3.1 Microstructure analysis
The SEM observation is conducted on the mixed powder after grinding, as shown in Fig. 1. Most of the aluminum particles in the figure are spherical or elliptic, indicating that there is no obvious coarsening of aluminum powder during ball grinding. The results show that no graphene agglomeration is found in the mixed powder. At high magnification, as shown by the red arrow in Fig. 1b), embedded and suspended graphene were observed on the particle surface. It can be considered that the shear stress of the stainless steel ball makes graphene embedded into the aluminum powder particles during the high-energy ball grinding process, effectively preventing the agglomeration of graphene.
Moreover, the surface morphology of the prepared GAMCs is observed, as shown in Fig. 2. According to Fig. 2a), the surface morphology at low magnification suggests that the grain size is relatively uniform and there is no abnormally coarse grain. No cavity or micro-crack is found on the surface of the material by high energy ball grinding and powder metallurgy. Surface morphology at high multiples and the interface is tightly bound, as indicated in Fig. 2b). Meanwhile, Energy Dispersive Spectrometer (EDS) is used to analyze the elements and content of the selected location. Point A and point B represent grains and grain boundaries respectively. Their element distribution and content are clearly visible on the right side of the image. It can be seen that there is only Al and C content at the grain A, which proves that there is no oxygen on the grain surface and inside. This kind of situation means there is no oxidation of aluminum. However, the oxygen content at grain boundary point B is detected as 3.72%, indicating a high probability of alumina formation. The oxygen element exists because the air in the powder gap is not easy to be discharged during the process of pressing powder.
Figure 3 depicts the X-ray diffraction of the sintered sample. The diffraction peaks of sample are near 38°, 44°, 65°, 78° and 82°, without significant deviation. The corresponding crystal planes are (111), (200), (220), (311) and (222), respectively. The absence of a diffraction peak indicates that there is no agglomeration of graphene. The GAMCs are prepared successfully by high energy ball milling and powder metallurgy. At the same time, the peak of alumina is not shown, which is helpful for the next experiment.
Figure 4 reveals the representative microscopic structure of aluminum matrix composites. According to the Fig. 4a), the gray area is matrix aluminum and the grain boundary is clearly visible. The graphene can be observed at the grain boundary. Some black matter in the matrix may be the product of the reaction between the matrix and graphene or oxygen in the preparation process. In order to determine its composition, the circular region is subjected to electron diffraction. The selected electron diffraction spots are obtained and the d value is measured. Then it is compared with the standard card. The results show that the diffraction spots corresponded to the crystal faces of AlC3: (\(\stackrel{-}{2}\)00), (\(\stackrel{-}{1}\)1\(\stackrel{-}{1}\)) and (11\(\stackrel{-}{1}\)), as presented in Fig. 4b). Figure 4c) clearly displays the direction of the [110] crystal band axis in the AlC3 crystal unit. In addition, it is obvious from Fig. 4d) that the position of (\(\stackrel{-}{1}\)1\(\stackrel{-}{1}\)) and (11\(\stackrel{-}{1}\)) crystal planes at the equator plane can be realized more intuitively by using the method of stereographic projection. They are located to the left and right of the central axis, respectively. After careful comparison and analysis, the calibrated crystal indices are self-consistent, proving that the black matter is AlC3. In the existing research, AlC3 is easy to be produced when most carbon materials are used for reinforcement of aluminum matrix composites [15–17]. The reaction process of graphene and aluminum can be analyzed from the perspective of thermodynamics. The chemical reaction formula and the relative free energy formula are as follows [18]:
