Deformation Behaviour of Cold-Rolled Ni/CNT Nanocomposites

Abstract

Metal matrix nanocomposites (MMNCs) reinforced by carbon nanotubes (CNTs) are good candidates to produce structural components in the mobility industry, given their unique properties. The manufacture of these components can involve plastic deformation. Therefore, it is crucial to understand whether reinforcement can influence the deformation behaviour of these nanocomposites. Thus, this work aims to study the deformation behaviour of MMNCs, given their importance and the lack of studies on this topic. Although nickel is not the most widely used metal as a matrix of nanocomposites, it presents mechanical properties superior to other matrices, such as aluminium. In addition, this metal has proven to establish a strong interface and integration of carbon nanotubes, making it an exciting material for the production and study of these nanocomposites. In that sense, nickel matrix nanocomposites are reinforced by 1.00 %vol. CNTs were produced by powder metallurgy using ultrasonication as a dispersion/mixture method. For comparison purposes, a nickel matrix was produced under the same conditions. Samples with and without CNTs were cold-rolled with thickness reductions between 10 and 60% (logarithmic strains between 0.11 and 0.92) to investigate the deformation behaviour. Microstructural characterization was performed using scanning electron microscopy (SEM) and electron backscattered diffraction (EBSD). Microhardness tests were applied to evaluate their mechanical properties. The results revealed that the nanocomposites exhibited a softening for small strains (0.11 and 0.22). This decrease in hardness was attributed to the decline in dislocation density observed by EBSD, due to the rearrangement and annihilation of pre-existing dislocations that originated during production. A possible inversion can explain the decrease in dislocation density when minor strains are applied in the dislocation or deformation trajectory, known as the Bauschinger effect. The difference in the texture evolution of the nanocomposites can be explained by the initial crystallographic orientations, which are influenced by the presence of CNTs.

1. Introduction

The development of nanocomposites comes from the growing need to obtain advanced materials capable of combining low weight with high mechanical properties. In this sense, metal matrix nanocomposites are especially attractive for producing structural components in the mobility industry [1]. The range of reinforcement materials is vast, with carbon nanotubes (CNTs) standing out due to their unique properties when integrated into metal matrices. Furthermore, the heterointerfaces of nanocomposites can improve the properties of some components of the energy industry due to the possibility of increasing the ionic and/or electronic conductivity, electrocatalytic activity, and stability of electrolytes and electrodes. These heterointerfaces are characterized normally by large variations in local composition, lattice strain, electronic structure, and other properties [2].

The promising mechanical properties of these nanocomposites depend on the contribution of several strengthening mechanisms, which act simultaneously. If they do not occur properly, they can compromise the strengthening effect, as analysed in previous work [3]. Since the final components’ obtention is often dependent on some processing to obtain their final geometry, the knowledge of the effect of CNTs on the deformation behaviour of nanocomposites is essential for their implementation in the industry. In that sense, understanding how CNTs can influence the deformation behaviour of the nanocomposites and their interaction with the matrix during the process represents an added value in optimizing their properties and processing. This deformation alters their microstructure and, consequently, their mechanical properties, requiring detailed studies. For the heterointerface, the dislocation network along the interface also plays a role in the transport pathways in the ionic conductivity [2].

During the plastic deformation process, the grain morphology change is accompanied by grain rotation, so the structure accommodates the imposed deformation. The movement and multiplication of dislocations allow the plastic deformation process to continue until fracture. These aspects can increase domains, dislocation cells, and the appearance of one or more preferential crystallographic orientations. This texture is more evident with the greater degree of deformation of the material.

There are two main plastic deformation mechanisms in metallic materials: dislocation slip or twinning. However, dislocation movement is the dominant plastic deformation process, most often the only one, and mainly responsible for grain rotation, usually occurring in the crystal’s more compact planes and directions. When a polycrystalline material is plastically deformed, each grain tends to deform differently from its neighbours, as they have different orientations. This would lead to the formation of voids or overlaps between grains, which does not occur. When deforming polycrystalline metals, continuity is observed between the grains. However, for this continuity to occur, the grains deform in the direction of the applied strain. In turn, this deformation is guaranteed by geometric necessary dislocations (GNDs), whose movement and multiplication are due to local internal stresses and can accommodate the applied strains [4,5]. All this contributes to a change in grain morphology during plastic deformation.

