We are IntechOpen, the world's leading publisher of Open Access books Built by scientists, for scientists

Open access books available 5,300

130,000 155M

International authors and editors

Downloads

Our authors are among the

most cited scientists 154 TOP 1%

Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI)

# Interested in publishing with us? Contact book.department@intechopen.com

Numbers displayed above are based on latest data collected. For more information visit www.intechopen.com

# **Carbon Nanotube (CNT)-Reinforced Metal Matrix Bulk Composites: Manufacturing and Evaluation**

Sebastian Suárez, Leander Reinert and Frank Mücklich

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/63886

#### **Abstract**

This chapter deals with the blending and processing methods of CNT-reinforced metal matrix bulk composites (Al/CNT, Cu/CNT and Ni/CNT) in terms of solid-state processing, referring mainly to the research works of the last ten years in this research field. The main methods are depicted in a brief way, and the pros and cons of each method are discussed. Furthermore, a tabular summary of the research work of the mentioned three systems is given, including the blending methods, sintering meth‐ ods, the used amount of CNTs and the finally achieved relative density of the composite. Finally, a brief discussion of each system is attached, which deals with the distribu‐ tion and interaction of the CNTs with the matrix material.

**Keywords:** Carbon nanotubes, metal matrix composites, reinforcement effect, solidstate processing.

# **1. Introduction**

Composite materials have been in the spotlight of material science and engineering for a long time already. They provide the capability of tailoring their properties by managing very simple variables such as the reinforcement fraction or the processing parameters, among others. Their application area is found in a wide range of dissimilar fields, ranging from bioengineering [1,2] up to aerospace [3]. The applicability relies on the proper selection of both, the matrix phase and the reinforcing phase. By matrix, the phase with the largest volume fraction is meant, whereas the opposite is valid for the reinforcing phase. In the former – according to the specific applica‐ tion – polymers, ceramics or metals are used. Polymer-matrix composites are mainly used in

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

applications where lightweight is required, working in environments that do not present high temperatures. On the opposite, ceramic matrices are used where inertness under high temper‐ atures is required as well as high mechanical properties. Finally, metal matrix composites (MMC) lie in between both application fields, presenting in most cases tailored microstructures (and subsequently, tailored physical properties) and, in certain cases, lightweight.

Reinforcing phases can be of very dissimilar nature. The most widespread phases are usually ceramic fibres and/or particles (i.e. Al2O<sup>3</sup> , TiC), which show very good mechanical properties, thus enhancing the overall mechanical properties of the composite. However, when consid‐ ering the transfer properties (electrical and thermal), their ceramic nature plays a detrimental role. Furthermore, the fact that the material being subjected to improvement is a metal, with usually very good transport properties, the task becomes indeed non-trivial.

In recent years, the appearance of carbon nanotubes (CNTs) has opened an interesting new field. Since CNTs show intrinsically outstanding physical properties, the aforementioned drawback brought by ceramic reinforcements might be straightforwardly overcome. Yet, the predicted physical properties of CNTs are only realisable if the CNT is in a "perfect" structural state. By "perfect" structural state, it is meant that there are (a) no structural defects on its lattice, (b) no exo- or endohedral contaminants (synthesis/catalysis residues) and (c) the CNTs are in an isolated state (no CNT agglomeration). Those three conditions are quite challenging to achieve in the praxis.

When it comes to CNTs, different synthesis methods have to be considered. CNT synthesis renders unavoidably contaminants (sulphur/amorphous carbon) and catalyst particles. The most common synthesis methods employed are chemical vapour deposition (CVD), arc discharge and laser ablation. The CVD synthesis is the one with the highest capacity to be scaled to industrial quantities and provides a better control on the morphology of the obtained CNTs [4,5] in comparison with other standard synthesis methods. Yet, the defect state is usually high and should be thoroughly analysed before the application.

Most of the recent research works are leaning towards commercially available MWCNTs as starting material for the composite production. Some of the most recurring suppliers are Nanocyl (Belgium), Nanolab Inc. (USA), Iljin Nanotechnology Co. Ltd (Korea), Bayer Materi‐ alScience (Germany), Chengdu Organic Chemicals Co. Ltd (China), Chinananotech Co. Ltd (China) and Hanhwa Nanotech Co. Ltd (Korea) to name just a few [6–31]. But also selfproduced CNTs are used for composite manufacturing. For this purpose, catalytic chemical vapour deposition (CCVD) is the most common way to synthesize CNTs that are used as reinforcement phase in composites [6,7,10,32–39]. The MWCNTs generally show a purity between 90% [10] and 99.5% [19], but mostly around 95%. The diameter ranges from 10 nm to 80 nm with a length of 0.5 µm–50 µm [6–39]. All the information about CNT-MMCs in this book chapter is derived from the research works of the last ten years, including some relevant exceptions.

