Fabrication of Copper Matrix Composites Reinforced with Carbon Nanotubes Using an Innovational Self-Reduction Molecular-Level-Mixing Method

An innovational self-reduction molecular-level-mixing method was proposed as a simplified manufacturing technique for the production of carbon nanotube copper matrix composites (CNT/Cu). Copper matrix composites reinforced with varying amounts of (0.1, 0.3, 0.5 and 0.7 wt%) carbon nanotubes were fabricated by using this method combined with hot-pressing sintering technology. The surface structure and elemental distribution during the preparation of CNT/Cu mixing powder were investigated. The microstructure and comprehensive properties of the CNT/Cu composites were examined by metallography, mechanical and electrical conductivity tests. The results revealed that the CNT/Cu could be produced by a high temperature reaction at 900 degrees under vacuum, during which the carbon atoms in the carbon nanotubes reduced the divalent copper on the surface to zero-valent copper monomers. The decrease in the ratio of D and G peaks on the Raman spectra indicated that the defective spots on the carbon nanotubes were wrapped and covered by the copper atoms after a self-reduction reaction. The prepared CNT/Cu powders were uniformly embedded in the grain boundaries of the copper matrix materials and effectively hindered the tensile fracture. The overall characteristics of the CNT/Cu composites steadily increased with increasing CNT until the maximum at 0.7 wt%. The performance was achieved with a hardness of 86.1 HV, an electrical conductivity of 81.8% IACS, and tensile strength of 227.5 MPa.


Introduction
Copper plays an essential function in life because of its good electrical and thermal properties and is extensively applied in electronics, electrical, construction, military and other fields [1]. Accordingly, an increasing amount of research is being devoted to the exploitation of copper-based composite [2,3] materials. A nanomaterial called a carbon nanotube (CNT) [4,5] has remarkable mechanical, electrical and thermal properties [6,7] due to its high strength and specific modulus. Therefore, it is an excellent reinforcing phase for the fabrication of new copper-based composites, and it has great potential in its application prospects [8]. Combining the features of these two materials, carbon nanotube copper matrix composites (CNT/Cu) [9] present a series of advantages, such as good mechanical hardness [10,11], electrical [12][13][14] and thermal conductivity [15,16], friction and wear resistance [17,18], and compression and tensile resistance [19][20][21] properties. It has been reported that the resulting microstructural arrays have an important role upon the material's distinctive properties [22]. Additionally, the distributed particles also have positive effects on the resulting mechanical behavior of the composites [23].
However, the poor interfacial wetting ability of CNT and the clustering effect in the metal matrix limit the application of CNT in composites. The presence of defects such as inclusions and pores in clusters can degrade the performance of the composites. Recently, compared to the conventional preparation process, it has been well demonstrated that the molecular level mixing (MLM) method [24][25][26] successfully improved the interfacial bonding [27,28] and dispersion [29] of CNT with substrates at the molecular level by using surface modification [30]. Unfortunately, the requirement of using hydrogen as a reducing agent [31,32] in a variety of studies is accompanied by serious safety hazards and a complicated preparation process. A more convenient and efficient restoration method has not yet been studied.
In this paper, an innovative molecular-level-mixing self-reduction method was proposed and validated. Utilizing the reducibility of carbon atoms, the copper oxide on the surface of CNT was reduced to copper under high temperature and vacuum conditions. The feasibility and regularity of the self-reduction process was investigated and discussed through verification of characterizing CNT for different stages of the process by X-ray diffraction (XRD), infrared spectroscopy, optical microscopy, scanning electron microscope (SEM) and transmission electron microscope (TEM). The molecular-level-dispersed CNT/Cu composites with different CNT contents were successfully prepared by ballmilling [33,34] and hot-pressing processes [35,36]. The conductivity, mechanical properties and tensile behavior [37] of the fabricated composites were tested and analyzed.