The crystallographic orientations of individual grains in a polycrystalline material are modified during plastic deformation due to grain rotation. They can result in preferred orientation or textures, which become stronger with increasing deformation [6]. Texture formation can be influenced by several factors, such as (1) the chemical composition, namely the presence of alloying elements, which consequently can influence the atomic structure and energy faults; (2) the type of applied deformation (its temperature and type of strain path imposed); and (3) the initial crystallographic orientation of the grains, i.e., before deformation [6]. Some works investigate how these conditions can influence the texture evolution of some alloys [7,8,9,10,11,12].

Included in a wide range of processing techniques, rolling is a very recurrent plastic deformation process in industrial environments. It is also commonly used to study the formation of crystallographic textures in metals. Thus, this deformation technique is used in the present work. The texture of a rolled sheet metal is often represented by its components, i.e., characteristic grain orientations. Each texture component is defined by a crystallographic plane {hkl}, parallel to the rolling plane; and a crystallographic direction <uvw>, parallel to the rolling direction [6].

Sun et al. [13] addressed the evolution of the texture and microstructure of cold-rolled nickel. The authors used backscatter electron diffraction (EBSD) to characterize recrystallized and deformed samples (with thickness reductions between 20 and 90%). This study concluded that during cold-rolling, the grains became more elongated, the average misorientation of the grains decreased as shown by the boundaries’ character distribution, and the main texture components observed varied with the applied deformation. For thickness reduction greater than 50%, the main components observed were of the shear type. The deformation structures and mechanical properties of 2519 aluminium when cold-rolled with a thickness reduction of up to 80% were also studied by Zuiko and Kaibyshev [14]. Among other techniques, the authors relied on EBSD analysis, which allowed the analysis of the influence of cold-rolling on the microstructure, dislocation density, and crystallographic texture of the samples. The dislocation density increased with the applied strain, which increased in the tensile strength of the samples. The authors also observed dislocations organized in cellular structures (up to 40% thickness reduction) and shear bands for greater deformations. The present texture components also changed and evolved during the deformation, with the texture after 80% thickness reduction mainly composed of the copper component. Other studies were carried out on the deformation of metals and their influence on the dislocation structures and textures formed, and they concluded that the dislocations organize themselves into cell walls, whose interior contains a small dislocation density (see, for example, Fernandes and Schmitt [15]).

In this context, studying how reinforcement can influence texture evolution with plastic deformation is crucial. Few studies investigate the deformation behaviour during the tensile test in nanocomposites, but these works do not consider texture evolution [16,17,18,19,20].

The main objective of this work is to contribute to a detailed understanding of the deformation behaviour of nanocomposites and how the presence of CNTs reinforcement can affect texture evolution. The nanocomposites and the matrix were produced by powder metallurgy and deformed by cold-rolling, and the texture evolution was studied by an EBSD analysis.

2. Materials and Methods

The nanocomposites were produced by powder metallurgy. The nickel powders used were from the supplier Goodfellow (Goodfellow Cambridge Ltd., Huntingdon, UK), whose morphology and microstructure were thoroughly studied by advanced characterization techniques in other works [21,22,23,24]. The carbon nanotubes used as reinforcing material were multi-walled and non-functionalized, supplied by Fibermax (Fibermax Ltd., London, UK) and have also been characterized in detail in previous work [25].

The metallic powders and carbon nanotubes (1.00 %vol.) were mixed. The dispersion and mixture of powders and CNTs were performed simultaneously in a single step, consisting of 15 min of ultrasonication with isopropanol; followed by cold pressing; and, finally, sintered in a vacuum. This process has already been described elsewhere [3,23,24,25,26,27], as well as its effect on the carbon nanotubes [28] and metallic powders [22]. Unreinforced samples (matrix only) were produced and deformed under the same conditions for comparison purposes.

