The Reinforcement Mechanisms of Graphene Oxide in Laser Directed Energy Deposition Fabricated Metal and Ceramic Matrix Composites: A Comparison Study

Carbon-based nanomaterials mainly including carbon nanotubes (CNTs), graphene, and graphene oxide (GO) have superior properties of low density, outstanding strength, and high hardness. Compared with ceramic reinforcements, a small amount of carbon-based nanomaterials can signicantly improve the mechanical properties of metal matrix composites (MMCs) and ceramic matrix composites (CMCs). However, CNTs and graphite always aggregate or degrade during the fabrication with a high temperature, especially in MMCs. GO has the advantages of easier to be dispersed in other materials and better high-temperature stability. Laser directed energy deposition (DED), has been used to fabricate GO-MMCs and GO-CMCs due to the unique capabilities of coating, remanufacturing, and producing functionally graded materials. Laser DED, as a fusion manufacturing process, could fully melt the material powders, which could rene the microstructure and increase the density and mechanical properties. However, GO could react with matrix materials at high temperatures. The survival, degradation, and reactions of GO in laser DED fabricated GO-MMCs and GO-CMCs are still unknown. There is also no investigation on the reinforcement mechanisms of GO in metal matrix materials and ceramic matrix materials in the laser DED process. In this study, GO reinforced Ti (GO-Ti) and GO reinforced zirconia toughened alumina (GO-ZTA) parts were fabricated by laser DED process. Raman spectrum, XRD analysis, and EDS analysis have been applied to investigate the forms of GO in both DED fabricated GO-MMCs and GO-CMCs. The reinforcement mechanisms of GO on microhardness and compressive properties of MMCs and CMCs have been analyzed.


Introduction
Carbon-based nanomaterials, including carbon nanotubes (CNTs), graphene, and graphene oxide (GO), have the advantages of demonstrated low density, high strength, and high hardness, which made them preferable reinforcements to improve the hardness, wear resistance, and compressive properties of metal and ceramic materials [1,2]. Research works have been carried out to study carbon-based nanomaterials reinforced metal matrix composites (MMCs) and ceramic matrix composites (CMCs) [2][3][4][5]. Recently, carbon-based nanomaterials reinforced MMCs and CMCs have been successfully fabricated by different manufacturing processes [6][7][8][9][10][11][12][13][14][15]. The results showed that these nanomaterials could improve the hardness, compressive properties, and tensile properties due to their high strength and hardness [4,5,[16][17][18][19]. The self-lubrication properties of carbon-based nanomaterials could also improve the wear resistance of carbon-based nanomaterial reinforced MMCs and CMCs. However, it has been reported that CNTs and graphene are easily decomposed and can react with some matrix materials during the hightemperature fabrication processes [3,20]. In addition, CNTs and graphene are di cult to be uniformly mixed with metal or ceramic powders [12,13]. To reduce these disadvantages of CNTs and graphene, GO has been applied to serve as a reinforcement material in metal matrix materials and ceramic matrix materials to improve the mechanical properties. GO has a one-layer carbon atom plate structure with functional groups (such as -OH, =O, -COOH, etc.) that are attached at the edges of GO plates [21]. The functional groups could signi cantly change the Van der Waals interactions between different plates, which made GO to be easily dispersed in water and mixed with matrix materials [14]. The functional groups could also increase the stability of GO, which could reduce the possibility to react with matrix materials at a high temperature. In addition, compared with CNTs and graphene, the cost of GO was much lower.
Traditional GO reinforced MMCs and CMCs manufacturing processes had the shortcomings of high energy consumption and shape-restriction [22]. Compared with traditional fabrication processes, laser additive manufacturing (LAM) had the advantages of easy controllability, high stability, and the capabilities of surface modi cation. Recently, GO reinforced MMCs and CMCs have been fabricated by several LAM processes, including powder bed fusion (PBF) and laser directed energy deposition (DED) [16,[23][24][25]. Selected laser sintering (SLS), as a kind of PBF process, has been successfully used to fabricate GO reinforced MMCs (such as GO-Ti and GO-Fe) and GO reinforced CMCs (such as GO-Al 2 O 3 and GO-ZrO 2 ) with a relatively low fabrication temperature, which would result in the lower density of the fabricated parts [3,23,25,26]. Compared with the SLS process, the temperature in the SLM and laser DED processes was much higher. The material powers could be fully melted, which contributed to the development of the parts with ne-grained microstructure and high density [27]. Compared with SLM, laser DED had the capabilities of coating, remanufacturing, and producing functionally graded materials. However, it has been reported that GO plates might react with matrix materials during the fabrication due to the high fabrication temperature during the laser DED process [3,26]. The survival, reaction, and degradation of GO in laser DED fabricated GO reinforced MMCs and CMCs had not been studied. There was also no comparison study on the reinforcement mechanisms of GO on the mechanical properties of laser DED fabricated MMCs and CMCs.
In this study, GO reinforced Ti (GO-Ti) and GO reinforced zirconia toughened alumina (GO-ZTA) parts were fabricated by laser DED process. The survival of GO in Ti and ZTA was analyzed by Raman spectra results, EDS results, XRD results, and microstructure morphology. The different reinforcement mechanisms of GO on the microhardness and compressive properties of MMCs and CMCs were further investigated.

