Investigating the Effects of Annealing on the Mechanical Properties of FFF-Printed Thermoplastics

Fused filament fabrication (FFF) is a cost-effective additive manufacturing method that makes use of thermoplastics to produce customised products. However, there are several limitations associated with FFF that are adversely affecting its growth including variety of materials, rough surface finish and poor mechanical properties. This has resulted in the development of metal-infused thermoplastics that can provide better properties. Furthermore, FFF-printed parts can be subjected to post-processes to improve their surface finish and mechanical properties. This work takes into consideration two commonly used polymeric materials, i.e., ABS (acrylonitrile butadiene styrene) and PLA (polylactic acid) and compares the results with two metal-infused thermoplastics i.e., copper-enhanced PLA and aluminium-enhanced ASA (acrylonitrile styrene acrylate). The four different materials were subjected to a post-process called annealing to enhance their mechanical properties. The effect of annealing on these four materials was investigated through dimensional analysis, ultrasonic testing, tensile testing, microstructural analysis and hardness testing. The results showed that annealing affects the materials differently. However, a correlation among ultrasonic testing, tensile testing and microstructural analysis was observed for all the materials based on their crystallinity. It was found that the semi-crystalline materials (i.e., PLA and copper enhanced PLA) showed a considerable increase in tensile strength post-annealing. However, the amorphous materials (ABS and aluminium-enhanced ASA) showed a comparatively lower increase in tensile strength, demonstrating that they were less receptive to annealing. These results were supported by higher transmission times and a high percentage of voids in the amorphous materials. The highest hardness values were observed for the ASA material and the lowest for the ABS material. This work provides a good comparison for the metal-infused thermoplastics and their applicability with the commonly used PLA and ABS materials.


Introduction
The production of customised products at low costs and required mechanical properties is in demand as it leads to higher customer satisfaction and profits for manufacturers. The impact of additive manufacturing (AM) in this context is undeniable as the key features of these methods include customisation, design freedom and cost-effectiveness. There are seven main categories of AM; vat photopolymerization, powder bed fusion, binder jetting, material jetting, sheet lamination, material extrusion and directed energy deposition [1][2][3]. These technologies can work with different materials (e.g., metals, polymers, ceramics) to manufacture customised products. However, most of these AM systems are expensive and require considerable knowledge to achieve the desired results from the products. One of the most commonly used and cost-effective methods is called fused deposition

Dimensional Analysis
FFF-printed samples are prone to warping [27] due to the nature of the process and are also affected by the dimensional precision of the printing system (Anet ET4 has a precision of ± 0.1 mm). Annealing could also lead to shrinkage and expansion [28] in any of the three axes (X, Y, and Z). Therefore, it is crucial to compare the dimensions of the FFF-printed sample before and after annealing. A Vernier calliper was used to measure the dimensions of the samples. Keeping in mind

Dimensional Analysis
FFF-printed samples are prone to warping [27] due to the nature of the process and are also affected by the dimensional precision of the printing system (Anet ET4 has a precision of ± 0.1 mm). Annealing could also lead to shrinkage and expansion [28] in any of the three axes (X, Y, and Z). Therefore, it is crucial to compare the dimensions of the FFF-printed sample before and after annealing. A Vernier calliper was used to measure the dimensions of the samples. Keeping in mind After building the parts for testing, they were subjected to a post-process called annealing. The samples were placed in a laboratory convection oven from CONTROLS Group (Milano, Italy) at a specified temperature for one hour. After the time had elapsed, the oven was turned off and the samples were cooled inside for two hours. The oven was not opened during this time, to allow for the gradual cooling of the samples. A convective oven circulates heated air around an object, providing a 360 • heat source that produces more uniform results compared to gas-fired or electric ovens. The temperatures chosen for the different materials were based on their glass transition temperatures and three different values were used for analysis. Table 1 shows the parameters used for the build operation and the annealing process. It is to be noted that different nozzles and bed temperatures can affect the properties of the materials. Therefore, values recommended by the material datasheets have been used in this work.

Dimensional Analysis
FFF-printed samples are prone to warping [27] due to the nature of the process and are also affected by the dimensional precision of the printing system (Anet ET4 has a precision of ±0.1 mm). Annealing could also lead to shrinkage and expansion [28] in any of the three axes (X, Y, and Z). Therefore, it is crucial to compare the dimensions of the FFF-printed sample before and after annealing. A Vernier calliper was used to measure the dimensions of the samples. Keeping in mind that the samples will be subjected to tensile testing afterward, they were manufactured according to BS EN ISO 527-2:2012 [29]. The dimensions of the samples are given in Figure 3.