#### **1.1. Blending methods**

Considering that commercial CNTs are usually delivered in agglomerated form, different methods have to be employed to disaggregate and blend them with the metallic matrix material. There are a lot of dispersing and blending processes like magnetic stirring [40], nanoscale dispersion processing [18,21,41], colloidal mixing [6,16,22,26–31,41–49], molecularlevel mixing [7,9,11,12,14,24,25,50–54], particle composite system mixing [15,55], friction stir processing [56], layer stacking [57], ball milling [8,10,17,19,20,22–24,32–36,38,50,53,58–74], dipping [75] or roller mixing [37,76]. However, three of them have to be pointed out as the most commonly used ways to disperse and blend CNT metal matrix systems. These three processing methods are ball milling, molecular-level mixing and colloidal mixing. The methods will be briefly commented in the following paragraphs.

#### *Ball milling*

It is usually performed using a planetary or attrition ball mill. The mixing is done by filling in the Metal powders and reinforcement phase together with some hard balls into mixing jars and rotating the jar with a certain rotational speed (**Figure 1 a**).

**Figure 1.** Schematic draft of the ball milling process (a). The used balls are hitting the CNTs and the matrix powder material (b), welding and integrating the two components (c) and bounce of the powder particle to restart this process again at another spot (d).

During the rotational movement, the added balls are falling on top of the powder material (**Figure 1 b**), thus leading to a size reduction, particle welding and integration of CNTs into the matrix powder material (**Figure 1 c** and **d**) generated by the impact energy. Different ball materials and sizes, rotational speeds, balls to powder ratios, gas atmospheres and mixing

times can be chosen as main mixing parameters. Usually, a process control agent, for example ethanol, is added in order to prevent cold welding of the matrix material powder particles. In other cases, like mechanical alloying, a cold welding of the particles is desired. Ball milling is known to produce a homogenous distribution of reinforcement phase in metal matrix composites as particle agglomerates are segmented, and dispersed CNTs are partially welded together with the matrix powder material (**Figure 1 d**). In general, high energy ball milling and low energy ball milling are distinguished in literature when it comes to the mixing process of CNTs with metal matrix material. This is because of the main drawback of the method, which is the increasing defect density of CNTs during the mixing process by the direct application of large contact pressures (up to 30 GPa) [77]. Using low energy ball milling, this unwanted effect can be reduced, but it will not vanish. The importance of a low defect density of CNTs will be discussed later on in entry 2.2 [8,10,17,19,20,22–24,32–36,38,50,53,58–74].

#### *Molecular-level mixing*

For this mixing method, it is of utmost importance to functionalize the CNTs, for example with an acid treatment (**Figure 2 a**). After this, the CNTs can be dispersed in various solvents, for example using ultrasonic agitation to obtain a stable suspension.

**Figure 2.** Schematic draft of the molecular-level mixing method. CNTs are functionalized by functional groups, which will be covalently bond onto the CNTs surface (a). After this, metal ions can electrostatically interact with the function‐ al groups, thus coating the CNT surface with metal ions (b). These ions are transformed to a pure metal layer by calci‐ nation and reduction under temperature and H<sup>2</sup> , N<sup>2</sup> or CO atmosphere (c). Finally, the CNTs are fully decorated or even integrated in the metal matrix material (d).

A metal salt is added and reduced by an added reducing agent to form a metal oxide CNT suspension with the CNTs acting as nucleation centres for the metal oxide formation (**Figure 2 b**). Finally, after washing off all the remaining chemicals, the powder is calcinated and then reduced, for example under hydrogen atmosphere to reduce it to the metal/CNT powder (**Figure 2 c** and **d**). The advantage of this process is that the CNT particles are embedded or coated by the metal matrix, thus resulting in very homogeneous distributions within the metal matrix (**Figure 2 d**). In many other mixing methods, the reinforcement phase can only be situated at the grain boundaries, offering weak interfacial bonding between the reinforcement phase and the matrix material and reducing the homogeneity of the distribution. For the molecular-level mixing process, this is not true, therefore being favoured for applications where a good connectivity of the reinforcement phase network is needed like thermal or

electrical applications. The main drawback of the method is the required functionalization, which involves breaking up covalent carbon bonds to add functional groups to the surface of CNTs, thus diminishing their outstanding properties [7,9,11,12,14,24,25,50–54].

