Mechanical and Thermal Properties of 3D Printed Polycarbonate

: This study aims at showing the potential of additive manufacturing as a new processing route for designing future insulators in the building sector. Polycarbonate (PC) is studied as a possible candidate for designing these new insulators. This polymer offers several advantages, among them ﬁre resistance and stability of its physical properties at high temperatures. The 3D printing of PC is attempted using fused deposition modelling technology. The printing temperature and inﬁll rate are varied to achieve optimal mechanical and thermal characteristics. The results show that an optimal printing temperature of 280 ◦ C is needed to achieve high tensile performance. In addition, thermal properties including thermal conductivity and effusivity increase with the increase of the inﬁll rate in opposition to the thermal diffusivity decrease.


Introduction
Poorly insulated buildings are known to be a major source of huge energy consumption including heating and cooling [1]. Choosing adequate insulation has thus been made mandatory in recent legislation. For instance, The Energy Performance of Buildings Directive (EPBD) set by the European Union is meant to improve the energy performance of buildings in Europe [2]. Intensive work has been carried out recently in order to develop new and more efficient and environmentally friendly materials to be used in the field of insulation [3,4].
An interesting method of improving the insolation capabilities of a building is to design new insulators with local control of material architecture allowing, for instance, to adapt the shape, connectivity and local content of porosities [5][6][7][8]. Finite element computations are well suited for exploring new material designs [9]. The design of insulators with complex material architecture is, however, difficult to address by conventional manufacturing processes such as moulding or foaming. With the recent development of additive manufacturing (AM) technology, there is a strong tendency to rethink the material design based on required part functionality using a layer-by-layer deposition approach. This can be a game-changer for future building insulation. Indeed, AM is a promising new solution in the sustainable building sector, more particularly for developing biosourced polymeric solutions for the new generation of insulators. With a combination of computer-based control and a local deposition process, AM responds optimally to insulation requirements in the context of energy-saving policies being adopted worldwide.
However, there are challenges that face AM in the building sector. The potential of the printed parts at a large scale would be undermined by the large time needed to reach production at the industrial level. In order to ensure the viability of AM technology on a large scale, several successful attempts have been made both at the laboratory scale and at the industrial level. Because the printing speed remains a major limitation of 3D printing processes [10], several routes have been explored recently to overcome this issue, among which is the design of innovative deposition strategies such as nozzles that allow large filament deposition rates [11] and multi-nozzle systems [12]. A realistic strategy that is currently tested is to obtain a balance between resolution and part size. For prints on a centimetre scale, it makes sense to use nozzles as small as 0.4 mm. For large realizations as in the case of the building sector, nozzles that are used for instance in crafting are on the centimetre scale [13].
FDM (Fused Deposition Modelling) is one of the most widespread AM technologies [14,15]. FDM combines simplicity and cost-effectiveness for processing polymeric structures without further need for extra-tooling or post-processing of manufactured parts [16]. A large amount of work was carried out in recent decades to understand the mechanical and thermal properties of printed parts and their dependency on printing parameters such as the printing temperature and the slicing conditions [17][18][19][20].
The present study explores the use of FDM to achieve new insulators from polycarbonate (PC). This material is generally known to be a tough, transparent plastic material with outstanding strength, stiffness, and impact. PC maintains its rigidity up to 140 • C and toughness down to −20 • C. This material is also reputed to be a slow-burning polymer. It passes, for instance, several severe flammability tests when combined with a special flame retardant. In this study, optimal printing conditions are explored for this material with a focus on two main leverages: printing temperature ranging from 270 • C to 300 • C, and infill rate from 10% up to 100%. For these conditions, both the mechanical and thermal properties are determined and compared to standard insulation materials used in the building sector. Figure 1 shows the general scheme adopted in this study to achieve results on the mechanical and thermal properties of PC. However, there are challenges that face AM in the building sector. The potential of the printed parts at a large scale would be undermined by the large time needed to reach production at the industrial level. In order to ensure the viability of AM technology on a large scale, several successful attempts have been made both at the laboratory scale and at the industrial level. Because the printing speed remains a major limitation of 3D printing processes [10], several routes have been explored recently to overcome this issue, among which is the design of innovative deposition strategies such as nozzles that allow large filament deposition rates [11] and multi-nozzle systems [12]. A realistic strategy that is currently tested is to obtain a balance between resolution and part size. For prints on a centimetre scale, it makes sense to use nozzles as small as 0.4 mm. For large realizations as in the case of the building sector, nozzles that are used for instance in crafting are on the centimetre scale [13].
FDM (Fused Deposition Modelling) is one of the most widespread AM technologies [14,15]. FDM combines simplicity and cost-effectiveness for processing polymeric structures without further need for extra-tooling or post-processing of manufactured parts [16]. A large amount of work was carried out in recent decades to understand the mechanical and thermal properties of printed parts and their dependency on printing parameters such as the printing temperature and the slicing conditions [17][18][19][20].
The present study explores the use of FDM to achieve new insulators from polycarbonate (PC). This material is generally known to be a tough, transparent plastic material with outstanding strength, stiffness, and impact. PC maintains its rigidity up to 140 °C and toughness down to −20 °C. This material is also reputed to be a slow-burning polymer. It passes, for instance, several severe flammability tests when combined with a special flame retardant. In this study, optimal printing conditions are explored for this material with a focus on two main leverages: printing temperature ranging from 270 °C to 300 °C, and infill rate from 10% up to 100%. For these conditions, both the mechanical and thermal properties are determined and compared to standard insulation materials used in the building sector. Figure 1 shows the general scheme adopted in this study to achieve results on the mechanical and thermal properties of PC.