Materials and powder treatment
GO was in the state of 0.4 wt.% water suspension (Graphenea Inc., MA, USA) with a thickness less than 15 nm. The commercially pure Ti (CP-Ti) powder (99.7% purity) (Atlantic Equipment Engineers Inc., USA) with an average particle size of 50 µm was used in this study. The particle size of Al 2 O 3 power (99.9% purity) (Atlantic Equipment Engineers Inc., USA) was 45-75 µm. The particle size of ZrO 2 (99.9% purity) (Atlantic Equipment Engineers Inc., USA) was 1-5 µm.
The powder treatment processes of GO-Ti and GO-ZTA powders were shown in Fig. 2. For the GO-Ti powder, the weight ratio of Ti and GO was 99:1. For the GO-ZTA powder, the weight contents of GO, ZrO 2 , and Al 2 O 3 were 1 wt.%, 19.8 wt.%, and 79.2 wt.%. Before the mixing, the GO water dispersion was prepared by the ultrasonic treatment for two hours to suppress the aggregation of GO plates. After that, Ti / Al 2 O 3 powder was mixed with GO water dispersion with the assistance of ultrasonic vibration for four hours. Then the mixture was dried by a vacuum Oven (DFA-7000, MTI Corporation, USA) for two days.
After that, a planetary ball-milling machine (ND2L, Torrey Hills Technologies LLC., USA) was used to further dry and mix GO-Ti / GO-Al 2 O 3 powder. During the ball-milling processes, the sun wheel and the milling jars rotated in opposite directions with a speed of 200 rpm for two hours. The weight ratio between powder and milling balls was 3:1. For GO-ZTA powder, ZrO 2 powder was mixed with prepared GO-Al 2 O 3 powder and the mixture was further ball-milled for four hours. After the ball-milling process, GO plates were embedded on the surface of Ti particles. For GO-ZTA powders, GO plates and part of ZrO 2 particles were embedded on the surface of Al 2 O 3 particles.

Experiment set-up
A laser DED system (LENS 450, Optomec Inc., USA) was used to conduct the experiments. To avoid GO plates and Ti reacted with O 2 , the chamber system was purged by argon gas to a low oxygen level (< 50 ppm) before the fabrication. During the fabrication, the powder and gas delivery system delivered the feedstock powders, and the laser system generated and transferred the laser beam to the workpiece at the same time. A molten pool was generated on the surface of the substrate, which caught and melted the materials powders from the delivered powder ow. The molten pool solidi ed when the laser beam moved away. The deposition head movement was controlled by the control system to fabricate the part following the prepared trajectory. When the rst layer was deposited, the deposition head moved up onelayer thickness distance (Z increment), and the second layer was fabricated based on the rst layer. By repeating the deposition processes, the parts were deposited layer by layer. The input parameters of GO-Ti and GO-ZTA were listed in Table 1.  and GO-ZTA parts were increased. As shown in Fig. 3(a), there was an overlapping between the error bars of CP-Ti parts and GO-Ti parts. A paired sample t-test was conducted to compare the microhardness. The associated P-value (0.019) of these two groups of data was less than 0.05, which indicated that the microhardness value of GO-Ti parts was signi cantly larger than that of Ti parts. The microhardness of ZTA parts and GO-ZTA parts were shown in the right gure. With the addition of GO, the microhardness value of GO-ZTA parts (1980 HV1.0) was much larger than that of ZTA parts (1680 HV1.0). The additional t-test was not necessary to be conducted. The reasons for the signi cant increase of microhardness would be further discussed in Sect. 5.1. Figure 4 shows the stress-strain curve and the compressive properties of GO-Ti parts and GO-ZTA parts. As shown in Fig. 4(a1) and Fig. 4(a2), compared with Ti parts, the compressive strength and Young's modulus of GO-Ti parts were higher. However, the addition of GO would slightly decrease the toughness and ductility of GO-Ti parts. The compressive strength, Young's modulus, toughness, and ductility of ZTA parts and GO-ZTA parts were shown in Fig. 4(b2). GO could signi cantly improve the compressive properties of CMCs. With the addition of GO, the compressive strength and toughness were increased by two times to three times. In addition, Young's modulus and ductility were also signi cantly increased. and 1600 cm − 1 corresponded to the D peak and G peak of GO, respectively. The D peak was caused by the presence of disorder in sp 2 -hybridized carbon systems in GO. G peak existed from the stretching of the C-C bond in graphitic materials, which was common to all sp 2 carbon systems. Figure 5(b) shows the Raman spectrum of laser DED fabricated GO-Ti parts. Raman peak was not observed in the spectrum from this sample, indicating no GO survived in GO-Ti parts after the fabrication. The possible reason was that most GO decomposed and reacted with Ti during the fabrication [28]. Figure 5(c) shows the Raman spectrum of laser DED fabricated GO-ZTA parts. The emergence of the D peak and G peak in the Raman spectrum of GO-ZTA showed that GO plates survived during the fabrication. It should be noticed that the intensities of D and G peaks were reduced. One of the possible reasons was that during the hightemperature laser DED process, part of GO was decomposed or lost [20]. The second reason was that GO plates were wrapped in liquid state ZTA during the fabrication. After the solidi cation, the GO plates were distributed in the ZTA matrix and became harder to be detected [25].