Ultrasonic Testing
After dimensional analysis, the dog-bone samples were also subjected to ultrasonic testing before and after annealing. This is a non-destructive test (NDT) that makes use of high-frequency sound waves to detect flaws and defects in products as well as take measurements, e.g., thickness. Since annealing is used to relieve internal stresses, this test will help to quantify that reduction. Proceq PUNDIT ® PL-200 was used for the test that comprises two transducers (54 kHz) as shown in Figure 4. The use of couplant gel is recommended as it helps in facilitating the transmission of ultrasonic waves from the traducers into the test piece. Reflection and attenuation are the two methods of receiving the ultrasonic waveform. The former method makes use of only one transducer whereas the latter requires two transducers. This work used two transducers with one sending the signal through the test piece and the other receiving the signal, thus following the attenuation method. The transmission time (μs) was measured for all the different materials as shown in Figure 3. The dog-bone samples before and after annealing were subjected to ultrasonic testing at three points along their length to ascertain an average value of transmission time.

Ultrasonic Testing
After dimensional analysis, the dog-bone samples were also subjected to ultrasonic testing before and after annealing. This is a non-destructive test (NDT) that makes use of high-frequency sound waves to detect flaws and defects in products as well as take measurements, e.g., thickness. Since annealing is used to relieve internal stresses, this test will help to quantify that reduction. Proceq PUNDIT ® PL-200 was used for the test that comprises two transducers (54 kHz) as shown in Figure 4. The use of couplant gel is recommended as it helps in facilitating the transmission of ultrasonic waves from the traducers into the test piece. Reflection and attenuation are the two methods of receiving the ultrasonic waveform. The former method makes use of only one transducer whereas the latter requires two transducers. This work used two transducers with one sending the signal through the test piece and the other receiving the signal, thus following the attenuation method. The transmission time (µs) was measured for all the different materials as shown in Figure 3. The dog-bone samples before and after annealing were subjected to ultrasonic testing at three points along their length to ascertain an average value of transmission time.

Tensile testing
The samples for tensile testing were produced in accordance with BS EN ISO 527-2:2012 [29]. The dimensions of the samples are shown in Section 3.1 ( Figure 3). All the samples were of the same geometry and this helped in ensuring consistency of testing practices. The tensile testing was carried out using a Hounsfield 200 Series Universal Tensile Testing Machine with a load cell of 5 kN. The crosshead speed was 1.5 mm/s according to the standard.

Microstructural Analysis
After undertaking tensile testing on all the four different types of materials, their fractured surfaces were analysed using a JCM-5000 NeoScope™ (JEOL, Boston, MA, USA) tabletop scanning electron microscope (SEM). This analysis was conducted to investigate the fracture mechanism, stress relief (as a result of annealing) and interaction of metal particles with the plastic materials in case of Cu-PLA and Al-ASA. The fractured surfaces were carefully cut to size to fit atop the platform of the SEM. No surface treatment was applied to avoid contamination of the fractured surfaces.

Hardness Testing
There are three categories of hardness measurements, i.e., scratch, indentation and rebound. Each category comprises individual scales for measurements. This work makes use of indentation hardness to test FFF samples. Indentation hardness measures the resistance of a sample to material deformation due to a constant compression load from a sharp object. Common indentation hardness scales are Rockwell, Vickers, Brinell and Shore, amongst others. In this case, the Rockwell hardness scale L was utilized. The preparation of samples and testing procedures were based on BS EN ISO 2039-2: 2000 [30]. This test focus on determining the hardness number and whether the process of annealing has made the samples harder. The parts were 40 mm in length, 40 mm in width with 6 mm thickness. An Avery type 6402 Rockwell hardness testing machine was utilized for the samples. According to Rockwell hardness scale L, the indenter diameter was 6.35 mm. The minor load was

Tensile Testing
The samples for tensile testing were produced in accordance with BS EN ISO 527-2:2012 [29]. The dimensions of the samples are shown in Section 3.1 ( Figure 3). All the samples were of the same geometry and this helped in ensuring consistency of testing practices. The tensile testing was carried out using a Hounsfield 200 Series Universal Tensile Testing Machine with a load cell of 5 kN. The crosshead speed was 1.5 mm/s according to the standard.

Microstructural Analysis
After undertaking tensile testing on all the four different types of materials, their fractured surfaces were analysed using a JCM-5000 NeoScope™ (JEOL, Boston, MA, USA) tabletop scanning electron microscope (SEM). This analysis was conducted to investigate the fracture mechanism, stress relief (as a result of annealing) and interaction of metal particles with the plastic materials in case of Cu-PLA and Al-ASA. The fractured surfaces were carefully cut to size to fit atop the platform of the SEM. No surface treatment was applied to avoid contamination of the fractured surfaces.

