Strengthening Mechanisms of Aluminum Matrix Nanocomposites Reinforced with CNTs Produced by Powder Metallurgy

: The present work aims to investigate the strengthening mechanisms in aluminum matrix nanocomposites reinforced by carbon nanotubes (CNTs). A classical powder metallurgy route produced Al-CNT nanocomposites using ultrasonication and ball milling as dispersion/mixture techniques. The microstructural characterization is crucial for this study to reach the objective, being performed mainly by electron backscattered diffraction (EBSD), transmission electron microscopy (TEM), and high-resolution TEM (HRTEM). Uniform dispersion without damaging the CNTs structure is the key for the nanocomposite by powder metallurgy production process. The reinforcement effect occurs due to several strengthening mechanisms that act simultaneously. For the Al-CNT nanocomposites produced by ultrasonication as a dispersion/mixture technique, the observed improvement in the mechanical properties of nanocomposites can be attributed to the load transfer from the matrix to the CNTs. The strain hardening and the second-phase hardening can also have a small contribution to the strengthening of the nanocomposites. This work presents an overall view of the strengthening mechanisms of Al-CNT nanocomposites produced with 1.00 vol % CNTs. A classical powder metallurgy route was performed, using two different dispersion/mixture methods: ultrasonication and its combination with ball milling. This step of the production of nanocomposites is crucial to the dispersion of CNTs, ensuring its efﬁciency while obtaining a damage-free CNT structure so that the strengthening effect occurs. In this sense, the uniform dispersion of the CNTs with minimal damage is the key to improving the mechanical properties of the nanocomposites.

To successfully produce these types of nanocomposites and assure the efficiency of the strengthening mechanisms, it is crucial to obtain a uniform MWCNTs dispersion in the aluminum matrix. In this sense, it is essential to carefully choose the dispersion/mixing methods and conditions in the production route to obtain the expected result. Ultrasonication and its combination with ball milling were the mixture and dispersion techniques performed. The ultrasonic technique was used for the simultaneous dispersion and mixing of the MWCNTs with Al powders, using the best conditions as reported in previous work [5,30,45]; that is, this method was applied simultaneously to Al and CNTs powders in isopropanol for 15 min. The ultrasonic method is used for short periods under the same conditions to untangle some CNT clusters in a second dispersion route. After that, the dispersion and mixture of Al powder and the untangled MWCNTs were conducted by ball milling, for 360 min, with 150 rpm and a ball-to-powder ratio of 20:1. The mixtures were pressed at 300 MPa and sintered at 640 • C for 90 and 120 min under a vacuum at 10 −2 Pa. Heat treatments were also conducted at 480 • C for 90 min. Figure 1 shows a summary of the production conditions of the nanocomposites and the samples without reinforcement. Al and MWCNTs (functionalized and not functionalized) were fully characterized in previous works [30,32,51,52], including the size and respective distribution of aluminum powder particles, as received and after dispersion, as well as the morphology of MWCNTs, which can influence the result.
To successfully produce these types of nanocomposites and assure the efficiency of the strengthening mechanisms, it is crucial to obtain a uniform MWCNTs dispersion in the aluminum matrix. In this sense, it is essential to carefully choose the dispersion/mixing methods and conditions in the production route to obtain the expected result. Ultrasonication and its combination with ball milling were the mixture and dispersion techniques performed. The ultrasonic technique was used for the simultaneous dispersion and mixing of the MWCNTs with Al powders, using the best conditions as reported in previous work [5,30,45]; that is, this method was applied simultaneously to Al and CNTs powders in isopropanol for 15 min. The ultrasonic method is used for short periods under the same conditions to untangle some CNT clusters in a second dispersion route. After that, the dispersion and mixture of Al powder and the untangled MWCNTs were conducted by ball milling, for 360 min, with 150 rpm and a ball-to-powder ratio of 20:1. The mixtures were pressed at 300 MPa and sintered at 640 °C for 90 and 120 min under a vacuum at 10 −2 Pa. Heat treatments were also conducted at 480 °C for 90 min. Figure 1 shows a summary of the production conditions of the nanocomposites and the samples without reinforcement. Optical microscopy (OM) and the image analysis software was used to overview all the produced samples and quantify the pores and agglomerates present in each one. For this, a DM 4000 M optical microscope equipped with a Leica DFC 420 camera (Leica Microsystems, Wetzlar, Germany) and the Leica Application Suite software (Leica Microsystems, Wetzlar, Germany) were used for the image analysis. The average grain value was obtained by measuring up to 500 grains per sample. Evaluating the number and average diameter of the CNT agglomerates used five fields of 489 × 653 µm 2 of OM imagens.
