high hardness of substrate. This crack was caused by the high-temperature gradients through the whole thickness and surface thermal stress on the original coating [25]. It was also noticeable that the crack passed through a large number of defects, such as micro-pores and inclusions, as can be seen from Figure 13. Stress concentration is often produced at the coating defects, especially at sharp angles, which results in the local stress exceeding the yield strength of the coating material to produce the local plastic strain. When the local plastic strain accumulates to a certain extent, the material ductility is exhausted and coating material is no longer able to absorb the deformation during thermal cycling. After this, a crack will form, which is typical for low-cycle fatigue [32]. At the same time, the thermal expansion coefficients of TiC-reinforced phase and Ni binder phase in the coating are strikingly different, which further aggravates the crack formation.  Figure 12 shows the cracks of coating surface after 600 thermal fatigue cycles. As shown in Figure 12, a severe crack could be observed on the surface of the original coating. Figure 13 shows the micrograph of the cross section of the original coating after 600 thermal fatigue cycles. From Figure 13, there was a main crack that was located from the surface to the substrate interface and was perpendicular to the interface. However, the crack did not extend to the H13 steel substrate or along the coating-substrate interface due to the high hardness of substrate. This crack was caused by the high-temperature gradients through the whole thickness and surface thermal stress on the original coating [25]. It was also noticeable that the crack passed through a large number of defects, such as micro-pores and inclusions, as can be seen from Figure 13. Stress concentration is often produced at the coating defects, especially at sharp angles, which results in the local stress exceeding the yield strength of the coating material to produce the local plastic strain. When the local plastic strain accumulates to a certain extent, the material ductility is exhausted and coating material is no longer able to absorb the deformation during thermal cycling. After this, a crack will form, which is typical for low-cycle fatigue [32]. At the same time, the thermal expansion coefficients of TiC-reinforced phase and Ni binder phase in the coating are strikingly different, which further aggravates the crack formation. Figure 14 shows the crack propagation in the coating. As shown in Figure 14, the crack propagation mainly occurred in the TiC grains rather than along the grain boundaries of the TiC-reinforced phase and Ni binder phase, which is a typical feature of trans-granular fractures rather than intergranular fractures. At the same time, it also showed that the interfacial bonding strength between the in situ TiC grains and Ni grains was larger than that of TiC grains. Due to better wetting of TiC and Ni (wetting angle is 30) [33], there is a thin diffusion layer at the interface between TiC and Ni. This resulted in strong interfacial bonding strength between TiC and Ni phases. TiC ceramics are brittle and, therefore, crack propagation occurred mainly in TiC grains.

Thermal Fatigue Test
coating [25]. It was also noticeable that the crack passed through a large number of defects, such as micro-pores and inclusions, as can be seen from Figure 13. Stress concentration is often produced at the coating defects, especially at sharp angles, which results in the local stress exceeding the yield strength of the coating material to produce the local plastic strain. When the local plastic strain accumulates to a certain extent, the material ductility is exhausted and coating material is no longer able to absorb the deformation during thermal cycling. After this, a crack will form, which is typical for low-cycle fatigue [32]. At the same time, the thermal expansion coefficients of TiC-reinforced phase and Ni binder phase in the coating are strikingly different, which further aggravates the crack formation.    Figure 14 shows the crack propagation in the coating. As shown in Figure 14, the crack propagation mainly occurred in the TiC grains rather than along the grain boundaries of the TiC-reinforced phase and Ni binder phase, which is a typical feature of trans-granular fractures rather than intergranular fractures. At the same time, it also showed that the interfacial bonding strength between the in situ TiC grains and Ni grains was larger than that of TiC grains. Due to better wetting of TiC and Ni (wetting angle is 30) [33], there is a thin diffusion layer at the interface between TiC and Ni. This resulted in strong interfacial bonding strength between TiC and Ni phases. TiC ceramics are brittle and, therefore, crack propagation occurred mainly in TiC grains.   Figure 14 shows the crack propagation in the coating. As shown in Figure 14, the crack propagation mainly occurred in the TiC grains rather than along the grain boundaries of the TiC-reinforced phase and Ni binder phase, which is a typical feature of trans-granular fractures rather than intergranular fractures. At the same time, it also showed that the interfacial bonding strength between the in situ TiC grains and Ni grains was larger than that of TiC grains. Due to better wetting of TiC and Ni (wetting angle is 30) [33], there is a thin diffusion layer at the interface between TiC and Ni. This resulted in strong interfacial bonding strength between TiC and Ni phases. TiC ceramics are brittle and, therefore, crack propagation occurred mainly in TiC grains.

Conclusions
The TiC/Ni composite coatings on H13 steel plate were fabricated by SHS/PHIP. The microstructure, microhardness, interfacial bonding performance, corrosion, and thermal fatigue behaviors of the original coating were investigated. The following conclusions can be drawn:


The in situ coating consisted of the TiC-reinforced phase and Ni binder phase. Spheroidal TiC particles were enveloped by a nearly-continuous Ni binder phase.

Conclusions
The TiC/Ni composite coatings on H13 steel plate were fabricated by SHS/PHIP. The microstructure, microhardness, interfacial bonding performance, corrosion, and thermal fatigue behaviors of the original coating were investigated. The following conclusions can be drawn:

•
The in situ coating consisted of the TiC-reinforced phase and Ni binder phase. Spheroidal TiC particles were enveloped by a nearly-continuous Ni binder phase.

•
Coating shows good metallurgical bonding in the interface.

•
The corrosive mechanism of the coating surface in molten aluminum alloy involves the Ni binder phase being etched by aluminum to form AlNi 3 and the oxidization of the TiC-reinforced phase. The corrosive mechanism that occurred at the corrosive front involves the Ni binder phase of the coating being etched by aluminum to form AlNi 3 , while the TiC skeleton still maintains the original organizational structure. • Hot fatigue cracks began at the defective tips of coating and propagated in the TiC-reinforced phase. The crack is a trans-granular fracture, which is the result of brittle rupture.