$$\frac{\text{4}}{\text{3}}\text{Al+C}\text{=}\frac{\text{1}}{\text{3}}{\text{Al}}_{\text{4}}{\text{C}}_{\text{3}}\text{, }\text{∆}{\text{G}}_{{\text{Al}}_{\text{4}}{\text{C}}_{\text{3}}}^{\text{free}}\text{=∆}{\text{G}}_{{\text{Al}}_{\text{4}}{\text{C}}_{\text{3}}}^{\text{0}}\text{+}\frac{\text{1}}{\text{3}}\text{RT}\text{ln}\frac{{\text{a}}_{{\text{Al}}_{\text{4}}{\text{C}}_{\text{3}}}}{{\left({\text{a}}_{\text{Al}}\right)}^{\text{4}}}$$
1
where the \(\varDelta {\text{G}}^{0}\) refers to the standard free energy of formation per mol of carbon, \(\text{a}\) stands for activity, expressed in atomic fractions under ideal conditions [18]. R is the ideal gas constant and T is the absolute temperature. Similarly, the chemical reactions and relative free energies of AlC3 in the form above can be expressed as below:
$$\frac{\text{1}}{\text{3}}\text{Al+C}\text{=}\frac{\text{1}}{\text{3}}\text{Al}{\text{C}}_{\text{3}}\text{, }\text{∆}{\text{G}}_{\text{Al}{\text{C}}_{\text{3}}}^{\text{free}}\text{=∆}{\text{G}}_{\text{Al}{\text{C}}_{\text{3}}}^{\text{0}}\text{+}\frac{\text{1}}{\text{3}}\text{RT}\text{ln}\frac{{\text{a}}_{\text{Al}{\text{C}}_{\text{3}}}}{{\text{a}}_{\text{Al}}}$$
2
By comparing equations (1) and (2), it can be observed that the free energy of both of them is calculated according to the reaction rate of C per mole. Therefore, this work does not consider the free energy calculation process of products and only observe the difference of individual values between the two formulas. For every mole of carbon, \(\varDelta {\text{G}}_{\text{A}\text{l}{\text{C}}_{3}}^{0}\) is less than \(\varDelta {\text{G}}_{\text{A}{\text{l}}_{4}{\text{C}}_{3}}^{0}\). R and T can be considered constant under the same circumstances. Then the size of the exponential function in the free energy equation can be compared. Since \({\text{a}}_{\text{Al}}\) is determined by atomic fraction, then \({\text{a}}_{\text{Al}}\) should be less than 1. Substituting (1) and (2), then \(\text{ln}\frac{{\text{a}}_{{\text{Al}}_{\text{4}}{\text{C}}_{\text{3}}}}{{\left({\text{a}}_{\text{Al}}\right)}^{\text{4}}}\) should be greater than \(\text{ln}\frac{{\text{a}}_{\text{Al}{\text{C}}_{\text{3}}}}{{\text{a}}_{\text{Al}}}\). Therefore, it is known that the free energy of AlC3 is less than the free energy of Al4C3. Also, the formation of carbides is related to the structural integrity of carbon materials [19]. During ball milling, graphene is more likely to form AlC3 in the material because of the impact of graphene, which tends to break the graphene bonds. As mentioned above, carbides are not easily formed during ball milling, and if any are formed, the content is very small. It is the reason that the XRD does not detect the carbide peak.
It can be observed from the Fig. 4 that the distribution of graphene powder and grain boundary do not show obvious agglomeration, indicating that the high-energy ball mill plays a good dispersion role. The graphene sheets at grain boundaries effectively inhibit grain expansion and hinder grain growth. Meanwhile, with its unique structure and excellent mechanical properties, graphene can realize load transfer from matrix to interface, effectively improving the mechanical properties of materials [10]. As mentioned in the previous section, graphene has payload transfer during frictional etching and slow down the destruction speed of the matrix surface.
In order to further study the fine structure of microstructure, the high-resolution transmission at grain boundary is shown in the Fig. 5. The distribution of graphene with fewer layers is clearly visible. GAMCs prepared by high energy ball milling and powder metallurgy have about 4–5 layers of graphene. By measurement, the thickness of single-layer graphene is about 0.334 nm. The shear stress of the stainless-steel grinding ball results in fewer layers of graphene. Graphene aggregation and multilayer graphene are not observed in the Fig. 5. The white dotted line shows the distribution of the graphene sheets. It can be seen that the curved distribution of graphene is well adapted to the orientation of grain boundaries. The high matching of graphene orientation and interfacial trend not only promotes the strong interfacial bond between graphene and matrix but also effectively inhibits the growth of grains.