EBSD is a promising characterization technique with great potential that allows obtaining a wide variety of results to study in detail the nanocomposites microstructures. Therefore, the samples were analyzed by high-resolution SEM (Thermo Fisher Scientific QUANTA 400 FEG SEM, Thermo Fisher Scientific, Hillsboro, OR, USA), which is coupled to an EBSD detector TSL-EDAX EBSD Unit (EDAX Inc. (Ametek), Mahwah, NJ, USA). The raw data obtained are submitted to a dilatation routine clean-up. A grain tolerance angle of 15° and 2 points as minimum grain size is determined, which is crucial for preventing doubtful results. The cleaned data are used to elaborate the wide range of maps and Optical microscopy (OM) and the image analysis software was used to overview all the produced samples and quantify the pores and agglomerates present in each one. For this, a DM 4000 M optical microscope equipped with a Leica DFC 420 camera (Leica Microsystems, Wetzlar, Germany) and the Leica Application Suite software (Leica Microsystems, Wetzlar, Germany) were used for the image analysis. The average grain value was obtained by measuring up to 500 grains per sample. Evaluating the number and average diameter of the CNT agglomerates used five fields of 489 × 653 µm 2 of OM imagens.
EBSD is a promising characterization technique with great potential that allows obtaining a wide variety of results to study in detail the nanocomposites microstructures. Therefore, the samples were analyzed by high-resolution SEM (Thermo Fisher Scientific QUANTA 400 FEG SEM, Thermo Fisher Scientific, Hillsboro, OR, USA), which is coupled to an EBSD detector TSL-EDAX EBSD Unit (EDAX Inc. (Ametek), Mahwah, NJ, USA). The raw data obtained are submitted to a dilatation routine clean-up. A grain tolerance angle of 15 • and 2 points as minimum grain size is determined, which is crucial for preventing doubtful results. The cleaned data are used to elaborate the wide range of maps and graphs with TSL OIM Analysis 5.2. The overall EBSD maps-inverse pole figures (IPFs) and grain orientation spread (GOS)-were obtained under conditions similar to those described in previous work [45].
Along with this, the grain average misorientation (GAM) maps were used to separate the grains into different categories (recrystallized, recovered, deformed). Using the same software, pole figures (PF) were also drawn to understand preferential crystallographic orientations and even the presence of sharp textures. The ATEX software was used, similarly to previous work [53], to elaborate maps and estimate the geometrically necessary dislocations (GNDs) density [54].
High-resolution TEM was very advantageous for the in-depth study of the bonding of CNTs to the metal matrix, the formation of second-phase particles resulting from the reaction between the CNTs and matrix, and the presence of dislocations and dislocations cells. A high-resolution TEM (HRTEM) (Thermo Fisher Scientific, Hillsboro, OR, USA) was used for this purpose.
Mechanical characterization was conducted to comprehend the effect of the CNTs on reinforced Al. In this sense, the samples were submitted to Vickers microhardness tests, which were performed using a Duramin-1 durometer (Duramin-1; Struers A/S, Ballerup, Denmark) and used for approximately 12 indentations in each sample, with a load of 98 mN. The tensile tests were performed at an initial strain rate of about 8 × 10 −3 s −1 ; three specimens were tested for each condition. The geometry of the tensile specimens is described elsewhere [51]. Figure 2 shows the optical microscopy (OM) images of the nanocomposite produced by ultrasonication or ball milling as dispersion/mixture techniques, using CNTs functionalized or non-functionalized, and during different sintering times (90 and 120 min). All nanocomposites exhibit a microstructure consisting of equiaxed aluminum grains with pores and CNT agglomerates mainly in grain boundaries, which are not distinguishable by OM (dark phase). With enhanced detail, Figure 3 shows a scanning electron microscopy (SEM) image of the nanocomposite produced using ultrasonication, confirming the presence of CNTs in the pores and grain boundaries of the Al matrix. The presence of pores characterizes the samples produced by powder metallurgy; nevertheless, it is not guaranteed that the CNTs fill all pores.