Preparation Procedure
The CNT/Cu powder was subsequently obtained by the specific experimental schematic in Figure 1. Firstly, carboxylated multi-walled carbon nanotubes were mixed with saturated copper acetate alcohol solution and sonicated for 2 h by an Ultrasonic Cleaner (model: YL-060ST). Then, the suspension was filtered and dried at 80 • C (±2 • C) for 24 h. After grinding, the powder was oxidized at 280 • C (±5 • C) for 5 h to obtain carbon nanotube/copper oxide (CNT/CuO) powder. Finally, the CNT/CuO powder was subjected to self-reduction reactions under vacuum conditions at 80, 280, 800, and 900 degrees with a holding time of 2 h. The equipment used was a vacuum melting furnace (model: HZCZ-240).
The CNT/Cu composites were produced by using the mechanical ball-milling method and vacuum hot-pressing method. The prepared CNT/Cu powder and copper powder were put into a planetary ball mill (model: QM-3SP4, from Nanjing University Instrument Factory, Nanjing, China) with a ball-to-material ratio of 5:1. The rotating speed was 200 r/min. After 4 h of mechanical ball milling, a uniformly mixed composite powder was obtained. After that, the powder was heated to a temperature of 700 • C (±10 • C) at a heating rate of 10 • C/min and sintered at a pressure of 30 MPa. The holding period was 6 h. Finally, four groups of CNT/Cu composites were completed, containing CNT at 0.1, 0.3, 0.5, and 0.7 wt%. Materials 2022, 15, x FOR PEER REVIEW

Characterization
A Fourier transform near-infrared spectrometer (model: MPA0304040) and confocal micro-Raman spectrometer (inVia Qontor 03040405) were used to exami molecular structure and the structural integrity of the surface of the composite po at different stages of treatment. CNTs/CuO and carbon nanotube powder samples analyzed for the physical phase of the samples using an X-ray diffractometer (m XRD-600003030502). The morphology of the CNTs/Cu composite samples were eva by metallographic microscopy (Leica DMi8) and the hardness, compression and t properties of the composites were tested by using a universal testing machine (CSS-Five data points for hardness and conductivity were taken and averaged. The len the tensile specimen was 40 mm, the width was 10 mm, and the thickness was 3 mm tensile test was performed at a strain rate of 3.7 × 10 −4 s −1 . Molecular binding of cop carbon nanotubes at the microscopic level was observed by using field emission tran sion electron microscope (JEM-2100F, from JEOL, Tokyo, Japan) and field emission ning electron microscopy (SU5000, from HITAXHI, Tokyo, Japan). Figure 2 shows the comparison of the IR spectra of the powders in two states. the carbon nanotube powder mixed with saturated copper acetate alcohol solution ultrasonic treatment (CNT + C4H8CuO5), and the other is the original carbon nan powder (CNT). From the figure, it can be observed that CNT + C4H8CuO5 exhibi obvious peak segment of Cu 2+ ions after sonication. This situation proves that the c nation of Cu 2+ ions and functional groups on the surface of CNT could be achiev using ultrasonication.

Characterization
A Fourier transform near-infrared spectrometer (model: MPA0304040) and Laser confocal micro-Raman spectrometer (inVia Qontor 03040405) were used to examine the molecular structure and the structural integrity of the surface of the composite powders at different stages of treatment. CNTs/CuO and carbon nanotube powder samples were analyzed for the physical phase of the samples using an X-ray diffractometer (model: XRD-600003030502). The morphology of the CNTs/Cu composite samples were evaluated by metallographic microscopy (Leica DMi8) and the hardness, compression and tensile properties of the composites were tested by using a universal testing machine (CSS-44200). Five data points for hardness and conductivity were taken and averaged. The length of the tensile specimen was 40 mm, the width was 10 mm, and the thickness was 3 mm. The tensile test was performed at a strain rate of 3.7 × 10 −4 s −1 . Molecular binding of copper to carbon nanotubes at the microscopic level was observed by using field emission transmission electron microscope (JEM-2100F, from JEOL, Tokyo, Japan) and field emission scanning electron microscopy (SU5000, from HITAXHI, Tokyo, Japan). Figure 2 shows the comparison of the IR spectra of the powders in two states. One is the carbon nanotube powder mixed with saturated copper acetate alcohol solution after ultrasonic treatment (CNT + C 4 H 8 CuO 5 ), and the other is the original carbon nanotube powder (CNT). From the figure, it can be observed that CNT + C 4 H 8 CuO 5 exhibited an obvious peak segment of Cu 2+ ions after sonication. This situation proves that the combination of Cu 2+ ions and functional groups on the surface of CNT could be achieved by using ultrasonication.