To study the deformation behaviour, nanocomposites, and the matrix, produced under the same conditions, were submitted to cold-rolling, with different thickness reduction percentages (from 10 to 60%), as schematized in Figure 1a. The cold-rolling process was performed at room temperature (≈25 °C) with 10 rpm, and each thickness reduction was gradually achieved in a maximum of 5 passes. The microstructure of rolling samples was analysed in the cross-section, parallel to the rolling direction, as represented by the letter A in Figure 1b. The thickness reduction can be associated with the respective logarithmic strain (ε) by Equation (1) [29]:

$$\epsilon =\ln \left(\frac{h_0}{h}\right)$$

(1)

where the h₀ represents the initial thickness of the sample and h its thickness after rolling.

All the samples were microstructurally characterized using scanning electron microscopy (SEM) with high-resolution equipment FEI QUANTA 400 FEG SEM (Thermo Fisher Scientific, Waltham, MA, USA) and a TSL-EDAX EBSD detector (Ametek Inc., Devon-Berwyn, PA, USA). While acquiring the raw EBSD data, small step size was used to obtain highly detailed information. In this case, the minimum step size used was 0.03 µm. The analysis of the EBSD data was performed using two software with different functionalities: the TSL OIM Analysis 5.2 (Ametek Inc., Devon-Berwyn, PA, USA) and the ATEX version 3.28 (University of Lorraine, Metz, France) [30]. The first one was used for most maps, for example, inverse pole figure (IPF) and Kernel average misorientation (KAM) maps, and to distinguish between different boundary characters. The conditions applied in each map, as well as the clean-up routine applied to avoid misleading results of this software, are described in previous works [3,22]. The ATEX software allowed the elaboration of estimated maps of average geometrically necessary dislocations (GNDs), grain disorientation maps, as well as pole figures (PF), and orientation distribution functions (ODF).

The mechanical properties of the deformed samples (matrix and nanocomposite submitted to each applied strain) were evaluated by microhardness tests. The condition used was a 98 mN load during the normalized time (15 s). The calculated average values refer to about 15 indentations per sample, measured on cross-section A, illustrated in Figure 1b. The microhardness tests were performed using Duramin-1 equipment (Struers, Ballerup, Denmark).

3. Results and Discussion

3.1. Microstructural Characterization of Sintered Samples

The initial characterization of the samples was carried out to assess whether the reinforcement has any influence on the crystallographic orientation of the matrix during production, as this is a factor that certainly influences the evolution of the texture during plastic deformation. Figure 2 shows the IPF maps, the {111} pole figures, and the inverse pole figures for the Ni matrix and Ni/CNT composite after sintering. Based on these figures, the samples are not characterized by having a marked texture. The introduction of reinforcement changes the crystallographic orientation of the matrix. This can be seen in the pole figures and inverse pole figures of the two samples, shown in Figure 2b,d.

For instance, the IPF figures reveal that the Ni matrix has a preferential orientation of directions <111> and <212> parallel to ND, while nanocomposites show a preferential orientation of directions <101> parallel to ND. The IPF of transversal direction (TD) exhibits differences between matrix and nanocomposites. The matrix shows a higher intensity in the direction <001> parallel to TD, while the nanocomposite has its preferential orientation divided into two, around <112> and <012>, with lower intensity. Another visible difference is that in Ni, there is a preferential orientation to the direction <101> parallel to the RD. In contrast, in the Ni/CNT, despite <101> also having some intensity, the principal orientation is observed in <111> parallel to RD. The respective PF of the samples illustrates these differences between the two samples. The texture difference can be related to the fact that the presence of CNTs alters the grain rotation during sintering and, in addition, affects the movement and rearrangement of dislocations originating during the dispersion/mixture process, which directly influences the extension of recovery and recrystallization occurring during the sintering of samples, as reported in previous works [3,22]. Several authors have already used the grain boundary misorientation angle to study the deformation of metallic materials [13,31,32,33]. In that sense, a study of the grain boundaries’ character and misorientation was also carried out in the as-sintered samples to understand to what extent the reinforcement can influence these important parameters to be evaluated when the samples are subjected to plastic deformation. The maps with the different types of grain boundaries and the distribution of grain boundary misorientation, referring to the Ni and Ni/CNTs samples before deformation, are presented in Figure 3.