Results and Discussion
Dispersion/mixture techniques, sintering time, and the use of CNTs functionalized treatment strongly influence the dispersion of the CNTs. The nanocomposites produced by ball milling and ultrasonication with functionalized CNTs exhibit a smaller amount (2.68% and 3.05%, respectively) of CNT agglomerates. This can be an indication of a uniform dispersion obtained for these nanocomposites. The nanocomposites produced by ultrasonication revealed a percentage of CNT agglomerates of 5.95% and a maximum size of 78 µm [30]. Increasing the sintering time promotes the decrease of the amount and size of CNT agglomerates to 4.49% and 61.4 µm, respectively. Figure 4 shows the influence of dispersion/mixture processes, sintering time, and functionalization of the CNTs on the mechanical properties of nanocomposites. This evolution shows an increase in the hardness and ultimate tensile strength and a decrease in the nanocomposites' elongation concerning the Al matrix's same properties produced under the same conditions (yield strength of 23 MPa, an ultimate tensile strength of 66 MPa, an elongation of 28%, and 34 HV 0.1 [51]). Although the properties reported for Al-CNT nanocomposites depend significantly on the type of dispersion/mixing and processing methods, these properties are according to the expected ones [31,[34][35][36][37][38][39].
nanocomposites exhibit a microstructure consisting of equiaxed aluminum grains with pores and CNT agglomerates mainly in grain boundaries, which are not distinguishable by OM (dark phase). With enhanced detail, Figure 3 shows a scanning electron microscopy (SEM) image of the nanocomposite produced using ultrasonication, confirming the presence of CNTs in the pores and grain boundaries of the Al matrix. The presence of pores characterizes the samples produced by powder metallurgy; nevertheless, it is not guaranteed that the CNTs fill all pores.  Dispersion/mixture techniques, sintering time, and the use of CNTs functionalized treatment strongly influence the dispersion of the CNTs. The nanocomposites produced by ball milling and ultrasonication with functionalized CNTs exhibit a smaller amount (2.68% and 3.05%, respectively) of CNT agglomerates. This can be an indication of a uniform dispersion obtained for these nanocomposites. The nanocomposites produced by ultrasonication revealed a percentage of CNT agglomerates of 5.95% and a maximum size of 78 µm [30]. Increasing the sintering time promotes the decrease of the amount and size of CNT agglomerates to 4.49% and 61.4 µm, respectively. Figure 4 shows the influence of dispersion/mixture processes, sintering time, and functionalization of the CNTs on the mechanical properties of nanocomposites. This evolution shows an increase in the hardness and ultimate tensile strength and a decrease in the nanocomposites' elongation concerning the Al matrix's same properties produced under the same conditions (yield strength of 23 MPa, an ultimate tensile strength of 66 MPa, an elongation of 28%, and 34 HV 0.1 [51]). Although the properties reported for Al-CNT nanocomposites depend significantly on the type of dispersion/mixing and processing methods, these properties are according to the expected ones [31,[34][35][36][37][38][39]. All nanocomposites revealed a higher hardness and ultimate tensile strength than Al produced in the same conditions, corroborating the strengthening effect of introducing All nanocomposites revealed a higher hardness and ultimate tensile strength than Al produced in the same conditions, corroborating the strengthening effect of introducing the CNTs into the metal matrix. Furthermore, these values also confirm the quality of the CNTs dispersion.
Although having a more significant amount of CNT clusters, the nanocomposites produced with ultrasonication as the dispersion/mixture process present a greater increase in hardness (from 34 to 50 HV0.01) and ultimate tensile strength (from 66 to 196 MPa for 90 min of sintering and from 72 to 257 MPa, increasing the sintering for 120 min), revealing the most favorable conditions to make these nanocomposites.
While the Al-CNT nanocomposites produced with functionalized CNTs present a better dispersion, the structural damage caused by the functionalization treatment affects the CNTs' strengthening potential. The increase in hardness was only from 34 to 42 HV0.01, and the increase in ultimate tensile strength was from 66 to 143 MPa. These results were similar to those observed for the Ni-CNTs nanocomposites in our previous work [52] and to the work of Xu et al. [36]. It was clear that the functionalization treatment of CNTs and ball milling promotes defects in the structure, which affected the hardness value and strength obtained for the nanocomposites.