Characterization of CNT/Cu Composite Powder
The XRD patterns of the samples of CNT/Cu powder at different stages of the selfreducing molecular-level-mixing process are shown in Figure 3. The corresponding peak of CNT (002) at 2θ = 25.9 • appeared in all four groups of samples. The structure of CNT remained in a stable state and was not disrupted under the low to high temperature process conditions. After oxidization at 280 • C, the diffraction peak of CuO appeared on the curve in addition to the diffraction peak of CNT; the diffraction peak of CuO still existed in the plot when the reduction temperature was 800 • C, but the diffraction peak of Cu 2 O also existed, which means that the reduction process had already begun, and divalent copper had been partially reduced to univalent copper; the XRD plot displayed the diffraction peak of a copper monomer when the self-reduction reaction temperature was 900 • C. With the increase in temperature, the metal structure attached to the surface of CNT reacted from divalent Cu ions to Cu. It was apparent that the temperature strongly influenced the formation of the powders. The self-reduction method was feasible fpr achieving the Materials 2022, 15, 6488 4 of 11 molecular-level mixing of carbon nanotubes and copper, but it needed to reach a sufficiently high temperature because the reaction did not proceed sufficiently at temperatures below 900 • C. Therefore, the obvious self-reduction reaction appeared only when the temperature reached at least 900 • C. The XRD patterns of the samples of CNT/Cu powder at different stages of the reducing molecular-level-mixing process are shown in Figure 3. The corresponding of CNT (002) at 2θ = 25.9° appeared in all four groups of samples. The structure of remained in a stable state and was not disrupted under the low to high temperature cess conditions. After oxidization at 280 °C , the diffraction peak of CuO appeared o curve in addition to the diffraction peak of CNT; the diffraction peak of CuO still ex in the plot when the reduction temperature was 800 °C , but the diffraction peak of C also existed, which means that the reduction process had already begun, and div copper had been partially reduced to univalent copper; the XRD plot displayed the fraction peak of a copper monomer when the self-reduction reaction temperature 900 °C . With the increase in temperature, the metal structure attached to the surfa CNT reacted from divalent Cu ions to Cu. It was apparent that the temperature stro influenced the formation of the powders. The self-reduction method was feasibl achieving the molecular-level mixing of carbon nanotubes and copper, but it need reach a sufficiently high temperature because the reaction did not proceed sufficien temperatures below 900 °C . Therefore, the obvious self-reduction reaction appeared when the temperature reached at least 900 °C .  The XRD patterns of the samples of CNT/Cu powder at different stages of the selfreducing molecular-level-mixing process are shown in Figure 3. The corresponding peak of CNT (002) at 2θ = 25.9° appeared in all four groups of samples. The structure of CNT remained in a stable state and was not disrupted under the low to high temperature process conditions. After oxidization at 280 °C , the diffraction peak of CuO appeared on the curve in addition to the diffraction peak of CNT; the diffraction peak of CuO still existed in the plot when the reduction temperature was 800 °C , but the diffraction peak of Cu2O also existed, which means that the reduction process had already begun, and divalent copper had been partially reduced to univalent copper; the XRD plot displayed the diffraction peak of a copper monomer when the self-reduction reaction temperature was 900 °C . With the increase in temperature, the metal structure attached to the surface of CNT reacted from divalent Cu ions to Cu. It was apparent that the temperature strongly influenced the formation of the powders. The self-reduction method was feasible fpr achieving the molecular-level mixing of carbon nanotubes and copper, but it needed to reach a sufficiently high temperature because the reaction did not proceed sufficiently at temperatures below 900 °C . Therefore, the obvious self-reduction reaction appeared only when the temperature reached at least 900 °C . The four groups of sample powders with different treatments were analyzed by Raman spectroscopy, and the results are shown in Figure 4. R is the ratio of I D /I G , which means the intensity ratio between the D-peak and G-peak in the Raman spectra. This ratio can be used to describe the intensity relationship between these two peaks. A larger ratio indicates that the more defects there are in the CNT, the more unstable the structure. It can be seen from the figure that the ratio of the peaks of carboxylated CNT was the lowest, and the ratio increased gradually with the sonication and oxidation treatment. This is attributed to the damage of the CNT surface by the shock in the sonication. Alternatively, Cu 2+ binds to functional groups at the molecular level resulting in defect problems due to the uneven alignment of carbon atoms. However, the R value of the powders after self-reduction reaction decreased instead. It can be inferred that the defective spots after the original surface treatment were wrapped and uniformly covered by the copper. This is the primary reason for the decrease in the intensity ratio of the D and G peaks. and the ratio increased gradually with the sonication and oxidation treatment. This is attributed to the damage of the CNT surface by the shock in the sonication. Alternatively, Cu 2+ binds to functional groups at the molecular level resulting in defect problems due to the uneven alignment of carbon atoms. However, the R value of the powders after selfreduction reaction decreased instead. It can be inferred that the defective spots after the original surface treatment were wrapped and uniformly covered by the copper. This is the primary reason for the decrease in the intensity ratio of the D and G peaks.  (Figure 5c) as copper monomers. In addition, it was visible that the deposited copper particles in Figure 6a,b gathered together to form spherical metal particles. These particles were attached to the surface of CNT, forming a point-like distribution structure. As previously judged, when the reduction reaction of carbon atoms with copper oxide occured at high temperature, the surface functional groups gradually wetted the surface. The produced copper partially wrapped the dotted sites of carbon atoms after the reaction (Figure 6d). The interface between the copper and carbon was well bonded and tightly connected (Figure 6f). Moreover, the lattice stripes with inconsistent orientation directions of metallic Cu can be observed through Figure 6e. By analyzing and comparing the SEM and TEM images, it was confirmed that the molecular-level self-reduction reaction at 900 °C could be used to obtain CNT/Cu powders with good interfacial bonding.  (Figure 5c) as copper monomers. In addition, it was visible that the deposited copper particles in Figure 6a,b gathered together to form spherical metal particles. These particles were attached to the surface of CNT, forming a point-like distribution structure. As previously judged, when the reduction reaction of carbon atoms with copper oxide occured at high temperature, the surface functional groups gradually wetted the surface. The produced copper partially wrapped the dotted sites of carbon atoms after the reaction (Figure 6d). The interface between the copper and carbon was well bonded and tightly connected (Figure 6f). Moreover, the lattice stripes with inconsistent orientation directions of metallic Cu can be observed through Figure 6e. By analyzing and comparing the SEM and TEM images, it was confirmed that the molecular-level self-reduction reaction at 900 • C could be used to obtain CNT/Cu powders with good interfacial bonding.   It can be observed that a large area was covered by carbon, accounting for 90%. The uniformly attached spherical clusters are Cu elements, accounting for about 8%. The surface of the CNT was slightly oxidized, accounting for about 2%. (e,f) The boundary junction between copper and carbon tube was tightly fitted, with large lattice stripes of copper clearly visible.