This type of analysis was used by Sun et al. [13] in the study of deformed nickel, and by Ciemiorek et al. [33], and is common when investigating deformation studies in detail since it allows us to understand the consequences and extension of deformation, for example, through the LAGB fraction, associated with dislocation cells, and the misorientation consequence of this deformation. Although, as mentioned, this analysis is more common on deformed samples, the present work performs it also on sintered samples. This allows investigating the effect of the cold-rolling process, since it characterizes the initial stage (as-sintered). After sintering, some deformation is still observed in the microstructure (especially in the nanocomposite). In the maps of Figure 3, the HAGBs are outlined in black, ∑3 in red, the remaining CSLs marked in green, and the LAGBs in purple. The fraction of CSLs in the Ni samples, especially the ∑3 twins, is higher than the values shown by the nanocomposites. This may be related to the fact that during sintering, a rearrangement of dislocations occurs that promotes the formation of HAGBs and CSLs, promoting a restored microstructure. In nanocomposites, the mobility of dislocations is hindered due to the presence of reinforcement, which affects the kinetics of the restoration processes and, consequently, the fraction of this type of boundaries.

Based on these results, the fraction of LAGBs is higher in the nanocomposites than in Ni without reinforcement. This fraction of LAGBs is related to the higher dislocation density observed for the nanocomposites, as mentioned above. In the grain boundary misorientation distribution of Figure 3b,d, the red columns correspond to the calculated misorientation relative to neighbouring points, and the blue line is relative to a random point in the sample analysis area. The green line, also known as Mackenzie’s distribution [34], is the theoretical distribution of misorientation between grains of a polycrystal assuming a random orientation [34,35]. The Ni/CNTs nanocomposite exhibits smaller average misorientation angles (θ_Avg.) than the Ni sample due to its high fraction of LAGBs.

3.2. Microstructural Characterization of Cold-Rolled Samples

The microstructural evolution of the deformed samples is evaluated by unique grain colour maps shown in Figure 4. This figure reveals that the grains elongate in the rolling direction with increasing strain. However, the grains seem more equiaxed for the nanocomposites than for the matrix until 0.36. The grain elongation in the deformation direction is expected and was already mentioned in other authors’ works [33].

The hardness evolution is measured to understand this behaviour and is presented in Figure 5. In the case of the Ni matrix, the hardness increases with increasing strain, as expected. However, for nanocomposites, the observed behaviour is different. For strains of 0.11 and 0.22, a softening is observed, and, for higher values, the hardness of the nanocomposites is lower than for the matrix. A more detailed study on the microstructure may explain this behaviour of nanocomposites when subjected to cold-rolling.

A more detailed microstructural characterization is carried out on samples of Ni matrix and Ni/CNT nanocomposites cold-rolled up to various strain values. Figure 6 shows maps with the different types of boundaries and grain boundary misorientations’ distribution for the Ni and Ni/CNTs samples after deformation. For a strain of 0.11, the Ni matrix and the nanocomposites show an increase in the fraction of CSL and HAGBs and a decrease in the fraction of LAGBs. However, this difference is noticeably more significant for nanocomposites. The Ni matrix cold-rolled with 0.36 is characterized by a high fraction of LAGBs (f_LAGB = 70%), a low average disorientation angle (θ_avg = 17°), and elongated grains with a few twins, as can be seen in Figure 5b,e. These effects on the microstructure can be explained by the increase in dislocations necessary to accommodate the imposed deformation, which leads to the formation of dislocation tangles and dislocations walls, which consequently are revealed in the rise in the LAGB fraction.

Additionally, since the boundaries with lower angle fractions increase, and consequent high angle boundary and CSL decrease, the average disorientation also decreases. For the nanocomposites subjected to the same strain, a few different microstructural changes are observed concerning the condition after sintering, with a similar fraction of LAGBs. For the nanocomposites subjected to the same strain, a few different microstructural changes are observed concerning the condition after sintering, with a similar fraction of LAGBs.