#### *Colloidal mixing*

For colloidal mixing, the CNTs are dispersed using an ultrasonic bath, homogenizer or magnetic stirrer in a solvent (for the most part, ultrasonic agitation is used (**Figure 3 a**)). There are many different solvents that allow for a fine and stable dispersion of the CNTs, which is discussed in [81] (e.g. DMF or ethylene glycol). The dispersion grade and stability of the CNT suspensions are not only depending on the used solvent, but also on the surface of the used CNTs. Some research works functionalize the surface of the CNTs to obtain electrochemically stable dispersions, which works great but influences the physical properties of CNTs as already discussed for the molecular-level mixing. Other works avoid functionalization, thus using the pristine CNTs, retaining their properties in detriment of the dispersion quality.

**Figure 3.** Schematic draft of the colloidal mixing method. First, the CNTs are dispersed in a liquid solvent using ultra‐ sonic agitation, after which the matrix material powder is added (a). The two components are mixed again by ultrason‐ ic agitation (c), and the solvent is evaporated (c) to finally obtain the mixed CNT/metal matrix powder (d).

One point that is controversially discussed is the impact of ultrasonic agitation on the defect density of CNTs. There are studies claiming for a rise in the defect density as a function of the time spent in the ultrasonic bath and others which report the opposite. The observed decrease in the defect density might be a misinterpretation of Raman spectra of disentangled CNTs (which would render improved ID/IG ratio stemming from the avoidance of intertube interac‐ tions). After the dispersion of CNTs in the solvent, the metal powder is added and mixed with the dispersed CNTs again by ultrasonic agitation, stirring or a homogenizer (**Figure 3 b**). Finally, the solvent is evaporated to obtain a dry mixed powder (**Figure 3 c** and **d**). A significant advantage of this method is that it can be very easily upscaled and still yield the same results [6,16,22,26–31,41–49].

#### **1.2. Processing methods**

After merging the CNT/metal composite powders, different densification methods are used for the consolidation of the final samples. Examples of these are cold pressed sintering (CPS) [8,14,17,20,27,28,34,35,47,53,63,74,78], hot uniaxial pressing (HUP) [10,19,26,27,29–31,39,40,47, 49,62,67–70], spark plasma sintering (SPS) [6,7,9,11,12,14–16,18,21,23–25,42,44–46,48,50–52,55, 64,75,79,80], hot or cold rolling [7,50,59,60,64], hot extrusion [18,20,23,40,58,79], high pressure torsion (HPT) [22,31–33,41,54,61,65,66,71,73], friction stir processing [56], hot isostatic pressing (HIP) [22,57], microwave sintering [43], laser engineered net shaping (LENS) [36–38,76] or a combination of those methods. Methods like cold or hot rolling, hot extrusion or HPT are often used for post-processing of the already consolidated samples [7,22,31,50,59,60,64]. The most often used methods for the production of metal matrix composites are CPS, HUP and SPS, which will be described briefly in the following.

#### *Cold Pressed Sintering (CPS)*

This is by far the simplest method to densify the blended powder. Using a uniaxial press or an isostatic press, the powder is pre-compacted to the desired shape (**Figure 4**).

**Figure 4.** Schematic draft of pressureless sintering or cold pressed sintering. A die is filled with the powder material, which is then pressed uniaxial to a green pellet. The sample is then removed and sintered without pressure in a fur‐ nace under vacuum or inert gas atmosphere.

After this, the sample is sintered without further pressure under vacuum or an inert gas atmosphere to form the consolidated sample (**Figure 4**). Even though one heating–cooling cycle might be very time-consuming (the resistive furnaces usually have a very limited heating rate), it has the advantage that many samples of dissimilar shapes can be sintered at the same time, making this sintering method very time efficient. However, the densification mechanism is mainly based on lattice and grain boundary diffusion [81], and large porosities cannot be closed without applying pressure, thus resulting in a poor final density of the composite [8,14,17,20,27,28,34,35,47,53,63,74,78].