Morphology and Mechanical Properties of CNT/Cu Composites
The microstructures of the CNT-reinforced copper matrix composites with different contents of CNT by using a combination of the molecular-level-mixing self-reduction method and hot-pressing process are shown in Figure 7. The well-defined grain boundaries and complete coarse grains are obviously visible in Figure 7a. The black dot-like distribution of the CNT particles is regularly and uniformly dispersed and embedded in the grain boundaries. On the contrary, the grains were refined with the increase in CNT content (Figure 7b,c). However, black-phase carbon aggregation appeared with increasing the content of CNT (Figure 7d). Furthermore, pores, inclusions, and cracking defects tended to appear at the locations where carbon nanotubes were aggregated. Therefore, the results indicate that a limited amount of CNT could effectively refine the grains, while an excessive increase in the content of CNT leads to aggregation. This problem can be alleviated by adjusting the ball-milling process. grain boundaries. On the contrary, the grains were refined with the increase in CNT content (Figure 7b,c). However, black-phase carbon aggregation appeared with increasing the content of CNT (Figure 7d). Furthermore, pores, inclusions, and cracking defects tended to appear at the locations where carbon nanotubes were aggregated. Therefore, the results indicate that a limited amount of CNT could effectively refine the grains, while an excessive increase in the content of CNT leads to aggregation. This problem can be alleviated by adjusting the ball-milling process. where ρ c : density of the composite material; ρ m : density of the substrate; ρ f : density of the enhanced system; and w m : Mass fraction of matrix in the composite material, theoretically, it is known that the higher the CNT content, the lower the density of the material. When the CNT content was 0.1 wt%, 0.3 wt% and 0.5 wt%, the CNT was distributed at the grain boundaries, refining the metal grains and making the structure compact. As a result, the density gradually increased. However, when the CNT content reached 0.7 wt%, the According to the density formula of composite materials as shown in the equation, where ρ c : density of the composite material; ρ m : density of the substrate; ρ f : density of the enhanced system; and w m : Mass fraction of matrix in the composite material, theoretically, it is known that the higher the CNT content, the lower the density of the material. When the CNT content was 0.1 wt%, 0.3 wt% and 0.5 wt%, the CNT was distributed at the grain boundaries, refining the metal grains and making the structure compact. As a result, the density gradually increased. However, when the CNT content reached 0.7 wt%, the high content of CNT was more likely to agglomerate, leading to some tiny pores in the middle of the material, which made a slight decrease in density.
The results shown in Figure 8 visually represent that the hardness increased rapidly with the increase in CNT content, reaching a maximum hardness of 86.1 HV at 0.7 wt%, which was 58.9% higher than that at 0.1 wt%. The carbon nanotubes as reinforcing materials are equivalent to "fibers" firmly adhered to the inside of the matrix, hindering the dislocation movement and enhancing the performance of the composite. Although the conductivity continued to increase when the CNT content increased from 0.1 wt% to 0.3 wt%, the conductivity decreased slightly at 0.7 wt% due to the increased porosity of the composites. Defects hinder electron migration, which largely affects the conductivity of the copper matrix composites. It can be seen that there was a regular increase in yield strength and tensile strength with increasing the CNT content. The performance surges and elongation were the best at 0.7 wt%. strain curve (Figure 9) highlights the problem that the poor dispersion of CNT in the metal matrix lead to increased agglomeration phenomena and defects in a small area. The sample had difficulty in withstanding higher tensile forces and fractured prematurely, so that the plasticity of the material decreased significantly. By optimizing the hot-pressing process, the dispersion of CNT and the properties of CNT/Cu composites could be further improved.  However, the low elongation at 0.5 wt% of the carbon nanotube content in the stressstrain curve (Figure 9) highlights the problem that the poor dispersion of CNT in the metal matrix lead to increased agglomeration phenomena and defects in a small area. The sample had difficulty in withstanding higher tensile forces and fractured prematurely, so that the plasticity of the material decreased significantly. By optimizing the hot-pressing process, the dispersion of CNT and the properties of CNT/Cu composites could be further improved.  Figure 10 shows the morphological characteristics of the fracture of the tensile sample. At a CNT content of only 0.1 wt%, the fracture level was a relatively flat plane interlocking and no apparent presence of CNT can be observed (Figure 10a). When the content of CNT increased to 0.3 wt%, the fracture morphology appeared to be an obvious microporous  Figure 10 shows the morphological characteristics of the fracture of the tensile sample. At a CNT content of only 0.1 wt%, the fracture level was a relatively flat plane interlocking and no apparent presence of CNT can be observed (Figure 10a). When the content of CNT increased to 0.3 wt%, the fracture morphology appeared to be an obvious microporous aggregation type ductile fracture with unevenness ( Figure 10b). The microscopic morphology of the fracture was honeycomb-like, and the fracture surface consisted of a significant amount of tiny tough nests, which were about 1µm black nest pits (Figure 10c). A large number of CNT was found to be aggregated at the fracture site as the content increased (Figure 10d), with some clustered together (Figure 10e). The agglomerated carbon nanotubes were defective weak parts in the material, which were easy to fracture. The fracture surface had CNT agglomeration and entanglement phenomena. Plenty of CNT extended inside the matrix to bear the load through the interior of the matrix (Figure 10e), which effectively increased the tensile strength.

