1 Introduction

Graphene has attracted tremendous attention in the field of metal matrix composites in recent years because of its outstanding physical and mechanical properties [1, 2]. Many researches indicated that graphene was a promising solid lubricant to significantly improve the friction and wear properties of metal matrix composites [3,4,5,6,7,8]. As compared with traditional solid lubrication materials such as graphite, two-dimensional graphene with atomic-scale thickness possess high specific surface area and extremely high strength. Thus, large interfacial contact area with metal matrix can offer a better self-lubricating efficiency as well as the enhanced mechanical properties. However, the agglomeration of graphene caused by strong van der Waals interactions during fabrication process and the weak interface binding due to the poor affinity of graphene to metal matrix limit its applications as lubricant and reinforcement [9, 10]. In order to solve this problem, the diverse methods such as ball-milling, mechanically agitation, and electrostatic self-assembly have been used to disperse graphene in metal matrix [11,12,13].

Recently, G–Ni composites were in-situ fabricated in our group through one-step growing graphene in bulk nickel matrix using a powder metallurgical process [14]. The in-situ graphene can notably enhance the mechanical properties and reduce friction coefficient of nickel matrix owing to uniform dispersion of graphene in the matrix and sufficient interface strength between graphene and nickel. Unfortunately, the wear resistance of the composites becomes worse when graphene content is higher than 1 wt.%, because in-situ growth of graphene leads to large number of pores in the composites. The graphene-oil synergistic lubricating design demonstrated that tribological properties of the G–Ni composite can be improved by prestoring oil in micro-nanoporous structures [15]. However, it is necessary to improve the wear resistance of the G–Ni composite to meet the requirements under dry sliding condition.

In this paper, the G–Ni composite was in-situ fabricated by a powder metallurgical process using nickel powders as host matrix and sucrose as carbon source. The fabricated composite was further repressed and annealed to increase densification. The effect of counterparts and applied load on the tribological properties of the samples was investigated. The friction and wear mechanisms were explored in detail.

2 Experiment Procedure

2.1 Sample Preparation

Nickel powders (300 meshes, spherical, 99.8% purity) and sucrose (320 mg, analytical reagent) were used as matrix and carbon source of graphene. Firstly, 10 g of Ni powders and 320 mg of sucrose were dispersed in 150 mL deionized water. The solution was stirred mechanically in an oil bath at 120 °C until water was entirely evaporated. Then, the dry powders were pressed for 10 min to prepare a green compact in a steel mold at a pressure of 380 MPa. Subsequently, the green compact was pre-sintered to in-situ grow graphene at 1100 °C for 1 h in a quartz tube furnace with a hydrogen and argon mixed atmosphere. Finally, the pre-fabricated composite was re-densified by a repressing process at the same condition and annealing process. The G–Ni composite was annealed at 600 °C for 12 h in argon atmosphere followed by furnace cooling.

2.2 Tribological Tests and Characterization

X-ray diffraction patterns (XRD) were recorded using X-ray diffractometer (Bruker D8 Advance, Cu Kα radiation). The operation voltage and current were 40 kV and 40 mA, respectively. Raman spectrum was acquired using a confocal Raman spectrometer (HORIBA, LabRam HR Evolution) with a wavelength of 532 nm. The measurement was carried out with the 600 g/mm grating and a microscope confocal hole setting of 100 µm. The laser was focused through a 50 objective. The power of the laser was set to 12 mW. Vickers hardness was measured at a load of 200 g with a dwell time of 10 s using a micro-hardness indenter. The bulk density of the samples was measured by the Archimedes’ principle. Tribological tests were conducted on a UMT-2MT ball-on-disk tribometer (Center, USA) using a reciprocating sliding mode at room temperature in ambient air with relative humidity of about 40% under the following conditions. The samples were oscillating with a stroke of 5 mm and a sliding speed of 9 mm/s. The number of cycles was 1620 and the total sliding distance was 16.2 m. The tests were performed under the normal loads of 2, 4, 6, and 8 N, repeating twice or three times to ensure the reliability of data. The samples were cylindrical with a diameter of 25 mm and thickness of 3 mm. Before testing, the samples were polished to a surface roughness (Ra) of less than 0.5 µm using 600, 800, 1000, 1200, and 1500 grade emery, and then cleaned ultrasonically in alcohol for 10 min to remove surface debris. GCr15, Si3N4, and ZrO2 balls (Ø 6.35 mm) were used as counterparts sliding against G–Ni composite. The mechanical properties and surface roughness of the balls were obtained from the balls company, as listed in Table 1. The worn surfaces of G–Ni composite and the counterparts were observed and analyzed by a JSM 6700F scanning electron microscope equipped with an Oxford INCA X-ray energy dispersive spectrometer (EDS). SEM was operated in a secondary electron imaging mode at an acceleration voltage of 18 kV. The working distance was set at 8 mm. The wear rate was calculated from the wear volume divided by the sliding distance and the normal load, where the wear volume was measured using a 3D optical surface profiler (NanoFocus AG, Germany).