Hardness Testing
There are three categories of hardness measurements, i.e., scratch, indentation and rebound. Each category comprises individual scales for measurements. This work makes use of indentation hardness to test FFF samples. Indentation hardness measures the resistance of a sample to material deformation due to a constant compression load from a sharp object. Common indentation hardness scales are Rockwell, Vickers, Brinell and Shore, amongst others. In this case, the Rockwell hardness scale L was utilized. The preparation of samples and testing procedures were based on BS EN ISO 2039-2: 2000 [30]. This test focus on determining the hardness number and whether the process of annealing has made the samples harder. The parts were 40 mm in length, 40 mm in width with 6 mm thickness. An Avery type 6402 Rockwell hardness testing machine was utilized for the samples. According to Rockwell hardness scale L, the indenter diameter was 6.35 mm. The minor load was 98.07 N and its dwell time was 8s, whereas the major load was 588.4 N with a dwell time of 15 s. The indentation was measured using a hand-held gauge. Each Rockwell hardness scale division represents a 0.002 mm vertical movement of the indenter. The Rockwell hardness number is derived from the following equation: where, HR = Rockwell number e = depth of impression after removal of the major load, in units of 0.002 mm.

Results from Dimensional Analysis
Dimensional changes in FFF-printed samples before and after annealing is a common phenomenon [31] due to the nature of this post-process. In FFF, thermoplastics are extruded out of a nozzle at their melting temperatures and are then cooled rapidly to form the desired shape. Being a poor conductor of heat, the extruded parts are cooled unevenly leading to internal stresses especially between layers. To relieve these stresses, annealing can be performed by heating the materials to their glass transition temperatures, but below their melting points. This will reorganize the internal crystalline structure and cause bigger grains to form, which will result in much stronger objects that are less prone to layer separation. In doing so, the dimensions could also be affected in any of the three axes (x, y, and z) through shrinkage or expansion. This is due to how the FFF-printed samples cool and how the tensions, stresses, and air bubbles get locked between the layers. 98.07 N and its dwell time was 8s, whereas the major load was 588.4 N with a dwell time of 15 s. The indentation was measured using a hand-held gauge. Each Rockwell hardness scale division represents a 0.002 mm vertical movement of the indenter. The Rockwell hardness number is derived from the following equation: where, HR = Rockwell number e = depth of impression after removal of the major load, in units of 0.002 mm.