Nanocomposites produced by ball milling also improve the mechanical properties, but the increase is slightly lower than expected, especially the ultimate tensile strength (from 87 to 216 MPa). The microstructural characterization revealed a more uniform CNTs dispersion with smaller-sized agglomerates, and therefore, a more significant improvement in mechanical properties than ultrasonication was expected. The reason for this is the damage suffered by CNTs during the functionalization process. The ball milling method promotes damage to the structure of CNTs, which was proved by HRTEM and Raman spectroscopy in previous work [32], reducing their potential for strengthening. For effective strengthening, there must be not only a uniform dispersion of the CNTs strongly bonded to the matrix but also a preserved CNTs structure without damage.
Since the Al-CNT nanocomposites produced by ball milling as the dispersion/mixture process or using functionalized CNTs reveal an increase in the mechanical properties that were lower than expected, the present work focuses on the study of the strengthening mechanisms mainly in the samples produced by ultrasonication.
The nanocomposite fracture surfaces were analyzed to investigate the load transfer mechanism. In this analysis, fractured and pulled-out CNTs were present in all nanocomposites. Figure 5 shows illustrative SEM images of the fracture surface of the Al-CNTs produced by ultrasonication.
High-magnification SEM images show that ruptured or pulled-out CNTs can be identified. These observations prove that one of the strengthening mechanisms present is load transfer. When the matrix undergoes plastic deformation, nucleation and cracks propagation occur. The presence of CNTs will allow them to act as bridges and restrain the crack growth.
The load transfer is one of the mechanisms that most contribute to the strengthening effect in nanocomposites reinforced with CNTs [9,55,56]. However, other mechanisms are simultaneously present, playing a role in increasing the mechanical properties of nanocomposites. In a previous study [53], the strengthening mechanisms occurring in nickel nanocomposites reinforced by CNTs were carefully investigated. Load transfer and increased dislocation density seem to strongly affect the properties and microstructure of the nanocomposite. The grain refinement and the presence of second phase particles have a small contribution to the strengthening of this nanocomposite.
Microstructural features such as grain size and orientation, increased dislocation density, and the formation of second-phase particles can also contribute to the nanocomposites strengthening [6,10,23,36,57]. The final microstructural features are perceived by inverse pole figures (IPF) maps, image quality (IQ) maps with delineated high-and lowangle boundaries, grain orientation spread (GOS) maps, and pole figures (PF) for the nanocomposites produced by ultrasonication, which can be observed in Figure 6. Figure 6 reveals that similar grain sizes, low-angle boundary densities, and grain orientations are observed for Al and Al-CNTs produced by ultrasonication. The GOS maps show the main difference between these samples. Although not very noticeable, the nanocomposites have a lower percentage of fully recrystallized grains (in blue), along with a high content of slightly deformed grains (represented by green) and some significantly deformed grains (orange and red). These results show that CNTs affect the extension of the recovery and recrystallization processes of the Al matrix during the sintering. In the case of Ni-CNT composites produced under similar conditions [45,53], more significant differences were observed in the texture and density of low-angle boundaries induced by the introduction of CNTs.
To further investigate the effect of the CNTs on the recovery and recrystallization processes, a heat treatment was conducted at 480 • C for 90 min under vacuum. Figure 7 shows the grain size, GOS, and GAM maps for the Al and Al-CNTs produced using ultrasonication, sintered for 90 min, and then heat-treated. Based on these results, recrystallized grains are observed for the heat-treated Al samples. Moreover, the grain size is smaller for the Al than Al-CNT samples due to the recovery and recrystallization process. The nanocomposites show a lower density of small, recrystallized grains and the presence of deformed grains.