Table 1 Mechanical properties and roughness of GCr15, Si3N4, and ZrO2 balls
Full size table

3 Results and Discussion

3.1 Characterization of Sample

Table 2 shows the component and mechanical properties of the G–Ni matrix composite. It was found that the repressing and annealing process improve significantly the density and hardness of the G–Ni composite. The relative density of the sample increased from 91.7 to 97.5%, and the hardness of sample increased considerably from 86.7 to 122.1 HV owing to reduction of porosity [15]. Figure 1a shows the XRD patterns of the sample before and after annealing. The appearance of only nickel diffraction peaks suggests that the annealing did not trigger the interfacial reaction between graphene and nickel matrix. Raman spectra of the samples show significant differences before and after annealing, as shown in Fig. 1b. Obviously, the annealing results in a sharp increase of intensity of D peak and broadening of G peak. The ID/IG ratio increases from 0.4 to 0.95, which indicates appearance of a large number of defects in graphene.

Table 2 Component and hardness of the G–Ni composite
Full size table
Fig. 1
figure 1

a XRD patterns and b Raman spectra of the composite before and after annealing. (Color figure online)

Full size image

3.2 Effect of Counterparts on the Tribological Behavior

Figure 2 shows the mean friction coefficient and wear rate of the G–Ni composite sliding against different counterparts at 2 N normal load. It can be found that counterparts have great influences on the friction coefficient and wear rate of the composite. The friction coefficient of the composite sliding against GCr 15 ball is about 0.74, which is quite similar to that against Si3N4 ball. Meanwhile, the composite sliding against ZrO2 ball exhibits the lowest friction coefficient of 0.52. The wear rate of the composite is the highest (3.15 × 10−4 mm3/Nm) when sliding against GCr 15 ball, which is about 3 times higher than that sliding against Si3N4 ball (1.07 × 10−4 mm3/Nm) and ZrO2 ball (9.58 × 10−5 mm3/Nm).

Fig. 2
figure 2

Friction coefficient and wear rate of the G–Ni composite sliding against different counterparts (The error bars represent the minimum and maximum values)

Full size image

Figure 3 shows the three-dimensional (3D) topographies and profiles of wear tracks of G–Ni composite sliding against different counterparts. From the cross-section profiles, it is seen that the wear tracks are deeper and broader with obvious grooves or scratch on the worn surface when sliding against GCr 15 (Fig. 3a). The roughness (Ra) of the worn surface is about 1.33 µm, which is higher than that of the others. The GCr 15 ball is severely worn with wear scar of about 650 µm, meanwhile the ZrO2 ball is only slightly worn with smaller and smoother scar (Fig. 4). Hence, it is concluded that the G–Ni/ZrO2 tribo-pair exhibits the excellent self-lubricating behavior.

Fig. 3
figure 3

3D topographies and profiles of the wear tracks of G–Ni composite sliding against counterparts: a GCr 15, b Si3N4 c ZrO2 (black line is sectional profiles of the wear tracks and red line is surface profiles of wear tracks). (Color figure online)