EDS analysis and XRD analysis on element and phase composition
As discussed in Sect. 4.1, Raman spectrum results only proved the decomposition of GO in GO-Ti parts after the laser DED process. The reaction products of GO and Ti were hard to be shown in Raman spectrum results. Figure 6 shows the composition analysis of GO-Ti parts. EDS analysis was used to investigate the element composition of the fabricated GO-Ti parts. It can be seen in Fig. 6 (a), there were black regions unevenly distributed in light regions. In the EDS results of probe 1, the atomic ratio of Ti to C was around 1:1, which was similar to that of TiC. In addition, the shape and size of the black regions were different from that of GO plates, indicating the occurrence of the reactions. It could be considered that these black regions were TiC particles that were generated during the fabrication. To con rm the phase compositions of these black regions, X-ray power diffraction technology (XRD) was conducted. The XRD patterns of laser DED fabricated Ti and GO-Ti parts were shown in Fig. 6(b). It can be seen that the GO-Ti pattern had the peaks at 2θ degree of 36.5°, 42.4°, and 73.7°, which could be indexed into the phase of TiC. It could be con rmed that most GO reacted with Ti at a high temperature, forming TiC particles. The major element at point 2 was Ti (> 90 at.%), which provided that the light region was Ti matrix. There was also a small composition of C element. The reason was the diffusion of C element during the fabrication. There were ultra ne TiC particles distributed in the Ti matrix [29]. The microstructure of Ti and GO-Ti parts fabricated by the laser DED process was shown in Fig. 8. It could be observed that some black particles were distributed in light regions. According to the results of Raman spectrum analysis, XRD analysis, and EDS analysis, these black particles were TiC particles that were formed during the fabrication. During the laser DED process, GO plates partly lost the functional groups in the molten pool due to the high temperature [32]. The content of H element and O element was decreased and the content of C element in GO plates became higher. According to the Ti-C phase diagram. When the temperature of the molten pool came to 1670°C, the liquid Ti could react with C to generate TiC [33]. The reinforcement mechanism of GO in GO-Ti was that TiC particles were formed during the fabrication. Due to the extremely high hardness of TiC, TiC could support the load force and reduce deformation when the indenter of the microhardness tester penetrated in GO-Ti parts during the microhardness tests. Figure 8 shows the microstructure of ZTA and graphene GO-ZTA parts fabricated by the laser DED process. GO plates were survived in GO-ZTA parts, and there was a clear separation between GO plates and ZTA ceramic matrix. There were two reasons for the survival of GO plates. First, the melting point of GO (about 3600°C) was much higher than that of Al 2 O 3 (2072°C) and ZrO 2 (2715°C). During the laser DED process, the ZTA ceramic powders were melted in the molten pool and re-solidi ed with GO plates, but GO plates were not melted and maintained their original shapes. Second, different from some of the metallic materials, ZTA ceramics could not react with GO even at a high temperature. As discussed in Sect. 3.1, the microhardness of GO-ZTA parts was much higher than ZTA parts. There were two reinforcement effects of GO on GO-ZTA. First, the laser absorbability of GO was lower than Al 2 O 3 . The addition of GO reduced the temperature of the molten pool and increased the cooling rate during the fabrication, which could re ne the grain size of GO-ZTA [27]. With the decrease of the grain size, the microhardness could be increased. The second reason was that GO plates had extremely high hardness and strength. They could support the load and reduce the deformation during the microhardness indentation tests. Figure 9 shows the features of the fracture interface of GO-Ti parts and GO-ZTA parts. As shown in Fig.  9(a), the fracture marks were relatively evenly distributed in the fracture interface of Ti parts. Some smallsized fracture marks extended to a certain area, forming larger-sized fracture marks. In GO-Ti parts, fracture marks were generated around the formed TiC particles. There were almost no fracture marks in the regions relatively far from TiC particles.