#### *Hot Uniaxial Pressing (HUP)*

In addition to lattice and grain boundary diffusion (mechanisms for pressureless sintering), plastic deformation and creep can be major sintering mechanisms. As the overall densification rate of a compact is a function of the sum of the active densification mechanisms, pressureassisted sintering is much more effective than pressureless sintering. The application of an external pressure leads to an increase in the densification driving force and kinetics. As grain growth is not related to the applied external pressure, it is more effective in systems with a

large grain growth to densification rate. To sum up, by using an external pressure, the sintering temperature as well as the sintering time can be reduced [81].

**Figure 5.** Schematic draft of the hot uniaxial pressing method. A pre-compacted green pellet is inserted in a die, and uniaxial pressure is applied while the pellet is heated by induction under vacuum or inert gas atmosphere. The densi‐ fied sample is finally removed.

When it comes to HUP, as with the CPS method, the mixed powder is usually pre-compacted using a uniaxial press or an isostatic press to obtain green pellets. After this, the green pellets are typically inserted in a die (often a steel die) where two punches (e.g. alumina) exert a uniaxial pressure onto the sample (**Figure 5**). The heating of the sample is usually conducted by induction, and thus, this system is very limited in the heating rate, rendering HUP as a very time-consuming sintering process. However, high pressures (several hundred MPa) can be applied with this method while sintering and almost full densification can be achieved (as punches with good mechanical properties can be used). To conclude, HUP is a time-consuming way to sinter samples, but it is also very effective when it comes to the final maximum densification of the sample [10,19,26,27,29–31,39,40,47,49,62,67–70].

#### *Spark Plasma Sintering (SPS)*

The active sintering mechanisms in SPS do not vary much from the active mechanisms in HUP. Yet, it provides a much quicker way to consolidate the composite powders [81].

**Figure 6.** Schematic draft of the Spark plasma sintering method. A graphitic die is filled with the powder material, and a uniaxial pressure is applied via two graphite punches. A pulsed electric DC is applicated, which leads to the heating of the sample by its electrical resistance. The process is conducted under vacuum or inert gas atmosphere.

As with HUP, the mixed powders are pre-compacted using a uniaxial press or an isostatic press to obtain green pellets, which are inserted in a graphite die. With this method, graphite punches are used to allow for inducing a pressure at the sample and conducting a pulsed electric DC through the sample at the same time (**Figure 6**). The sample is heated by its electrical resistance, which depends on the used material. By controlling the used current, the heating rate can be adjusted. This method allows for a very high heating rate of several hundred °C/ min, being therefore very time efficient. In contrast to HUP, the used pressure during sintering is much lower (typically about 50 MPa) because of the mechanically weak graphite punches. However, as with HUP, almost full densification can be reached with this method. Overall, this method offers a quick and effective way to consolidate the CNT-reinforced metal matrix powders, therefore being the most employed method in this area [6,7,9,11,12,14–16,18,21,23– 25,42,44–46,48,50–52,55,64,75,79,80].

#### **1.3. Potential applications**

The main application of the CNT-reinforced metal matrix composites is in structural applica‐ tions. This is due to the fact that most of the literature is devoted to the study of the mechanical properties of the composites. From the mechanical point of view, the addition of an intrinsically strong second phase would certainly improve the overall properties. Considering that, in the ideal state, the CNTs show a Young's modulus of approximately 1 TPa and a maximum tensile strength of over 60 GPa [82], it is easy to trace their influence. Additionally, a set of factors influences the mechanical behaviour of these composites. First, it has been demonstrated that the addition of CNTs acts on the grain boundary mobility by hindering their displacement during grain growth [83–85]. This effect influences the final microstructure (by refining it) and thus the mechanical behaviour (grain boundary strengthening). Second, a proper distribution of CNTs acts as an obstacle for dislocation movement, activating another strengthening mechanism known as particle dispersion strengthening (Orowan strengthening). Third, it has been shown that the CNTs present a very low or even negative coefficient of thermal expansion (CTE) in a wide temperature range [86,87]. When combined with high CTE materials (such as metals), this CTE mismatch acts also as a strengthening factor. Finally, considering the aforementioned values of the mechanical properties of the CNTs, the strengthening of the composite due to load transfer is expected to be significant. Summarizing, the addition of CNTs to MMCs is clearly expected to be beneficial in terms of the improvement of the mechanical behaviour, and subsequently, it is clear why most of the studies are focused towards this feature. The strengthening effect of the CNTs in MMCs is shown in **Figure 4**. There, the measured yield strength improvement of different systems is given as a function of the CNT volume concentration in the composite. The scattering of results is related to the different mixing and processing methods that were employed. In this context, it becomes clear that after a direct comparison of different research works involving a wide span of production methods for the same system, the task of correlating dissimilar results becomes non-trivial. Due to a lack of available information on the yield strength of Ni-CNT composites, this illustration provides only a summary of the mechanical reinforcement effect that was achieved so far with Al-CNT and Cu-CNT MMC systems.