Full size image
Fig. 4
figure 4

3D topographies of different counterparts

Full size image

Figure 5 shows the SEM images of worn surface of the G–Ni composite sliding against different counterparts. For G–Ni/GCr 15 pair, many grooves exist on the worn surface along the sliding direction (Fig. 5a), which reveals the three-body abrasive wear due to fine and hard debris. The plastic deformation and adhesive characteristic are also clearly observed on the worn surface sliding against GCr 15 ball. Besides, several micro-cracks are observed on the worn surface, which may be the pores induced brittle fracture. When coupling with Si3N4 and ZrO2 balls, worn morphologies of the composite are quiet similar. Some peelings and delamination are clearly seen on worn surface, as shown in Fig. 5b, c. Table 3 lists the chemical compositions in regions A, B, and C marked by red rectangles in Fig. 5. All of the worn surfaces have high carbon content, indicating the presence of carbon-rich tribofilms derived from graphene on the worn surface. The Fe, Si, and Zr element from counterparts can also be detected on the worn surface due to tribo-chemical reaction of the interface between G–Ni composite and counterparts. Especially, it can be found that the O content in worn surface sliding against Si4N3 and ZrO2 is much higher than that sliding against GCr 15, which provides a proof the interfacial tribo-chemistry. The XPS results confirmed that tribo-chemical reactions resulted in the formation of metal oxides such as NiO and Ni(OH)2 on worn surface in a previous paper [16]. Besides, it is proposed that and GCr 15 steel may react with oxygen and water molecules in ambient air with about 40% humidity under the shear stress and friction heat as follows [17, 18]:

Fig. 5
figure 5

SEM images of worn morphologies of the G–Ni composite sliding against different counterparts

Full size image
Table 3 Chemical compositions of worn surface analyzed by EDS in Fig. 5
Full size table
$${\text{4Fe}}+{\text{3}}{{\text{O}}_{\text{2}}}+{\text{6}}{{\text{H}}_{\text{2}}}{\text{O}}={\text{4Fe}}{\left( {{\text{OH}}} \right)_{\text{3}}}$$
(1)
$$\Delta G_{f}^{{298}}=~ - 360.8\,{\text{KJ/mol}},$$

where \(\Delta G_{f}^{{298}}\) is the Gibbs free energy of corresponding reaction at 298 K. The Fe(OH)3 products were much unstable and easy to be further oxidized to form Fe2O3, Fe3O4 and FeOOH components [18]. The fine and hard iron oxide can act as third-body at frictional interface leading to severe abrasive wear (see Fig. 5a). As a result, the G–Ni/GCr 15 pair exhibits a high friction coefficient and wear rate. As for ceramic counterpart materials such as Si3N4 and ZrO2, Si (OH)4 and Zr (OH)4 could be formed by the following tribo-chemical reactions [18,19,20]:

$${\text{S}}{{\text{i}}_{\text{3}}}{{\text{N}}_{\text{4}}}+{\text{6}}{{\text{H}}_{\text{2}}}{\text{O }}={\text{3Si}}{{\text{O}}_{\text{2}}}+{\text{4N}}{{\text{H}}_{\text{3}}}$$
(2)
$$\Delta G_{f}^{{298}}=~ - 570.4\,{\text{KJ/mol}}$$
$${\text{Si}}{{\text{O}}_{\text{2}}}+{\text{2}}{{\text{H}}_{\text{2}}}{\text{O}}={\text{Si }}{\left( {{\text{OH}}} \right)_{\text{4}}}$$
(3)
$${\text{Zr}}{{\text{O}}_{\text{2}}}+{\text{2}}{{\text{H}}_{\text{2}}}{\text{O}}={\text{Zr }}{\left( {{\text{OH}}} \right)_{\text{4}}}$$
(4)
$$\Delta G_{f}^{{298}} = - 372.9\,{\text{KJ/mol}}.$$

The Si(OH)4 and Zr (OH)4 gel have been proven to play an effective tribo-lubricants [21,22,23,24], which could prevent the scratch and adhesion (see Fig. 5b, c), reducing wear of the G–Ni composite. According to the above reactions, it seems that Si3N4 could more easily react with H2O than ZrO2. But the tribo-chemical reactions are related to temperature of the frictional surface. The surface temperature of ZrO2 is higher because thermal conductivity of ZrO2 (2.5 W/mK) is more than that of Si3N4 (12.5 W/mK) [19]. The higher temperature is favorable to the formation of products, which interprets the higher content of Zr on the worn surface from the reaction. Accordingly, G–Ni/ZrO2 pair exhibits the lowest friction coefficient and wear rate than other tribo-pairs.

3.3 Effect of Applied Load on the Tribological Behavior

Figure 6a shows the curves of friction coefficient of the G–Ni composite sliding against ZrO2 counterpart under different loads. The steady-state friction coefficient decreases from 0.50 to 0.41 as the load increases from 2 N to 6 N, and then it abnormally increases to 0.53 with the quite large fluctuation as the load increases up to 8 N. The similar tendency was reported by other groups for TiAl and Al matrix composites [24, 25]. Here, this dependence of friction coefficient at load could be attributed to the formation and breakdown of tribofilms on worn surface, which will be discussed in detail later. It can be seen from Fig. 6b that the wear rates are approximately 1.0 × 10−4 mm3/Nm without significant changes at low loads, whereas it sharply decreases to 5.5 × 10−5 mm3/Nm at high loads, which is about 45% lower than that at low loads.