Results from Dimensional Analysis
Dimensional changes in FFF-printed samples before and after annealing is a common phenomenon [31] due to the nature of this post-process. In FFF, thermoplastics are extruded out of a nozzle at their melting temperatures and are then cooled rapidly to form the desired shape. Being a poor conductor of heat, the extruded parts are cooled unevenly leading to internal stresses especially between layers. To relieve these stresses, annealing can be performed by heating the materials to their glass transition temperatures, but below their melting points. This will reorganize the internal crystalline structure and cause bigger grains to form, which will result in much stronger objects that are less prone to layer separation. In doing so, the dimensions could also be affected in any of the three axes (x, y, and z) through shrinkage or expansion. This is due to how the FFF-printed samples cool and how the tensions, stresses, and air bubbles get locked between the layers.   As evident from the dimensional changes, the materials were affected by the annealing conditions but to varying degrees. At 90 °C, PLA shrank by a maximum of 5.6% along the x-axis and 7% along the y-axis. On the other hand, it expanded by 5% at the same temperature along the z-axis. Cu-PLA was not affected significantly in the x and y axes, showing shrinkage of 1.3% and 0.9%, respectively, at 90 °C. However, its thickness was reduced by 8% at 90 °C, showing that annealing does not affect the dimensional tolerances of this material significantly. Like Cu-PLA, ABS also experienced minimum shrinkage along the x (2%) and y (0.9%) axes at 125 °C but it expanded by 4.25%. Al-ASA performed well until 90 °C with shrinkage of only 0.6% and 1% along x and y axes  As evident from the dimensional changes, the materials were affected by the annealing conditions but to varying degrees. At 90 °C, PLA shrank by a maximum of 5.6% along the x-axis and 7% along the y-axis. On the other hand, it expanded by 5% at the same temperature along the z-axis. Cu-PLA was not affected significantly in the x and y axes, showing shrinkage of 1.3% and 0.9%, respectively, at 90 °C. However, its thickness was reduced by 8% at 90 °C, showing that annealing does not affect the dimensional tolerances of this material significantly. Like Cu-PLA, ABS also experienced minimum shrinkage along the x (2%) and y (0.9%) axes at 125 °C but it expanded by 4.25%. Al-ASA performed well until 90 °C with shrinkage of only 0.6% and 1% along x and y axes As evident from the dimensional changes, the materials were affected by the annealing conditions but to varying degrees. At 90 • C, PLA shrank by a maximum of 5.6% along the x-axis and 7% along the y-axis. On the other hand, it expanded by 5% at the same temperature along the z-axis. Cu-PLA was not affected significantly in the x and y axes, showing shrinkage of 1.3% and 0.9%, respectively, at 90 • C. However, its thickness was reduced by 8% at 90 • C, showing that annealing does not affect the dimensional tolerances of this material significantly. Like Cu-PLA, ABS also experienced minimum shrinkage along the x (2%) and y (0.9%) axes at 125 • C but it expanded by 4.25%. Al-ASA performed well until 90 • C with shrinkage of only 0.6% and 1% along x and y axes respectively but deformed significantly as it was heated at higher temperatures (105 • C) as shown in Figure 8. This is the reason why the Al-ASA samples (annealed at 105 and 115 • C) were not subjected to tensile testing and should not be annealed at temperatures higher than 90 • C. This is a good estimation and can be used for future testing without adversely affecting the dimensional tolerances of FFF-printed Al-ASA products. Considering that 90 • C is a usable annealing temperature for Al-ASA material, it expanded by 2.5% along the z-axis. Al-ASA is considered to have high heat resistance and that is shown by the extremely small shrinkage/expansion values at 90 • C compared to other materials. However, higher temperatures can result in significant deformation that can render a product useless. The reason for such high shrinkage can be attributed to residual micro-stresses that arise from the inhomogeneity of properties in the microstructure. A common source of micro-stresses is the difference in the coefficients of thermal expansion in multiphase materials. Al-ASA is a combination of aluminium powder and ASA plastic, therefore, it is expected to develop stresses at elevated temperatures between the matrix and reinforcement as they contract at different rates. It is to be noted that different materials from different suppliers could result in varying percentages of shrinkage/expansion along the three axes. These values can be compensated in the original CAD design phase to ensure that the required dimensions are achieved after annealing.
J. Manuf. Mater. Process. 2020, 4, x FOR PEER REVIEW 10 of 21 respectively but deformed significantly as it was heated at higher temperatures (105 °C) as shown in Figure 8. This is the reason why the Al-ASA samples (annealed at 105 and 115 °C) were not subjected to tensile testing and should not be annealed at temperatures higher than 90 °C. This is a good estimation and can be used for future testing without adversely affecting the dimensional tolerances of FFF-printed Al-ASA products. Considering that 90 °C is a usable annealing temperature for Al-ASA material, it expanded by 2.5% along the z-axis. Al-ASA is considered to have high heat resistance and that is shown by the extremely small shrinkage/expansion values at 90 °C compared to other materials. However, higher temperatures can result in significant deformation that can render a product useless. The reason for such high shrinkage can be attributed to residual micro-stresses that arise from the inhomogeneity of properties in the microstructure. A common source of micro-stresses is the difference in the coefficients of thermal expansion in multiphase materials. Al-ASA is a combination of aluminium powder and ASA plastic, therefore, it is expected to develop stresses at elevated temperatures between the matrix and reinforcement as they contract at different rates. It is to be noted that different materials from different suppliers could result in varying percentages of shrinkage/expansion along the three axes. These values can be compensated in the original CAD design phase to ensure that the required dimensions are achieved after annealing.

Results from Ultrasonic Testing
All the materials used in this work are thermoplastics, i.e., they are polymer materials that become pliable or moldable at a certain elevated temperature and solidify upon cooling. These polymers comprise long and repetitive molecular chains, such molecular structures can be classified into semi-crystalline (partially organised structure) and amorphous (random structure). PLA and Cu-PLA fall under the former category whereas ABS and Al-ASA fall under the latter category. The reason for undertaking ultrasonic testing (UT) is to assess whether annealing can enhance the crystallinity of these polymers and subsequently form a smooth pathway for sound waves to travel. It is widely reported that annealing can increase crystallinity at elevated temperatures [14,15,32,33] and this test can provide insight into whether UT is a suitable option to assess the crystallinity of FFFprinted samples. Crystallinity has also been linked with the mechanical properties of FFF-printed PLA [16,18,34,35]. UT was conducted on all the materials and three measurements were taken based on the description given in Section 3.2. The time taken by the high-frequency acoustic sound waves to travel from one transducer through the test piece and to the second transducer was measured and the results for the different materials are shown in Figure 9.