Fracture interface characterizations
The reason for the higher ultimate compressive strength and Young's modulus of GO-Ti was that the ultra ne stiff particles could transfer the loading from the deformed Ti matrix to themselves during the compressive tests [28]. The TiC particles with higher hardness undertook more loading, which suppressed the dislocation movement and the generation of the fractures inside the Ti matrix (relatively far from TiC particles). Larger stress was required to deform the GO-Ti parts. The lower ductility of GO-Ti parts was caused by the uneven distribution of fractures, resulting in the growth of the fracture to a critical value. The uneven distribution was caused by two major reasons. First, the high-stress concentration occurred in the areas around TiC particles due to the irregular shape of TiC particles. Second, there was a large difference in deformation compatibility between TiC and Ti [34]. Under a similar load, the dislocation movement of TiC and Ti was different, which further intensi ed the fractures generation in the interfaces of TiC and Ti. GO has little in uence on toughness. The toughness was the ability of a material to absorb energy upon fracture failure, which could be calculated as the area under the strain-stress curve. Although the ultimate compressive strength of GO-Ti parts increased, their ductility was signi cantly decreased. Figure 9 shows the features of the fracture interface of laser DED fabricated ZTA parts and GO-ZTA parts.
In ZTA parts, the parallel distributed fracture marks had no obvious tendency of disappearance when they went through the grain boundaries. As a comparison, in GO-ZTA parts, fracture marks were signi cantly suppressed when they extended to the GO plates. In addition, compared with ZTA parts, the number and size of the fracture marks in GO-ZTA parts were signi cantly decreased.
As discussed in Sect. 4.2.1, GO plates were embedded in the matrix of ZTA. When the fractures extended to GO plates during the compressive tests, GO plates could connect both sides of the fractures to form a bridging condition. It could stop the dislocation movement during the compressive tests, which suppressed the further growth of the fractures. The number and size of fractures were reduced, which contributed to the improvement of compressive properties. In addition, GO plates could undertake the concentrated stress during the compressive tests. It was reported that the graphene-based nanomaterial tended to accumulate around the grain boundaries of ZTA ceramics where the stress was concentrated [13]. Due to the excellent hardness and strength, GO plates increased the energy requirement during the compressive tests. Finally, as discussed in Sect. 4.2.1, GO could re ne the grain size of GO-ZTA parts. The smaller grain size could increase the number of high-toughness grain boundaries, which could suppress the dislocations during the compressive tests to enhance the compressive properties [35].

Conclusion
To investigate the effects of GO on the compressive properties, and microhardness of MMCs and CMCs, GO-Ti and GO-ZTA parts were fabricated by the laser DED process in this study. The survival, degradation, and reactions of GO in laser DED fabricated GO-Ti and GO-ZTA parts were analyzed through the Raman spectrum analysis, EDS analysis, and XRD analysis. The microstructure and element distribution of GO-Ti and GO-ZTA composite parts were investigated. The reinforcement mechanism of GO in GO-Ti and GO-ZTA was further studied. The major conclusions are drawn as follows: 1. GO plates reacted with Ti to form TiC particles during the laser DED process. The major reason was that GO lost the oxygen-contained functional groups at a high temperature, promoting the reaction between Ti and C in the molten pool. As a comparison, GO could survive in GO-ZTA parts since it could not react with ZTA at a high temperature.
2. In GO-Ti parts, formed TiC was distributed in the Ti matrix, which had little in uence on the microstructure morphology. The microhardness of GO-Ti was improved due to the high hardness of TiC. As a comparison, the microhardness of GO-ZTA was improved by the existence of GO and the re nement of grain size.
3. The ultimate compressive strength and microhardness of GO-Ti were increased due to the high hardness of formed ultra ne TiC particles. The ductility of GO-Ti was decreased due to the uneven