Carbon Nanotube (CNT)-Reinforced Metal Matrix Bulk Composites: Manufacturing and Evaluation http://dx.doi.org/10.5772/63886 137

**Figure 7.** Yield strength improvement against the volume fraction of CNTs in (a) Al/CNT composites [18,20,56,57,60] and (b) Cu/CNT composites [7,9,11,53,64,66,70,75].

However, the transport properties occupy also a large amount of the research in this field. CNTs are predicted to have the highest thermal conductivity known (SWCNT: 6600 W/m.K [88], MWCNT: 3000 W/m.K [89]) and are expected to present a ballistic type of electrical conduction mechanism [90,91]. Both factors are of utmost importance when considering that the material to be improved is a metal (which usually shows very high thermal and electrical properties). However, it is critical to obtain individual CNTs after dispersion, since the transport properties could be reduced up to one order of magnitude If the CNTs are in agglomerated form [92]. In this regard, some parts of the research on Cu-CNT were aimed at the improvement of the thermal properties. Cho et al. observed an improvement of the thermal conductivity for very low CNT concentrations (up to 1 vol. %) [42]. For larger CNT concen‐ trations, the agglomeration of CNTs starts to play a fundamental role in the conductivity decrease. The same effect was observed by Yamanaka et al. for Ni-CNT [46]. Inversely, Firkowska et al. tested several functionalization routes so as to improve the interface in Cu-CNT composites and thus the thermal transport [14]. They observed that for all cases, the degradation of the CNT's intrinsic thermal properties was so relevant and that it was not able to increase the thermal conductivity of the composites in any case.

The electrical behaviour was also studied in several MMC-CNT systems, usually resulting in reduced conductivity. The major factor influencing this behaviour is the presence of agglom‐ erates that, analysed from an electrical transport point of view, are seen as voids. This happens mainly due to the fact that the interfaces between CNT agglomerates and the metal matrices are weak. Only marginal improvements in conductivity were observed for low CNT concen‐ trations in Ni/CNT composites [46].

The tribological properties of CNT-reinforced composites have also been reported in the literature. In most cases, a reduction in both the coefficient of friction (COF) and wear has been observed. In the case of Ni-CNT composites, the COF was reduced in margins from 40 [37] to 75 % [27]. Regarding the Cu-CNT system, the COF reductions ranged from 50 to 75 % [6,8,43, 51,63]. For both systems, the reduction in wear losses achieved in certain cases up to 6 [51], 7 [27] and 8 [43] times that of the reference metal under the same experimental conditions.

The former is usually traced back to a solid lubricant activity of the CNTs during the experi‐ ments [37]. Due to their high mechanical properties, they tend to act as rolling second phases at the interface between the rubbing surfaces, thus avoiding the direct contact between the surfaces. In some cases, also the development of a graphitic layer is observed, acting as a solid lubricant [37]. The wear reduction is mainly correlated to the increase in the mechanical properties of the composite, which—as already stated—is generated by a stabilization of the microstructure due to grain boundary pinning. Moreover, Kim et al. stated that a reduction in the wear loss might be due to the reduction in the grain peeling mechanism due to a CNT anchoring of the matrix grains [51].

## **2. Aluminium/CNT system**

#### **2.1. Solid-state processing**

Al/CNT composites are mainly interesting, because of their high potential being a lightweight, reinforced material, which can be used in manifold applications. Therefore, Al/CNT compo‐ sites have been in the focus of research since 1998 [40] and the interest is still growing.

There is a large variety of starting materials on the market that have been used to fabricate the composites. Some of the most mentionable suppliers for Al powders are ECKA Granules Japan Co. Ltd (Japan), Aluminium Powder Company Ltd (United Kingdom), Alpha Industries (South Korea) and AlfaAesar (Germany) [18,19,21,58,61,62]. The range of used powders goes from several µm up to 75 µm in mean particle size, having different particle shapes and various purity grades between >99% and >99.99% [18,22,39–41,50,56–59,79]. The most used blending method for Al/CNT composites is ball milling [19,20,22,23,50,58–62].