Results from Ultrasonic Testing
All the materials used in this work are thermoplastics, i.e., they are polymer materials that become pliable or moldable at a certain elevated temperature and solidify upon cooling. These polymers comprise long and repetitive molecular chains, such molecular structures can be classified into semi-crystalline (partially organised structure) and amorphous (random structure). PLA and Cu-PLA fall under the former category whereas ABS and Al-ASA fall under the latter category. The reason for undertaking ultrasonic testing (UT) is to assess whether annealing can enhance the crystallinity of these polymers and subsequently form a smooth pathway for sound waves to travel. It is widely reported that annealing can increase crystallinity at elevated temperatures [14,15,32,33] and this test can provide insight into whether UT is a suitable option to assess the crystallinity of FFF-printed samples. Crystallinity has also been linked with the mechanical properties of FFF-printed PLA [16,18,34,35]. UT was conducted on all the materials and three measurements were taken based on the description given in Section 3.2. The time taken by the high-frequency acoustic sound waves to travel from one transducer through the test piece and to the second transducer was measured and the results for the different materials are shown in Figure 9. As evident from the UT results, the time taken for sound waves to travel in each material type differs. PLA is a semi-crystalline material and has resulted in an average transmission time of 2.2 μs. These results are consistent with our previous work [36]. However, a significant reduction was observed at 70 °C annealing time with 2.06 μs indicating that the crystallinity of PLA has increased [14], internal stresses and air gaps/voids (as discussed in Section 4.4) have been reduced [15] resulting in faster transmission time. This time went up for 80 and 90 °C (2.16 μs for both) but was still lower than the control PLA sample values. This indicates that annealing of PLA can enhance its crystallinity and reduce its internal stresses significantly [18,20]. Therefore, it is a useful post-process for the PLA material [37,38]. Cu-PLA material followed a similar trend to PLA with the 70 °C samples showing slightly lower average values (3.36 μs) compared to control samples (3.43 μs). These values continued to fluctuate with increasing temperatures but stayed lower than the control samples. This shows that the crystallinity of Cu-PLA samples is affected by annealing but not to a large extent. Since Cu-PLA is about 80% copper, it was expected that the sound waves would travel much faster but that was not the case. It can be attributed to the way PLA and copper particles are mixed together to form the filaments.
With the amorphous materials (ABS and Al-ASA), the results of UT were more like Cu-PLA than PLA. For ABS, the lowest transmission time was observed for the sample annealed at 115 °C (2.16 μs) compared to the control sample (2.23 μs) and then returned to the same values at 125 °C. This shows that annealing can help fill interlayer gaps and porosity for ABS samples but not to a large extent As evident from the UT results, the time taken for sound waves to travel in each material type differs. PLA is a semi-crystalline material and has resulted in an average transmission time of 2.2 µs. These results are consistent with our previous work [36]. However, a significant reduction was observed at 70 • C annealing time with 2.06 µs indicating that the crystallinity of PLA has increased [14], internal stresses and air gaps/voids (as discussed in Section 4.4) have been reduced [15] resulting in faster transmission time. This time went up for 80 and 90 • C (2.16 µs for both) but was still lower than the control PLA sample values. This indicates that annealing of PLA can enhance its crystallinity and reduce its internal stresses significantly [18,20]. Therefore, it is a useful post-process for the PLA material [37,38]. Cu-PLA material followed a similar trend to PLA with the 70 • C samples showing slightly lower average values (3.36 µs) compared to control samples (3.43 µs). These values continued to fluctuate with increasing temperatures but stayed lower than the control samples. This shows that the crystallinity of Cu-PLA samples is affected by annealing but not to a large extent. Since Cu-PLA is about 80% copper, it was expected that the sound waves would travel much faster but that was not the case. It can be attributed to the way PLA and copper particles are mixed together to form the filaments.
With the amorphous materials (ABS and Al-ASA), the results of UT were more like Cu-PLA than PLA. For ABS, the lowest transmission time was observed for the sample annealed at 115 • C (2.16 µs) compared to the control sample (2.23 µs) and then returned to the same values at 125 • C. This shows that annealing can help fill interlayer gaps and porosity for ABS samples but not to a large extent [19,39]. Al-ASA also shows a similar pattern to ABS, but the lowest value was obtained for the sample annealed at 70 • C (2.13 µs) compared to the control sample (2.33 µs) and then the values increased again at 90 • C (2.33 µs). The only positive thing here is that the reduction in transmission time is about 9.3%, which shows the effect of annealing in reducing air gaps/voids within the Al-ASA sample at 70 • C but not at other values. Al-ASA is a unique material as it can be annealed at lower temperatures compared to ABS. One downside of this material is its high operating temperatures (as shown in Table 1) as most low-cost FFF systems would not be able to properly print at such high temperatures. Overall, in terms of crystallinity and differences in transmission times, annealed PLA and Cu-PLA samples showed lower values compared to the control samples whereas the values for annealed ABS and Al-ASA fluctuated with some lower and equivalent values compared to control samples.