3.2 Tribological corrosion properties
In the process of preparing and improving aluminum matrix composites, the wear resistance and corrosion resistance of the materials are important indexes to characterize the properties of the materials. Certain metal materials cannot avoid friction and wear during service, such as bearings, cutting tools, drilling equipment and certain precision structures[20–22]. On the one hand, these frictions are conducive to the smooth work. They affect the service life of metal materials, production cost and resource utilization[14]. The working environment of most metal materials is exposed to moist air or direct contact with liquid. The flow of air and liquid not only produces friction on the material, but also carries out continuous corrosion on it. Due to friction and wear, new contact surfaces are constantly produced on the metal surface, and the contact surface is involved in the corrosion caused by atmosphere or liquid, which aggravates the wear process. The synergy increases the surface damage rate of metal materials and seriously affects the service life of metal materials [22] .
Therefore, in order to study the friction and corrosion properties of materials, it is tested by electrochemical friction and wear coupling test system. The friction corrosion process of GAMCs (graphene accounts for 0.5 wt%) is comprehensively analyzed. Figure 6 shows the friction coefficient and OCP of GAMCs during the test. In the initial stage, the friction coefficient increases from about 0.10 to about 0.11. The maximum friction coefficient is 0.126 and the average friction coefficient is 0.027. On the whole, the friction coefficient of the material is relatively stable.
The schematic diagram of frictional corrosion is shown in Fig. 7. In corrosive environment, aluminum matrix composites tend to react with oxygen in electrolyte, forming oxide film to affect the subsequent process[23]. This film can be used as the surface strengthening layer, which is the source point of staggered enrichment and increases the internal stress of the material[24]. The strengthening layer on the initial material surface is not continuous. The friction coefficient tends to rise under the influence of load. As the friction continues, work hardening occurs on the surface of aluminum matrix composites, and the hardening area expands with time [14]. The load action promotes the surface strengthening to accumulate more energy, and the friction and wear require greater load. Under the condition of constant load, the friction coefficient of this experiment will be relatively reduced, and the friction coefficient can be as low as 0.10 or below. When the surface energy accumulates to a certain level, corrosion is more likely to occur [25, 26]. Tribo-corrosion acts on the material surface to promote the surface damage, so that the friction coefficient will be increased again for the next work hardening. In the case of constant load, the frictional corrosion is repeated, and the friction coefficient will fluctuate slightly over time, but the average friction coefficient is relatively low.
Frictional corrosion is the joint action of mechanical and chemical friction. Tribological corrosion is defined as degradation due to mechanical wear and corrosion electrochemical etching[14]. Changes in OCP are influenced by this process. A significant decrease in electric potential is observed at the start of the slide as the tribo-etched surface becomes more active under higher normal loads, resulting in lower OCP. In this process, the changes of surface energy and friction coefficient make OCP slightly fluctuate in the whole process. During the friction phase, the average CPO is about − 0.76 V. At the end of tribological corrosion, OCP will have a rising fluctuation due to the formation of tribological film and the decrease of surface energy.
Repeated friction of the surface from the alumina ball tends to generate a thin oxide film on the surface, resulting in stress concentration under load. In addition, the action of the corrosion solution causes micro-cracks in the oxide film area, as shown in Fig. 8. However, the matrix is prone to plastic deformation under high density loading times[11]. The graphene at and near the grain boundary rapidly transfers the load to the substrate near the surface[27, 28]. The plastic deformation near the crack results in the crack being deformed and squeezed, and then the crack is filled and repaired. The repaired crack prevents the electrolyte from entering the matrix, making the frictional corrosion stay only locally, slowing down the corrosion process and reducing the friction coefficient on the material surface. The reciprocating friction of GAMCs is nearly 20,000 times in 30 minutes. By analyzing the friction data, the friction coefficient and OCP potential of graphene /Al composite materials tend to be stable as a whole. GAMCs do not form in the whole process, which indicates that the material has good friction-corrosion resistance. Graphene is distributed at the grain boundary, which enhances the corrosion and friction resistance of the material, because it promotes load transfer and self-repair of the matrix.