A detailed overview of the research papers in the Al/CNT composite manufacturing can be found in **Table 1**.







**Table 1.** Summary of blending and sintering methods for the production of Al/CNT composites.

#### **2.2. Distribution and interaction with the matrix material**

In all the analysed articles, CNT agglomeration is observed, being the size and amount of those agglomerates strongly related to the chosen dispersion method. Specifically, in those reports where ball milling is used, a significant clustering is noticed. This might be due to the use of non-functionalized CNTs that tend to easily re-agglomerate during processing. Yet, the utilization of functionalized CNTs in ball milling mixed blends would not bring any further improvement, since the structural quality of the CNTs would be significantly lower. However, despite the presence of unavoidable agglomerates, very fine cluster dispersion and a homo‐ geneous distribution can be observed. Interestingly, one would assume that the formation of clusters would be detrimental to the proper densification of the composites. However, as already depicted in Table 1, the final densities of the composites were never below 94%. The formation of clusters is one of the most challenging problems to overcome when dealing with CNTs, since a trade-off between an optimal initial agglomerate disentanglement and the lowest possible damage to the CNT structural state must be found. There are two ways to observe the amount of agglomerates in the composites. Certain reports focus on the evaluation by electron microscopy on the polished surface of the samples, whereas other authors evaluate fracture surfaces of the composites. Both approaches are valid, in the sense that they show a good overview of the agglomeration. However, SEM imaging of the surface would allow a further quantification of the agglomerates size and distribution by segmenting the C-containing phases. Bakshi proposed an interesting methodology to quantify the dispersion, based on distance calculation algorithms from electron micrographs, which can be found in **Bakshi et al.** [93].

If we focus our attention on the CNT degradation during blending and after densification, we observe dissimilar results obtained by Raman spectroscopy, even for the same manufacturing method. For example, in samples densified by SPS (a technique known to exert strong thermomechanical stress on the CNTs), negligible variation of the ID/IG ratio is reported [23,79] as opposed to an increase in the defect density in other reports [18]. It is therefore very difficult to discern whether the sintering process has indeed any sort of influence on the CNT degra‐ dation. It has been already reported that given the strong stresses from SPS, CNTs are usually either damaged or degraded within the matrix material [94]. When strong shear forces are applied as in the case of HPT, sharp increases in the D band intensity are measured, depicting a modification of the graphitic structure. In all cases, an increased amount of structural damage would favour the reactivity of the CNT shells.

The CNT degradation is a major issue when working with metals which are strong carbide builders [95]. It is well known that aluminium tends to form a stable carbide (Al4C<sup>3</sup> ) in a wide compositional range [96], being of brittle nature and having the major drawback of being water soluble [97]. Some authors mention that the formation of this carbide would not be as detri‐ mental as expected [18,58], since it only sacrifices the outermost CNT layer, providing an optimal interface with the matrix. However, despite being able to efficiently transfer the applied load, the utilization of the composite in structural applications in humid environments would lead to an interfacial degradation. Another way to obtain this carbide is by transitioning from Al2O<sup>3</sup> to Al4C<sup>3</sup> [98]. Yet, this phase transition occurs only under certain circumstances that are rarely achieved in solid-state processing.

# **3. Copper/CNT system**

### **3.1. Solid-state processing**

Reports on the solid-state processing of Cu/CNT systems deal with a variety of different raw materials, blending methods and sintering techniques, therefore sticking out in the research field of CNT-reinforced metal matrix composites.

When it comes to the used Cu starting material, a great many of commercially available powders or chemicals are employed, making it very hard to compare the Cu/CNT systems of different publications. The materials used depend directly on the blending and production process that has been used. The range goes from nanosized to almost 100 µm sized Cu powder particles, having different particle shapes (spherical, dendritic) and various purity grades between >99.5% and >99.95% [6–12,14,15,32–35,42–44,50,52–55,63–71]. The most used blending methods are ball milling, molecular-level mixing or colloidal mixing, each having their advantages and disadvantages as discussed in the introduction [6–12,14–17,32–35,42–45,51– 55,63–75]. But especially for Cu, molecular-level mixing is employed very often, using copper compounds in solvents that have to be chemically treated in order to become pure copper (e.g. copper(II) acetate monohydrate, Cu(II) sulphate pentahydrate or simply CuO). Some promi‐ nent suppliers for the Cu starting materials are Sigma-Aldrich (USA), Alfa Aesar GmbH & Co. KG (Germany), Chang Sung Co. (Korea), Junsei Chemical Co. Ltd (Japan), Kojundo Chemical Lab. Co. Ltd (Japan), TLS Technik GmbH (Germany) or New Materials Research Co. Ltd (China) [6,12,14–16,32,33,42,54].