Results from Tensile Testing
Force vs. displacement curves for all the four materials are shown in Figure 10. It is to be noted that only the samples with the maximum load values have been plotted here. The fractured samples for all the four materials are shown in Figure 11. These samples were used for microscopic analysis in Section 4.3. Average load values with error bars for all the tested samples are shown in Figure 12 to ascertain the difference in test results. An increase in tensile strength of PLA material upon annealing is a widely reported phenomenon [14,17,18,20] and the same has happened here with the average fracture loads increasing with higher annealing temperatures. As shown in Section 4.2 for UT, lower transmission times were observed for annealed samples compared to the control PLA due to the annealed samples becoming more crystalline [15]. This has been supported by a significant increase in tensile strength as shown in Figure 12a. A similar trend of increase in tensile strength with increasing annealing temperatures was observed for Cu-PLA samples as well. The samples annealed at 90 • C show an increase of 21% in fracture load values as shown in Figure 12b. These are the same samples that showed lower transmission times before the tensile test and after annealing. The comparison between PLA and Cu-PLA also shows that the latter is inferior in terms of tensile strength compared to the former. Even though it has shown a substantial increase in load values after annealing, it is still incapable of competing with PLA in terms of strength. This is due to the inhomogeneity and the lack of cohesion between the metal and plastic particles at the microscopic level compared to the homogeneous structure of PLA. Regardless of the difference in the strength of the materials, these results demonstrate that UT is a viable option to ascertain the degree of crystallinity for semi-crystalline polymers.
On the other hand, ABS showed an increase of 3.3%, 6.75% and 2.8% at annealing temperatures of 105, 115 and 125 • C, respectively. This can also be supported by the UT, where the lowest transmission times were observed for the sample annealed at 115 • C, the ones showing the maximum increase in fracture load values in Figure 12c. Al-ASA shows an increase of 5.2%, 5.5% and 3% at temperatures of 70, 80 and 90 • C, respectively (Figure 12d). These results can also be supported by the UT but to a lesser extent compared to ABS. These results also show that UT can be used to assess the crystallinity of amorphous plastics. However, the results indicate quite a low increase in fracture load values after annealing, showing that ABS and ASA materials are not very receptive to annealing. Annealing is a complex process to begin with and requires a balance between time and temperature to achieve optimised results. In this context, a strong reason is needed to anneal these materials knowing the result would be a limited increase in mechanical properties.

Results from Microstructural Analysis
The fractured surfaces were cut to size to fit on the SEM platform and observed for the presence of voids, cracks and air gaps. First and foremost, all the tested samples exhibited a brittle failure, i.e., without showing any necking or plastic deformation before fracture. FFF is a process where layers of extruded materials are stacked on top of each other leading to the formation of voids or porosity [40]. These voids can be seen clearly in Figures 13-16. Voids represent empty areas with no connection and a higher percentage of voids could significantly degrade the mechanical properties [41]. PLA is a brittle material and its SEM micrographs at ×100 magnification are shown in Figure 13. Un-annealed PLA has some gaps due to the nature of the FFF process [42] as seen in Figure 13a, but these gaps reduce in 70 °C annealed samples, as shown in Figure 13b. These voids are still present at 80 (Figure 13c) and 90 °C (Figure 13d) annealed samples. However, they are very few and this is the reason for higher fracture load values (as discussed in Section 4.3) at elevated annealing temperatures for PLA. Furthermore, the low percentage of air gaps and voids is also the reason for the lower transmission times recorded for annealed samples compared to the control PLA (as discussed in Section 4.2). These results indicate a strong correlation among ultrasonic testing, tensile testing and microstructural analysis post-annealing for PLA materials manufactured by FFF.
Cu-PLA shows the clear presence of Cu particles in Figure 14a. At 70 °C, the distance among these particles has reduced and they have clustered together showing some voids in Figure 14b. With the increase in annealing temperature, these particles have amalgamated to a large extent and more voids have become visible (Figure 14c,d). This amalgamation is the reason for the higher fracture load values and lower transmission times of annealed Cu-PLA samples compared to the unannealed ones.