3.3 Strengthening mechanism
In this study, the Vickers hardness of pure aluminum was 36 Hv, while GAMCs was 61.8 Hv, an increase of 41.7%. The corresponding tensile strength is 211 MPa. During the sintering process, higher sintering temperature is conducive to nucleation, grain growth, atomic amplitude expansion, atomic diffusion velocity and plastic flow at grain boundaries. The holding time of 4 h makes the atoms at the grain boundary enter into the range of atomic force, which strengthens the bonding degree of grain boundary and improves the strength of the composite obviously.
At present, the strengthening mechanisms of GAMCs are mainly as follows: fine grain strengthening [29], thermal mismatch strengthening [30], Orowan strengthening[1], shear lag strengthening[31]. This paper will analyze the strengthening effect of GAMCs from the follow strengthening mechanisms.
Based on previous studies, the preparation of GAMCs and the binding of graphene at grain boundary can effectively inhibit grain growth and hinder grain boundary expansion. Hall-Petch [29] formula is used to calculate the strengthening effect of fine grain strengthening on aluminum matrix composites. The expression is as follows:
$$\text{∆}{\text{σ}}_{\text{G}}\text{=k(}{\text{d}}^{\text{-}\frac{\text{1}}{\text{2}}}\text{-}{\text{d}}_{\text{0}}^{\text{-}\frac{\text{1}}{\text{2}}}\text{)}$$
3
where d is the average grain size of aluminum matrix composites and \({\text{d}}_{0}\) is the grain size of pure aluminum. Scanning measurement with Image Pro shows that d and \({\text{d}}_{0}\)are 5.62 µm and 11.43 µm, respectively. The k is the constant of the influence degree of grain boundary on strength (0.04\(\text{M}\text{P}\text{a}/\sqrt{\text{m}}\)[32])
The enhancement of thermal mismatch is due to the difference of thermal expansion coefficient between the reinforce and the matrix when the temperature changes. The effect of this thermal mismatch is more obvious between the normal direction perpendicular to graphene and the matrix, due to the unique two-dimensional structure of graphene. It results in the residual thermal stress at the interface between the reinforce and the matrix, accompanied by the generation of height dislocation. The coefficient of thermal expansion of graphene is 1.1×10− 6 K− 1, while the coefficient of thermal expansion of aluminum is 23×10− 6 K− 1, which is an order of magnitude different. The intensification caused by the expansion difference between graphene and aluminum base is calculated by using the model formula proposed by Arsenault R J et al. [33]:
$${\text{∆σ}}_{\text{CTE}}\text{=K}{\text{G}}_{\text{m}}\text{b}\sqrt{\text{ρ}}$$
4
where \({\text{G}}_{\text{m}}\) represents the shear modulus of aluminum (2.45×104 MPa[34]) and K is the constant maturity factor (0.5[35]). b represents the Boggs vector (0.286 nm) and ρ presents the dislocation density. Following formula gives the calculation process of density.