is added and sonicated for 2h. The solution is vaporized with magnetic

Therefore, to review the progress in Cu/CNT composites, a detailed comparison of blending methods, sintering techniques and achieved final relative composite densities has to be conducted, which can be found in **Table 2**.








Carbon Nanotube (CNT)-Reinforced Metal Matrix Bulk Composites: Manufacturing and Evaluation http://dx.doi.org/10.5772/63886 153




**Table 2.** Summary of blending and sintering methods for the production of Cu/CNT composites.

#### **3.2. Distribution and interaction with the matrix material**

The only stable copper carbide known is the copper acetylide Cu2C<sup>2</sup> [99]. It is usually observed as a transition phase during the purification of Cu and after the reaction of cuprous oxide with water [99]. In the studied Cu-CNT composites, no carbide was detected whatsoever. However, in certain cases, CuO and Cu2O were observed as a result of the selected manufacturing process. This issue is overcome by utilizing reagents (such as EDTA) or a reducing atmosphere as a post-processing method.

Regarding interfacial features, there is a publication that reported on the influence of adsorbed oxygen on the CNT surface on the interfacial strength, observing an improved interface but a reduction in the transport properties [14]. Theoretical simulations have demonstrated that the addition of oxygen-containing functional groups results in improved interfaces. The assertion is based on the fact that chemically active oxygen on CNT surfaces might improve the binding of metals with CNTs by enhancing the electron exchange between the metal and the carbon atoms or by directly interacting with the metal [100].

In some cases, the presence of alloying elements in the matrix degraded the CNTs into carbides as for example: the presence of Al leads to the formation of Al4C<sup>3</sup> [50], and the presence of Cr leads to the formation of Cr3C<sup>2</sup> [70]. Another approach was considered by mixing the CNTs with a Cu-Ti alloy. Chu et al. showed that the CNTs reacted in the presence of Ti, resulting in a degradation of the CNTs into a TiC interphase, despite having only 0.85 wt% Ti in the mixture [68]. This reaction between the CNTs and the carbide forming alloying elements is usually referred to as an interfacial improvement, leaving aside the fact that the CNTs are likely degraded into a hardly improving second phase in the composites.

On the other side, there are reports in which a seamless interface was achieved even avoiding any phase formation, as shown by Cho et al. [42]. This coherent interface would then result in the reduction of detrimental features such as thermal resistance.

In certain cases, XRD is used to observe the possible interphase formation; however, it would not be sufficient to resolve it since the volume fraction of the formed phases would be low. HR-TEM is the most suitable way to characterize these interphases with the aid of SAED.

A very interesting way to overcome the reaction between the CNTs and the alloying elements was addressed by decorating the CNTs with Cu nanoparticles [14,43]. In this case, a good distribution was achieved, rendering improvements in hardness and thermal resistance, despite reducing the CNTs intrinsic properties.

The analysis of the agglomeration and distribution of CNTs was, in all cases, qualitatively reported. As a general case, the CNT distribution was acceptable, with low reagglomeration activity. Both, the reaction with alloying elements as well as the covalent functionalization of the CNTs (as in MLM) tends to reduce significantly the intrinsic properties of the nanotubes, thus hindering an optimal exploitation of the CNT usage.

Regarding the adhesion of the reinforcement to the matrix, the best outcomes are observed when covalent functionalization is used (MLM) or an interphase is formed. In the case of nonfunctionalized CNTs, the adhesion to the matrix is poor, mainly due to a poor metal-CNT wettability [95].