Regarding the strain of 0.69, the Ni and Ni/CNTs samples are identical. The average disorientation angle decreases with an increasing strain once LAGBs are formed, i.e., the formation of boundaries with lower disorientation causes the decrease in the average disorientation values in the sample, as can be seen in the blue line of the graphs in Figure 3. This effect was also mentioned in the works by Sun et al. [11] and by Ciemiorek et al. [33] where, after deformation, a decrease in HAGB was observed accompanied by an increase in the LAGB fraction, leading to a lower average misorientation angle. Although LAGBs are often associated with dislocation cell formation and sub-grain boundaries, a further analysis is performed using estimated GND maps. Figure 7 shows the average density values of the GNDs, and the respective maps observed for Ni and Ni/CNTs samples after sintering and cold-rolling up to different strains. In this figure, it is possible to observe a decrease in dislocation density for a strain value of 0.11. For strains of 0.36 and 0.69, the dislocation density increases similarly. As predicted by Ashby’s model, the slight growth of GNDs is observed for these strains [36]. This can be attributed to the fact that for high strains, the statistically stored dislocations (SSDs) become more significant, resulting in the stabilization of the density of GNDs.

The decrease in dislocation density observed for a strain of 0.11 can explain the relatively low hardness values observed for the samples when subjected to small strains by rolling (0.11 and 0.22). This can be related to the Bauschinger effect due to the annihilation and reorganization of dislocations by reversing their path during the initial rolling phase. The Bauschinger effect can be influenced by the presence of the phase in the matrix [37]. For nanocomposites, due to the presence of the CNTs and the few particles of Ni₃C, this effect can be pronounced and can result in a decrease in the strain hardening of the nanocomposites. Another aspect that needs to be pointed out is that, for the nanocomposites, the initial density of GNDs is higher than that of the Ni matrix, which will also affect dislocation motion during the cold-rolling and, consequently, the Bauschinger effect. The Bauschinger effect was also already observed by Xu et al. [38] in aluminium nanocomposites reinforced by CNTs through loading-unloading tests, where the authors observed an accentuated Bauschinger effect on the nanocomposites rather than in the aluminium matrix, showing that the CNTs can in fact influence this behaviour. Similar conclusions were also drawn by Chen et al. [39] regarding a 6061-aluminium alloy, and Sadeghi et al. [40] with Al, both reinforced by CNTs, and by some other authors studying similar nanocomposites [17].

This Bauschinger effect can also explain that the hardness values of the nanocomposites are lower than those of the nickel matrix. However, the texture evolution may also affect this variation in hardness values since the samples show different crystallographic orientations after sintering, which can affect the active slip planes during the rolling deformation process.

3.3. Texture Evolution

The texture evolution is investigated for the Ni matrix and Ni/CNTs nanocomposites at three strain values (0.11, 0.36, and 0.69). The orientation distribution functions (ODFs) and {111} PFs of these samples can be observed in Figure 8 for the Euler angle φ₂ = 0°, 15°, 30°, 45°, 65°, and 75°.

For face-centered cubic (CFC) structures, the rolling texture components are Copper {112} <111>, S {231} <346>, Brass {110}<112>, and Goss {110}<001> that are indicated in the figure; these texture components were also analysed in some authors’ works in order to characterize deformed samples [13,33]. For texture evolution, the initial grain orientation is an important parameter, and both samples have differences in their crystallographic orientation prior to deformation, as previously analysed in Figure 2. For the strain of 0.11, the two samples (Ni matrix and Ni/CNTs nanocomposites) are mostly characterized by random orientation. However, it becomes possible to determine some orientations with higher intensity. For the Ni matrix, the highest intensity occurs for the Cube and Brass rolling texture components, while for the nanocomposite, the Cube component has the highest intensity. With increasing strain up to 0.36, the Ni matrix develops texture components mainly composed of Copper and Goss, and starts to create other textures such as bb, A1, and C.

In contrast, the cold-rolled nanocomposite with the same strain mainly exhibits the Copper and Taylor texture components. For the higher strain (0.69), the texture shows similarities between Ni and Ni/CNTs samples. This is most evident in the {111} pole figure’s components, where the highest intensities are the Goss, Copper, and Taylor for both samples, with A and C being more significant for the Ni matrix, and Cube for nanocomposites. The Ni sample reveals that from 0.36, the shear texture components (bb, A1, and C) also occur beside the rolling texture components. Jeong et al. [41] revealed that for cold-rolled samples, the development of shear texture for inhomogeneous rolling due to the deformation geometry and friction between the rolls and sheet is possible. The main shear textures observed in FCC metals consist mainly of {001}<110>, {111}<112>, and {111}<110> orientations, which are in agreement to some extent in this study.