The nanocomposites reveal a slight increase in local misorientation spread than the Al samples produced under the same conditions, proving some effect of the CNTs on the dislocation arrangement during the sintering. However, the impact is not as significant as that observed for Ni-CNT nanocomposites [53]. Figure 8, using high magnifications, exhibit better a slight difference between the samples. This figure shows the IPF and IQ maps with high-and low-angle grain boundaries and GND estimated density maps for the Al and Al-CNT samples produced by ultrasonication. For the nanocomposites, a slightly higher density of low-angle grain boundaries was observed associated with a somewhat higher GND density (4.99 × 10 14 m −2 and 5.39 × 10 14 m −2 , respectively). The increase in the dislocation density can be attributed to the pinning effect of the CNTs, as mentioned by Chen et al. [50], or even to the impact of thermal mismatch strain [46]. the CNTs into the metal matrix. Furthermore, these values also confirm the quality of the CNTs dispersion.
Although having a more significant amount of CNT clusters, the nanocomposites produced with ultrasonication as the dispersion/mixture process present a greater increase in hardness (from 34 to 50 HV0.01) and ultimate tensile strength (from 66 to 196 MPa for 90 min of sintering and from 72 to 257 MPa, increasing the sintering for 120 min), revealing the most favorable conditions to make these nanocomposites.
While the Al-CNT nanocomposites produced with functionalized CNTs present a better dispersion, the structural damage caused by the functionalization treatment affects the CNTs' strengthening potential. The increase in hardness was only from 34 to 42 HV0.01, and the increase in ultimate tensile strength was from 66 to 143 MPa. These results were similar to those observed for the Ni-CNTs nanocomposites in our previous work [52] and to the work of Xu et al. [36]. It was clear that the functionalization treatment of CNTs and ball milling promotes defects in the structure, which affected the hardness value and strength obtained for the nanocomposites.
Nanocomposites produced by ball milling also improve the mechanical properties, but the increase is slightly lower than expected, especially the ultimate tensile strength (from 87 to 216 MPa). The microstructural characterization revealed a more uniform CNTs dispersion with smaller-sized agglomerates, and therefore, a more significant improvement in mechanical properties than ultrasonication was expected. The reason for this is the damage suffered by CNTs during the functionalization process. The ball milling method promotes damage to the structure of CNTs, which was proved by HRTEM and Raman spectroscopy in previous work [32], reducing their potential for strengthening. For effective strengthening, there must be not only a uniform dispersion of the CNTs strongly bonded to the matrix but also a preserved CNTs structure without damage.
Since the Al-CNT nanocomposites produced by ball milling as the dispersion/mixture process or using functionalized CNTs reveal an increase in the mechanical properties that were lower than expected, the present work focuses on the study of the strengthening mechanisms mainly in the samples produced by ultrasonication.
The nanocomposite fracture surfaces were analyzed to investigate the load transfer mechanism. In this analysis, fractured and pulled-out CNTs were present in all nanocomposites. Figure 5 shows illustrative SEM images of the fracture surface of the Al-CNTs produced by ultrasonication. Microstructural features such as grain size and orientation, increased dislocation density, and the formation of second-phase particles can also contribute to the nanocomposites strengthening [6,10,23,36,57]. The final microstructural features are perceived by inverse pole figures (IPF) maps, image quality (IQ) maps with delineated high-and lowangle boundaries, grain orientation spread (GOS) maps, and pole figures (PF) for the nanocomposites produced by ultrasonication, which can be observed in Figure 6.
Metals 2021, 11, x FOR PEER REVIEW 9 of 18 (g) (h) Figure 6. IPF maps, IQ maps with high-and low-angle boundaries delineated, GOS maps, and PF of (a-c,g) Al and (d-f,h) Al-CNTs produced by ultrasonication. Figure 6 reveals that similar grain sizes, low-angle boundary densities, and grain orientations are observed for Al and Al-CNTs produced by ultrasonication. The GOS maps show the main difference between these samples. Although not very noticeable, the nanocomposites have a lower percentage of fully recrystallized grains (in blue), along with a high content of slightly deformed grains (represented by green) and some significantly deformed grains (orange and red). These results show that CNTs affect the extension of the recovery and recrystallization processes of the Al matrix during the sintering. In the case of Ni-CNT composites produced under similar conditions [45,53], more significant differences were observed in the texture and density of low-angle boundaries induced by the introduction of CNTs.