## **4. Nickel/CNT system**

#### **4.1. Solid-state processing**

**(2008) [**76]

consisting of two rolls rotating in opposite directions (one clockwise and one anticlockwise). This mixing is carried

For Ni/CNT, as for the other composite materials, there is a large variety of starting materials that have been used to fabricate the composites. Suppliers that are found very frequent for Ni powders are Alfa Aesar (Germany) and Crucible research (USA) [26–31,36–38,47,76,78,80]. The range of used powders reaches from 120 nm up to 149 µm in particle size, having a dendritic or spherical morphology and showing a purity of 99.8% to 99.99% [24–31,36–38,46– 49,76,78,80]. The most used blending method for Ni/CNT composites is colloidal mixing and ball milling [24,26–31,36,38,46–49].

A detailed overview of the research papers in the Ni/CNT composite manufacturing can be found in **Table 3**.



The powder is finally


argon atmosphere under a

(850°C)





**Table 3.** Summary of blending and sintering methods for the production of Ni/CNT composites.

#### **4.2. Distribution and interaction with the matrix material**

As observed in the other metal-CNT systems, agglomeration is observed in all cases. If the different blending methods are considered, it is very unlikely to obtain a predominantly individual CNT dispersion rather than a homogeneous cluster distribution. Nevertheless, this cluster distribution provides also diverse improvements with regard to microstructural refinement and properties enhancement. Furthermore, when the initial agglomerate size (asreceived state) is considered, all dispersion methods render very good disaggregation. Regarding the damage to the CNTs during processing, it is clear that highly energetic proc‐ essing as ball milling increases the amount of defects on the CNTs, whereas milder processing routes as colloidal mixing present a fairly unmodified state of the CNTs structure. Interest‐ ingly, the sintering process tends to help in the defect healing [29]. The application of temper‐ ature in a non-reactive environment (vacuum sintering of non-carbide forming metal matrices) diminishes the ID/IG ratio as observed with Raman spectroscopy. Furthermore, the purity index IG'/ID also is reduced, evidencing some sort of contaminant removal (e.g. amorphous carbon).

Nickel does not form stable carbides [101]**;** however, it has been reported that under certain conditions it is possible to obtain Ni3C [102]. This carbide is brittle and, despite improving the interfacial cohesion to the CNTs, would have a detrimental influence on the transport prop‐ erties of the composite. Due to its metastable nature, it is quite complex to detect it in the composite. It has been shown that the most suitable way is to assess the crystallographic lattice of the Ni in the vicinity of the CNT-containing areas by selected area electron diffraction [29,102]. In the case of a C-Ni reaction, the originally FCC Ni phase would transition to an intermediate Ni3C (hcp) and later would stabilize in an hcp Ni phase. Thus, if this hcp lattice is detected in the near region of the CNT zones, it would mean that the Ni3C was previously present. To the date, Ni3C in Ni-CNT systems has been only detected once by Hwang et al. [102] under a very specific set of synthesis conditions.

Yamanaka et al. and Nguyen et al. show that the application of SPS rendered a very smooth interface between the CNTs and the matrix [46,48]. This was also observed in HR-TEM for hot pressed samples [29]. In general, all the different approaches have resulted in seamless interfaces that are later translated in improved mechanical properties [26,48], thermal prop‐ erties [46], tribological [27,37] and thermal expansion behaviour [47,78]. Additionally, a proper interface would favour a grain boundary drag that improves the microstructural control by achieving refined microstructures [26,28,48].

## **5. Outlook**

Although a large amount of efforts was directed towards the development of CNT-MMCs systems, there is still a significant room for improvement. This statement is supported by the fact that as described, dissimilar results have been obtained by even using the same processing methods as well as the same type and amount of CNTs. This is generated by a scarcity of proper knowledge of each particular system. As an example, there is still an ongoing discussion in the community about the most suitable dispersion and blending methods for a certain application. Furthermore, it can be noticed that the impact of interphases on the physical properties of the composites is still not well understood. Thus, we foresee a very high potential to gain new insights on each particular system and subsequently achieve further developments in this field.

### **Acknowledgements**

The present work is supported by funding from the Deutsche Forschungsgemeinschaft (DFG, projects: MU 959/38-1 and SU 911/1-1). The authors wish to acknowledge the EFRE Funds of the European Commission for support of activities. This work was also supported by the CREATe-Network Project, Horizon 2020 of the European Commission (RISE Project No. 644013).

## **Author details**

Sebastian Suárez\* , Leander Reinert and Frank Mücklich

\*Address all correspondence to: s.suarez@mx.uni-saarland.de

Functional Materials, Dept. of Materials Science and Engineering, Saarland University, Germany.

#### **References**