These results show that the Ni and Ni/CNTs exhibit different textures for the 0.11 and 0.36 strains and become similar for higher strains. This behaviour can also contribute to the differences observed between the hardness values of the cold-rolled nanocomposites and Ni matrix at the same deformations.

The different stages of microstructural evolution during the cold-rolling can give rise to these differences. Figure 9 shows the IPF maps and Kernel misorientation maps of the Ni and Ni/CNT samples deformed with strains of 0.36 and 0.69. It is possible to observe in this figure that the Ni matrix presents micro sheer bands for the strain of 0.36. For strain 0.69, these bands and regions with domains (sub-grains) are more evident. For the nanocomposites cold-rolled up to a strain of 0.36, the dislocation reorganization concerning the sintered sample is visible, and the dislocation density is lower than that of the Ni sample. For the strain of 0.69, the microstructure is similar to the Ni sample, making it possible to identify the micro shear bands. These observations are being referred to by other authors [14] regarding the microstructural evolution during cold-rolling.

4. Conclusions

This study evaluated the cold-rolling of Ni/CNT nanocomposites under different strains (between 0.11 and 0.92) to investigate the effect of the reinforcement on deformation behaviour. The Ni matrix was also produced and subjected to cold-rolling and analysed for comparison purposes. The samples were initially produced by powder metallurgy.

The microstructural characterization of cold-rolled specimens revealed that the nanocomposites exhibited a different microstructural evolution from the Ni matrix during the cold-rolling. This was attributed to the presence of CNTs that influenced the density of dislocations and initial crystallographic orientation of the matrix during production. The nanocomposites revealed a higher density of GNDs than the Ni matrix after the sintering process due to the clusters of the CNTs, mainly at grain boundaries, that hindered the recovery process during sintering, and that did not occur in the same proportion in both samples. Consequently, the nanocomposites showed a different crystallographic orientation than the Ni matrix after sintering, which affected the texture evolution when deformation was applied.

In both samples, when low strains were applied (ε = 0.11 and 0.22), a decrease in the density of dislocations was observed in GNDs’ density maps due to their rearrangement and annihilation during the application of these strains. This effect is attributed to the Bauschinger effect. With these lower strains, there were some differences between Ni and nanocomposite microstructures since the starting point (the sintering samples) had significant remaining deformation in the microstructures.

For higher strains (ε = 0.36 and 0.69), the dislocation density gradually increased until it reached a critical point, beyond which this density remained roughly stable. The sample’s microstructures also became similar.

The hardness of the matrix increased with increasing strain due to the strain hardening. However, for nanocomposites, the observed behaviour was different. For strains of 0.11 and 0.22, a softening (from 180 HV0.01 to 132 HV 0.01) of the sample was observed due to the significant dislocation density decrease—Bauschinger effect—as already mentioned. For higher strains (ε = 0.36 to 0.92), the overall hardness increased for both samples. However, the nanocomposites showed values lower than for the matrix under the same conditions (305 and 263 HV 0.01 for the Ni and Ni/CNT with strains of 0.92, for example), which was not foreseen.

These unpredicted hardness values can be justified by the different texture evolution of Ni and Ni/CNTs during the plastic deformation because the samples exhibited different initial crystalline orientations that affected the texture evolution during the cold-rolling.

The texture evolution was investigated using both ODFs and IPFs’ images. For the matrix, typical rolling components, such as Copper and Brass, were observed, resultant, and expected given the used deformation method. However, applying strains of 0.92 induced the formation of a shear texture (revealed by the components bb, A1, and C) that can be explained due to an inhomogeneous rolling in the sheet thickness. For the nanocomposites, the main rolling components observed were Copper and Taylor for the strain of 0.36, and the shear texture components were observed sooner, with strains from 0.69 up. This difference between Ni and nanocomposite texture evolution can be justified by the different initial crystalline orientations and microstructures that affect the texture evolution during cold-rolling.

In this sense, it is possible to conclude that the presence of CNTs can influence the deformation behaviour of nanocomposites, especially with lower strains, and especially due to the formation of different microstructures and crystallographic orientations after sintering. This influence also affects their mechanical properties, such as microhardness.