To further investigate the effect of the CNTs on the recovery and recrystallization processes, a heat treatment was conducted at 480 °C for 90 min under vacuum. Figure 7 shows the grain size, GOS, and GAM maps for the Al and Al-CNTs produced using ultrasonication, sintered for 90 min, and then heat-treated. Based on these results, recrystallized grains are observed for the heat-treated Al samples. Moreover, the grain size is smaller for the Al than Al-CNT samples due to the recovery and recrystallization process. The nanocomposites show a lower density of small, recrystallized grains and the presence of deformed grains. The TEM and HRTEM images were also used to characterize, with more detail, some of the microstructural features of the nanocomposites. Figure 9 shows images of Al and Al-CNT as produced by ultrasonication and heat-treated. For the nanocomposites, there is a higher density of dislocations. This confirms that the introduction of CNTs into the Al matrix affects the dislocation motion and rearrangement during the sintering. In the Al samples, the microstructure presents equiaxed grains. Although the CNT clusters are observed by OM and SEM images, CNTs are also embedded in the matrix. These CNTs strongly bound to the metallic matrix are essential for the strengthening effect.
TEM analysis shows the presence of the Al 4 C 3 phase, as shown in Figure 10. In these images, it is possible to see the presence of these second-phase particles, especially near the CNTs. Even in small quantities, these particles can contribute to the strengthening effect observed in the nanocomposites, improving the bonding between the CNTs and the matrix [37,[58][59][60] as well as acting as obstacles to the motion of dislocation [36,60] and grain boundaries [36], promoting grain refining.
The nanocomposite application may be compromised if, during the application, reaction or damage to the reinforcement occurs due to the service temperature. In this sense, the observations in TEM and HRTEM were made to evaluate the reaction and structure of the CNTs. The heat treatment of the nanocomposites at 480 • C did not significantly change the formation of Al 4 C 3 particles or even in the damage of CNTs. Figure 10e,f shows the results of the TEM images for the heat-treated nanocomposites samples. The nanocomposites with and without heat treatment show similar results regarding Al 4 C 3 particles close to the CNTs. The heat treatment did not significantly affect the reaction between the CNTs and the matrix.
Some investigations evaluated the effect of the Al 4 C 3 formation on the improvement also of the load transfer mechanism. Kwon et al. [48] observed two types of Al 4 C 3 particles in Al-CNT nanocomposites (1.00 vol % CNTs), which they defined as dumbbell and tube types. Dumbbell-shaped ones are derived from the CNT tip, while the tube shapes come from the reaction with an imperfect CNT. The authors also mentioned that these carbides play a vital role in load transfer from the matrix to the reinforcement, since the fracture surfaces observed after tensile tests showed bridging and some broken CNTs without any CNT pulled out. This CNTs behavior is a consequence of their strong bonding to the metallic matrix. Zhou et al. [61] also studied the formation of these particles and their relationship with CNTs. These authors mention that the Al 4 C 3 formed at the tip of MWCNTs can grow in the <100> direction of the CNTs, while those originated from nano defects can grow simultaneously in the <110> and <100> directions. An orientation relationship of Al(111)//CNT(002) was observed that means that the CNT reacted with the matrix and formed the carbine, which maintained the low-energy interface. The results confirmed that the Al 4 C 3 contributes to the enhanced load transfer mechanism to the CNTs.
(g) (h) Figure 6. IPF maps, IQ maps with high-and low-angle boundaries delineated, GOS maps, and PF of (a-c,g) Al and (d-f,h) Al-CNTs produced by ultrasonication. Figure 6 reveals that similar grain sizes, low-angle boundary densities, and grain orientations are observed for Al and Al-CNTs produced by ultrasonication. The GOS maps show the main difference between these samples. Although not very noticeable, the nanocomposites have a lower percentage of fully recrystallized grains (in blue), along with a high content of slightly deformed grains (represented by green) and some significantly deformed grains (orange and red). These results show that CNTs affect the extension of the recovery and recrystallization processes of the Al matrix during the sintering. In the case of Ni-CNT composites produced under similar conditions [45,53], more significant differences were observed in the texture and density of low-angle boundaries induced by the introduction of CNTs.