Results from Microstructural Analysis
The fractured surfaces were cut to size to fit on the SEM platform and observed for the presence of voids, cracks and air gaps. First and foremost, all the tested samples exhibited a brittle failure, i.e., without showing any necking or plastic deformation before fracture. FFF is a process where layers of extruded materials are stacked on top of each other leading to the formation of voids or porosity [40]. These voids can be seen clearly in Figures 13-16. Voids represent empty areas with no connection and a higher percentage of voids could significantly degrade the mechanical properties [41]. PLA is a brittle material and its SEM micrographs at ×100 magnification are shown in Figure 13. Un-annealed PLA has some gaps due to the nature of the FFF process [42] as seen in Figure 13a, but these gaps reduce in 70 • C annealed samples, as shown in Figure 13b. These voids are still present at 80 (Figure 13c) and 90 • C (Figure 13d) annealed samples. However, they are very few and this is the reason for higher fracture load values (as discussed in Section 4.3) at elevated annealing temperatures for PLA. Furthermore, the low percentage of air gaps and voids is also the reason for the lower transmission times recorded for annealed samples compared to the control PLA (as discussed in Section 4.2). These results indicate a strong correlation among ultrasonic testing, tensile testing and microstructural analysis post-annealing for PLA materials manufactured by FFF.
It is to be noted that the amalgamation should be controlled for better results. Both semi-crystalline materials, i.e., PLA and Cu-PLA, have shown consistent characteristics for all the experimental testing conducted in this work.   It is to be noted that the amalgamation should be controlled for better results. Both semi-crystalline materials, i.e., PLA and Cu-PLA, have shown consistent characteristics for all the experimental testing conducted in this work.   slightly in Figure 15b, but at the same time, the voids can be seen along a straight line in the sample as a result of the material being pulled during tensile testing. Figure 15c shows the smoothest weld joint characteristics at 115 °C; the temperature at which the maximum fracture load values and lowest transmission times were observed during tensile and ultrasonic testing, respectively. The pulled layers also became apparent in Figure 15d at 125 °C, showing a reduction in fracture load values and increased transmission times. The microstructural analysis of ABS has supported the results obtained through ultrasonic and tensile testing. As highlighted in Section 4.1, Al-ASA suffered significant deformation due to annealing at 105 °C. This material is heat resistant up to 93 °C and therefore should not be subjected to higher operating temperatures. In terms of the micrographs, the voids are quite visible in Figure 16. The sample annealed at 70 °C showed the maximum fracture load values and the lowest transmission times. Upon analysing Figure 16b and comparing it to Figure 16c,d, the size of the voids is smaller. However, the size of the voids continues to grow with elevated annealing temperatures. The sample annealed at 90 °C shows the largest voids and this is also the sample with the lowest fracture load values of all the annealed samples.

Results from Hardness Testing
Hardness measurements quantify the resistance of a material to plastic deformation and indentation hardness tests are used to determine material hardness. There are two main categories for indentation hardness tests, i.e., micro-indentation and macro-indentation tests. Micro-indentation tests typically have forces less than 2 N. The results obtained in this work are based on macroindentation as the applied minor and major loads were 98.07 N and 588.4 N respectively. As with any testing procedure, there are sources of errors that should be mitigated. These errors include but are not limited to poor technique, poor calibration of the equipment and the strain hardening effect of the process. Annealing can reduce surface roughness that can lead to higher hardness values [43,44]. It is to be noted that FFF parameters can also play a major role in determining the indentation Cu-PLA shows the clear presence of Cu particles in Figure 14a. At 70 • C, the distance among these particles has reduced and they have clustered together showing some voids in Figure 14b. With the increase in annealing temperature, these particles have amalgamated to a large extent and more voids have become visible (Figure 14c,d). This amalgamation is the reason for the higher fracture load values and lower transmission times of annealed Cu-PLA samples compared to the unannealed ones. It is to be noted that the amalgamation should be controlled for better results. Both semi-crystalline materials, i.e., PLA and Cu-PLA, have shown consistent characteristics for all the experimental testing conducted in this work. Figure 15 shows the SEM micrographs for the ABS material. It can be observed in Figure 15a that the weld joints of untreated ABS sample have uneven waviness. This may be due to the uneven cooling of the material that resulted in poor surface characteristics [19]. The unevenness has reduced slightly in Figure 15b, but at the same time, the voids can be seen along a straight line in the sample as a result of the material being pulled during tensile testing. Figure 15c shows the smoothest weld joint characteristics at 115 • C; the temperature at which the maximum fracture load values and lowest transmission times were observed during tensile and ultrasonic testing, respectively. The pulled layers also became apparent in Figure 15d at 125 • C, showing a reduction in fracture load values and increased transmission times. The microstructural analysis of ABS has supported the results obtained through ultrasonic and tensile testing.
As highlighted in Section 4.1, Al-ASA suffered significant deformation due to annealing at 105 • C. This material is heat resistant up to 93 • C and therefore should not be subjected to higher operating temperatures. In terms of the micrographs, the voids are quite visible in Figure 16. The sample annealed at 70 • C showed the maximum fracture load values and the lowest transmission times. Upon analysing Figure 16b and comparing it to Figure 16c,d, the size of the voids is smaller. However, the size of the voids continues to grow with elevated annealing temperatures. The sample annealed at 90 • C shows the largest voids and this is also the sample with the lowest fracture load values of all the annealed samples.