$$\text{ρ=}\frac{\text{12∆T∆α}{\text{f}}_{\text{v}}}{\text{(1-}{\text{f}}_{\text{v}}\text{)b}{\text{d}}_{\text{p}}}$$
5
where ∆T is the temperature change (575 K), and \(\text{∆α}\) is the difference in coefficient of thermal expansion between graphene and aluminum. \({\text{d}}_{\text{p}}\) is the average surface size of graphene and \({\text{f}}_{\text{v}}\) is the volume fraction of graphene, which is expressed as follows:
$${\text{f}}_{\text{V}}=\frac{{\text{f}}_{\text{m}}/{{\rho }}_{\text{G}\text{N}\text{S}}}{{\text{f}}_{\text{m}}/{{\rho }}_{\text{G}\text{N}\text{S}}+\left(1-{\text{f}}_{\text{m}}\right)/{{\rho }}_{\text{A}\text{l}}}\times 100\text{\%}$$
6
Accordingly, Orowan strengthening is also known as second phase strengthening. The graphene is distributed in the matrix as a fine second phase, which can act as a hindrance to dislocation movement. When the matrix is subjected to applied stress, graphene hinders dislocation movement and the dislocation is bent to form a dislocation ring, thus playing a strengthening role. In this paper, calculate its reinforcement value using the following model[36]:
$$\text{∆}{\text{σ}}_{\text{Oro}}\text{=}\frac{\text{0.13}{\text{G}}_{\text{m}}\text{b}}{\text{λ}}\text{ln}\frac{{\text{d}}_{\text{p}}}{\text{2b}}$$
7
where \(\text{λ}\) is the grain spacing (nm), which can be expressed in the following formula[37]:
$$\text{λ=}{\text{d}}_{\text{p}}\sqrt{\frac{\text{2}}{\text{3}}}\left(\text{1.25}\sqrt{\frac{\text{π}}{{\text{f}}_{\text{v}}}}\text{-2}\right)$$
8
Shear lag model proposed that the strengthening mechanism of graphene-enhanced aluminum matrix composites is that external load is transferred and dispersed through interfacial shear force. It is believed that the amount of graphene addition and the aspect/diameter ratio of graphene added in the composite material affect the properties of the composites[31]. The yield strength value of GAMCs is calculated by using this model, which is expressed as follows:
$${\text{σ}}_{\text{C}}\text{=}{\text{σ}}_{\text{0}}\text{(}\text{1}\text{+}\text{s}{\text{f}}_{\text{v}}\text{)}$$
9
where, \({\text{σ}}_{\text{0}}\)is the yield strength of the matrix (150 MPa[1]), s is the aspect ratio of graphene.
The calculation results of each enhancement effect are shown in Table 1. Due to the large size of the aluminum powder before ball grinding and the short ball grinding time, the aluminum powder is not fully crushed. The grain size is mostly in the micron level and only a small amount of fine grains exists near the grain boundary. This condition makes the calculated value of fine grain strengthening small, only 9.327 MPa. The distribution of graphene in matrix is idealized in the calculation of thermal mismatch reinforcement. In addition, the difference of coefficient of thermal expansion between them is obvious. The calculated result is 42.25 MPa, which tends to be an ideal value. This case shows that the strong effect of graphene under ideal conditions is very considerable, and the preparation process needs to be further improved. In general, graphene has great development value as a reinforcement. In terms of process innovation and preparation, effective dispersion and structural integrity of graphene are the top priorities.
Table 1
The calculated results of Graphene /Al composites
| Strengthening Mechanisms (MPa) | \(\varDelta {\sigma }_{G}\) | \({\text{∆σ}}_{\text{CTE}}\) | \(\varDelta {\sigma }_{\text{O}\text{r}\text{o}}\) | \({\text{∆σ}}_{\text{S}}\) |
| Graphene/Al composites | 9.327 | 45.25 | 26.9 | 16.89 |
| Al | | | | 150 |
In order to further illustrate the reliability of the reinforcement mechanism, shear-lag model modified by Nardone and Prewo[38] is used for calculate the yield strength.. \({{\phi }}_{\text{c}}\) is used to denote the yield strength value of GAMCs calculated by the modified model. Then the expression of the modified model can be expressed as follows:
$${\text{φ}}_{\text{c}}\text{=}\left({\text{σ}}_{\text{0}}\text{+}\text{∆}{\text{σ}}_{\text{G}}\text{+}{\text{∆σ}}_{\text{CTE}}\text{+}\text{∆}{\text{σ}}_{\text{Oro}}\right)\left(\frac{{\text{f}}_{\text{V}}\text{(s+4)}}{\text{4}}\text{+1-}{\text{f}}_{\text{V}}\right)$$
10
Referring to the relevant values, the yield strength of the material calculated by the modified model is 227.75 MPa. Compared with Formula (9), this value is slightly lower than the calculated value of the general Shear lag model (237.68 MPa). But it is closer to the yield strength value of GAMCs (211 MPa), and the experimental results are in good agreement with the calculated results.