Conclusions
(1) By performing the self-reduction reaction of CNT/CuO at different temperatur was shown that the preparation of CNT/Cu could be achieved under vacuu 900 °C. The copper bonding was tight on the surface of CNT without damage to overall structure. (2) With the increase in CNT content (0.1 wt%, 0.3 wt%, and 0.5 wt%), the copper m composites could progressively improve the density and electrical conducti However, the electrical conductivity and density of the composite with 0.7 wt% C content was reduced. The agglomeration defects were the most significant prob leading to cracking, which is detrimental to the performance. This problem ca alleviated by adjusting the ball-milling process.

Conclusions
(1) By performing the self-reduction reaction of CNT/CuO at different temperatures, it was shown that the preparation of CNT/Cu could be achieved under vacuum at 900 • C. The copper bonding was tight on the surface of CNT without damage to the overall structure. (2) With the increase in CNT content (0.1 wt%, 0.3 wt%, and 0.5 wt%), the copper matrix composites could progressively improve the density and electrical conductivity. However, the electrical conductivity and density of the composite with 0.7 wt% CNT content was reduced. The agglomeration defects were the most significant problem leading to cracking, which is detrimental to the performance. This problem can be alleviated by adjusting the ball-milling process. (3) With the increase in CNT content (0.1 wt%, 0.3 wt%, 0.5 wt%, and 0.7 wt%), the composites could effectively resist mechanical forces. The highest elongation at break was achieved with a CNT content of 0.7 wt%. Both yield strength and ultimate strength were enhanced with a rise in CNT content from 0.1 wt% to 0.7 wt%.