Results from Hardness Testing
Hardness measurements quantify the resistance of a material to plastic deformation and indentation hardness tests are used to determine material hardness. There are two main categories for indentation hardness tests, i.e., micro-indentation and macro-indentation tests. Micro-indentation tests typically have forces less than 2 N. The results obtained in this work are based on macro-indentation as the applied minor and major loads were 98.07 N and 588.4 N respectively. As with any testing procedure, there are sources of errors that should be mitigated. These errors include but are not limited to poor technique, poor calibration of the equipment and the strain hardening effect of the process. Annealing can reduce surface roughness that can lead to higher hardness values [43,44]. It is to be noted that FFF parameters can also play a major role in determining the indentation hardness, e.g., infill percentage, infill pattern and number of shells. An average of five values was taken from each sample that was equidistant from each other as shown in Figure 17, and the results of Rockwell hardness scale L are shown in Figure 18.

Results from Hardness Testing
Hardness measurements quantify the resistance of a material to plastic deformation and indentation hardness tests are used to determine material hardness. There are two main categories for indentation hardness tests, i.e., micro-indentation and macro-indentation tests. Micro-indentation tests typically have forces less than 2 N. The results obtained in this work are based on macroindentation as the applied minor and major loads were 98.07 N and 588.4 N respectively. As with any testing procedure, there are sources of errors that should be mitigated. These errors include but are not limited to poor technique, poor calibration of the equipment and the strain hardening effect of the process. Annealing can reduce surface roughness that can lead to higher hardness values [43,44]. It is to be noted that FFF parameters can also play a major role in determining the indentation hardness, e.g., infill percentage, infill pattern and number of shells. An average of five values was taken from each sample that was equidistant from each other as shown in Figure 17, and the results of Rockwell hardness scale L are shown in Figure 18.   As evident from the hardness test results, all the materials experienced an increase in hardness numbers with elevated temperatures. This is because annealing has reduced the internal stresses and the weld joins for the FFF samples have become more uniform. According to the results, Al-ASA samples have the highest hardness values with PLA as second, Cu-PLA as third and ABS with the lowest values. The difference between the maximum values of Cu-PLA (41.2 HRL) and ABS (39.1 HRL) is not significant, but ABS achieved these values at 125 °C whereas Cu-PLA was annealed to a maximum of 90 °C in this work. Hard materials are often also brittle. This can be seen in Figures 9  and 10 where only ABS offered a small degree of ductility as evident from the slight discoloration near the fractured area. Furthermore, the difference in measurements can be observed for all the materials. This can be attributed to the indenter not being able to compress a uniformly packed surface. Hardness values are greatly affected by surface roughness. FFF-printed samples are usually rougher than other types of plastic parts because of the layer by layer operation and these layers are visible on the surface of the samples after the print [45]. Even though annealing has reduced the surface roughness to an extent, two common situations may arise, i.e., the indenter meeting a peak or a valley. In the former, the non-uniform contact will increase the localized stress at the contact point and will deform the material to a greater depth at relatively low loads. This will result in greater penetration depth and lower calculated hardness value. In the case of the latter, when the indenter meets a valley the true contact area will be underestimated and consequently, the calculated hardness will be overestimated [46].

Conclusions
In this study, two categories of polymers were investigated, i.e., semi-crystalline and amorphous. Commonly available FFF materials in PLA and ABS along with metal-infused materials As evident from the hardness test results, all the materials experienced an increase in hardness numbers with elevated temperatures. This is because annealing has reduced the internal stresses and the weld joins for the FFF samples have become more uniform. According to the results, Al-ASA samples have the highest hardness values with PLA as second, Cu-PLA as third and ABS with the lowest values. The difference between the maximum values of Cu-PLA (41.2 HRL) and ABS (39.1 HRL) is not significant, but ABS achieved these values at 125 • C whereas Cu-PLA was annealed to a maximum of 90 • C in this work. Hard materials are often also brittle. This can be seen in Figures 9 and 10 where only ABS offered a small degree of ductility as evident from the slight discoloration near the fractured area. Furthermore, the difference in measurements can be observed for all the materials. This can be attributed to the indenter not being able to compress a uniformly packed surface. Hardness values are greatly affected by surface roughness. FFF-printed samples are usually rougher than other types of plastic parts because of the layer by layer operation and these layers are visible on the surface of the samples after the print [45]. Even though annealing has reduced the surface roughness to an extent, two common situations may arise, i.e., the indenter meeting a peak or a valley. In the former, the non-uniform contact will increase the localized stress at the contact point and will deform the material to a greater depth at relatively low loads. This will result in greater penetration depth and lower calculated hardness value. In the case of the latter, when the indenter meets a valley the true contact area will be underestimated and consequently, the calculated hardness will be overestimated [46].