To further investigate the effect of the CNTs on the recovery and recrystallization processes, a heat treatment was conducted at 480 °C for 90 min under vacuum. Figure 7 shows the grain size, GOS, and GAM maps for the Al and Al-CNTs produced using ultrasonication, sintered for 90 min, and then heat-treated. Based on these results, recrystallized grains are observed for the heat-treated Al samples. Moreover, the grain size is smaller for the Al than Al-CNT samples due to the recovery and recrystallization process. The nanocomposites show a lower density of small, recrystallized grains and the presence of deformed grains. The nanocomposites reveal a slight increase in local misorientation spread than the Al samples produced under the same conditions, proving some effect of the CNTs on the dislocation arrangement during the sintering. However, the impact is not as significant as that observed for Ni-CNT nanocomposites [53]. Figure 8, using high magnifications, exhibit better a slight difference between the  , using high magnifications, exhibit better a slight difference between the samples. This figure shows the IPF and IQ maps with high-and low-angle grain boundaries and GND estimated density maps for the Al and Al-CNT samples produced by ultrasonication. For the nanocomposites, a slightly higher density of low-angle grain boundaries was observed associated with a somewhat higher GND density (4.99 × 10 14 m −2 and 5.39 × 10 14 m −2 , respectively). The increase in the dislocation density can be attributed to the pinning effect of the CNTs, as mentioned by Chen et al. [50], or even to the impact of thermal mismatch strain [46]. The TEM and HRTEM images were also used to characterize, with more detail, some of the microstructural features of the nanocomposites. Figure 9 shows images of Al and Al-CNT as produced by ultrasonication and heat-treated. For the nanocomposites, there is a higher density of dislocations. This confirms that the introduction of CNTs into the Al matrix affects the dislocation motion and rearrangement during the sintering. In the Al samples, the microstructure presents equiaxed grains. Although the CNT clusters are observed by OM and SEM images, CNTs are also embedded in the matrix. These CNTs strongly bound to the metallic matrix are essential for the strengthening effect.  The TEM and HRTEM images were also used to characterize, with more detail, some of the microstructural features of the nanocomposites. Figure 9 shows images of Al and Al-CNT as produced by ultrasonication and heat-treated. For the nanocomposites, there is a higher density of dislocations. This confirms that the introduction of CNTs into the Al matrix affects the dislocation motion and rearrangement during the sintering. In the Al samples, the microstructure presents equiaxed grains. Although the CNT clusters are observed by OM and SEM images, CNTs are also embedded in the matrix. These CNTs strongly bound to the metallic matrix are essential for the strengthening effect. TEM analysis shows the presence of the Al4C3 phase, as shown in Figure 10. In these images, it is possible to see the presence of these second-phase particles, especially near the CNTs. Even in small quantities, these particles can contribute to the strengthening effect observed in the nanocomposites, improving the bonding between the CNTs and the matrix [37,[58][59][60] as well as acting as obstacles to the motion of dislocation [36,60] and grain boundaries [36], promoting grain refining. The nanocomposite application may be compromised if, during the application, reaction or damage to the reinforcement occurs due to the service temperature. In this sense, the observations in TEM and HRTEM were made to evaluate the reaction and structure of the CNTs. The heat treatment of the nanocomposites at 480 °C did not significantly change the formation of Al4C3 particles or even in the damage of CNTs. Figure 10e,f shows the results of the TEM images for the heat-treated nanocomposites samples. The nano- The formation of this second phase in Al-CNT nanocomposites was also observed in previous work [32,51] and by other authors [58,59]. Note that Yu et al. [37] observed, using HRTEM, the formation of an Al 4 C 3 phase closely connected to the open end of a CNT in a nanocomposite with an Al-Mg alloy as the matrix.