Fig. 6
figure 6

a Friction coefficient and b wear rate of the G–Ni sliding against ZrO2 counterpart under different loads (The error bars represent the minimum and maximum values). (Color figure online)

Full size image

Figure 7 shows topographies and profiles of the wear tracks of G–Ni composite sliding against ZrO2 under different loads. It is seen that the wear track became deeper, and the corresponding wear scar became greater with increasing load (Fig. 8). Interestingly, the roughness of the worn surface first decreases and then increases with increasing load, and reaches a minimum at load of 6 N, which presents the same trend with friction coefficient.

Fig. 7
figure 7

3D topographies and profiles of the wear tracks of G–Ni composite sliding against ZrO2 under different loads (black line is sectional profiles of the wear tracks and red line is surface profiles of wear tracks). (Color figure online)

Full size image
Fig. 8
figure 8

3D topographies of ZrO2 under different loads

Full size image

Figure 9 shows SEM images of worn surface of the G–Ni composite sliding against ZrO2 under different loads. The worn surfaces become gradually flatter and smoother when load increases from 2 N to 6 N, which is in well agreement with the above-measured roughness. Some smooth regions on the worn surface are ascribed to the formation of tribofilms from the product of tribo-chemical reaction. This is supported by EDS analysis of worn surface. From Table 4, it can be found that the O and Zr content in tribofilms are much higher than that in peeling region. The smooth and uniformly thick tribofilms cover surface as load increases to 6 N, which can provide an effective lubrication and prevent from the damaging composite matrix. Therefore, the G–Ni composite has the lowest friction coefficient. However, the higher loads could cause the cracking and peeling-off of the tribofilms due to low toughness of metal oxides. The damaged tribofilms and wear debris were aggressively ground by ZrO2 ball, forming a large number of delamination and steps on worn surface with higher surface roughness (Fig. 9 and Fig. 7d). Such a rough worn surface results in an increase of friction coefficient at load of 8 N. From this point of view, we can explain well the variations of initial running-in stage in Fig. 6. It can be seen that the running-in stage decreases from 300 to 70 s with increasing load, because the higher temperature at higher load can facilitate the formation of tribofilms. Nevertheless, the friction curve presents a long running-in stage with large fluctuations at a load of 8 N, suggesting that the breakdown of tribofilms play a negative role. Moreover, Raman spectrum acquired from the worn surface at load of 6 N is shown in Fig. 10. The broad peak between 800 and 1800 cm−1 can be attributed to the tribo-induced carbon product derived from graphene. The change of Raman spectrum before and after rubbing reveals the structural transformations of graphene due to tribo-chemical reactions on worn surface. The tribo-chemical mechanisms of graphene were discussed in detail in a recent paper [16]. The weak broad peak between 400 and 700 cm−1 can be assigned to Zr(OH)4 stretching mode [26]. This result provides the direct evidence that the tribofilms come from the tribo-induced chemical products.

Fig. 9
figure 9

SEM images of worn morphologies of the G–Ni composite sliding against ZrO2 under different loads

Full size image
Table 4 Chemical compositions of worn surface analyzed by EDS in Fig. 9
Full size table
Fig. 10
figure 10

Raman spectrum of worn surface at load of 6 N

Full size image

4 Conclusions

Effect of counterparts and applied load on the tribological behavior of the G–Ni matrix self-lubricating composite was investigated in this study. The main results were summarized as follows:

The G–Ni composite sliding against ZrO2 ball exhibits optimized self-lubricating properties with the lowest friction coefficient of 0.41 as well as the lowest wear resistance of 5.5 × 10−5 mm3/Nm, owing to lubricating and protection of the tribofilms derived from tribo-chemical reactions at the interface. The worn surface shows the severe peelings and delamination for G–Ni/ZrO2 pair. The wear rate of the composite sliding against GCr 15 ball is about 3 times higher than that against Si3N4 and ZrO2 ball due to the severe adhesion and three-body abrasive wear. The friction coefficient first decreases to a minimum, and then abnormally increases associating a large fluctuation with the increase of the load. This dependence of tribological behavior on the load is well interpreted in terms of the formation and breakdown of tribofilms on worn surface.