These aluminum carbides are formed by the total or partial chemical reaction between the CNTs and the matrix, as schematically illustrated in Figure 11. The partial reaction results in interfacial Al 4 C 3 particles between the matrix and the reinforcement, improving their bonding and contributing to the efficiency of the load transfer mechanism. When the reaction is total, the full transformation into individual Al 4 C 3 particles strengthens the nanocomposite, acting as a barrier to the dislocation motion and rearrangement and grain boundaries, preventing grain growth. Some investigations evaluated the effect of the Al4C3 formation on the improvement also of the load transfer mechanism. Kwon et al. [48] observed two types of Al4C3 particles in Al-CNT nanocomposites (1.00 vol % CNTs), which they defined as dumbbell and tube types. Dumbbell-shaped ones are derived from the CNT tip, while the tube shapes come from the reaction with an imperfect CNT. The authors also mentioned that these carbides play a vital role in load transfer from the matrix to the reinforcement, since the fracture surfaces observed after tensile tests showed bridging and some broken CNTs without any CNT pulled out. This CNTs behavior is a consequence of their strong bonding to the metallic matrix. Zhou et al. [61] also studied the formation of these particles and their relationship with CNTs. These authors mention that the Al4C3 formed at the tip of MWCNTs can grow in the <100> direction of the CNTs, while those originated from nano defects can grow simultaneously in the <110> and <100> directions. An orientation relationship of Al(111)//CNT(002) was observed that means that the CNT reacted with the matrix and formed the carbine, which maintained the low-energy interface. The results confirmed that the Al4C3 contributes to the enhanced load transfer mechanism to the CNTs.
The increased dislocation density close to the CNTs and the Al4C3 can also assist the strengthening of the nanocomposites. The nanocomposites produced by powder metallurgy increase dislocations during the different steps of production caused by deformation. Previous works [5,45,52,53] already demonstrated that the CNT presence significantly affects the recovery, recrystallization, and grain growth during the sintering of Ni-CNT nanocomposites.
In Figure 12, it is possible to see HRTEM images and FFT of three different regions The increased dislocation density close to the CNTs and the Al 4 C 3 can also assist the strengthening of the nanocomposites. The nanocomposites produced by powder metallurgy increase dislocations during the different steps of production caused by deformation. Previous works [5,45,52,53] already demonstrated that the CNT presence significantly affects the recovery, recrystallization, and grain growth during the sintering of Ni-CNT nanocomposites.
In Figure 12, it is possible to see HRTEM images and FFT of three different regions of the nanocomposites produced by ultrasonication: a region that corresponds to a CNT, a lattice of an Al 4 C 3 particle, and an area with the presence of dislocation around the Al 4 C 3 particle.

Conclusions
This work presents an overall view of the strengthening mechanisms of Al-CNT nanocomposites produced with 1.00 vol % CNTs. A classical powder metallurgy route was performed, using two different dispersion/mixture methods: ultrasonication and its combination with ball milling. This step of the production of nanocomposites is crucial to the dispersion of CNTs, ensuring its efficiency while obtaining a damage-free CNT structure so that the strengthening effect occurs. In this sense, the uniform dispersion of the CNTs with minimal damage is the key to improving the mechanical properties of the nanocomposites.
The mentioned improvement is a consequence of several strengthening mechanisms simultaneously, which are influenced by the presence of CNTs. The load transfer mechanism seems to contribute the most to the increase in the mechanical properties of the nanocomposites. The fracture surface showing pull-out and fractured CNTs confirms the critical role of this mechanism.
Strain hardening and second-phase hardening are also identified, since the formation of nanometric Al4C3 particles was observed together with a slight increase in dislocation density. Regarding grain size and texture hardening, as no significant differences were detected between the nanocomposites and the matrix, it was impossible to confirm their contributions to the Al-CNTs strengthening produced under these conditions.

Conclusions
This work presents an overall view of the strengthening mechanisms of Al-CNT nanocomposites produced with 1.00 vol % CNTs. A classical powder metallurgy route was performed, using two different dispersion/mixture methods: ultrasonication and its combination with ball milling. This step of the production of nanocomposites is crucial to the dispersion of CNTs, ensuring its efficiency while obtaining a damage-free CNT structure so that the strengthening effect occurs. In this sense, the uniform dispersion of the CNTs with minimal damage is the key to improving the mechanical properties of the nanocomposites.
The mentioned improvement is a consequence of several strengthening mechanisms simultaneously, which are influenced by the presence of CNTs. The load transfer mechanism seems to contribute the most to the increase in the mechanical properties of the nanocomposites. The fracture surface showing pull-out and fractured CNTs confirms the critical role of this mechanism.
Strain hardening and second-phase hardening are also identified, since the formation of nanometric Al 4 C 3 particles was observed together with a slight increase in dislocation density. Regarding grain size and texture hardening, as no significant differences were detected between the nanocomposites and the matrix, it was impossible to confirm their contributions to the Al-CNTs strengthening produced under these conditions.