Experimental Layout
A polycarbonate (PC) filament (diameter = 1.75 mm) from Nanovia company (Louargat, France) is used as a feedstock material for FDM. The main characteristics and printing conditions are given in Table 1. The FDM process is performed using the Volumic stream 30 Ultra printer from Volumic 3D [21] (Nice, France) with a printer volume of 290 × 200 × 300 mm, and an extruder reaching 420 • C and a building platform heating up to 150 • C. The fixed printing conditions are described in Table 2. For mechanical testing of filament intrinsic properties, the asreceived filament of a typical length of 80 mm was used with a gauge length of 40 mm ( Table 2). Specific shapes were adopted to perform mechanical and thermal analyses which obey the standards. For mechanical testing, gthe eometry of the specimens was adapted according to the ISO 527-1/-2 and ASTM D638 sample type I standards [22]. These are dogbone shapes used for tensile testing of 3D printed PC samples. The overall length of the sample was adapted to the FDM printing process to avoid adhesion problems to the printing platform noticed on larger samples. In addition, parallelepipedic samples were used for thermal conductivity with 2 infill conditions ( Table 2). As PC 3D printing requires particular heating conditions, both building platform T B and printing T P temperatures were adjusted to obtain successful prints without warpage or platform decohesion, which were the most challenging issues for high Tg materials such as PC. Preliminary tests demonstrated that T P of 260 • C does not allow successful prints although this is recommended by the supplier ( Table 1). Illustrations of the sample geometries and layups (−45 • /+45 • ) are shown in Figure 2. were the most challenging issues for high Tg materials such as PC. Preliminary tests demonstrated that TP of 260 °C does not allow successful prints although this is recommended by the supplier ( Table 1). Illustrations of the sample geometries and layups (−45°/+45°) are shown in Figure 2.  The selected layup corresponded to a printing angle (θ) of 0° following the definition reported in [23]. For all 3D printed PC structures, the building direction is aligned with the sample thickness.
Tensile testing experiments were performed using a universal testing machine from Zwick Roell Group (Ulm, Germany) [24]. The machine was equipped with a load cell of 10 kN (Figure 3). The selected layup corresponded to a printing angle (θ) of 0 • following the definition reported in [23]. For all 3D printed PC structures, the building direction is aligned with the sample thickness. Tensile testing experiments were performed using a universal testing machine from Zwick Roell Group (Ulm, Germany) [24]. The machine was equipped with a load cell of 10 kN (Figure 3). All tensile experiments were performed using a displacement rate of 5 mm/min up to the sample failure. From the reaction force (F)-cross-head displacement (d) response, the engineering stress (σ)-strain (ε) curve was built according to the following conversion where w, t, and are the width, thickness, and initial length of the specimen at the central part of the specimen, respectively.
Engineering constants were derived from the engineering stress-strain response, namely Young's modulus , yield stress , tensile strength , ultimate stress , and elongation at break .
The Young's modulus is derived from the initial slope of the stress-strain curve ⁄ where and refer to the engineering stress and strain in the elasticity stage according to Hooke's law.
In the testing setup, a high-speed camera was used to monitor the sample deformation during sample extension. The optical observations were performed using a Phantom V7.3 camera from the Photonline company (Marly Le Roi, 78-France) [22,25]. Both low rate and high rate image recordings were used to observe the entire deformation sequence and achieve higher time resolutions at the rupture point. The frame rate was adjusted accordingly between 100 and 5000 fps (frames per second), and the region of interest (ROI) area was modified between the full frame size (800 × 600 pixels) down to (112 × 320 pixels), with a typical pixel size of 130 μm.
Thermal properties of 3D printed PC samples were assessed using the technique of transient hot probe using the equipment FP2C provided by Neotim company (Albi, France) (Figure 4) according to the ASTM D5930-97 norm [26]. All tensile experiments were performed using a displacement rate of 5 mm/min up to the sample failure. From the reaction force (F)-cross-head displacement (d) response, the engineering stress (σ)-strain (ε) curve was built according to the following conversion and where w, t, and L 0 are the width, thickness, and initial length of the specimen at the central part of the specimen, respectively. Engineering constants were derived from the engineering stress-strain response, namely Young's modulus E Y , yield stress σ Y , tensile strength σ S , ultimate stress σ R , and elongation at break ε R .
The Young's modulus is derived from the initial slope of the stress-strain curve where σ e and ε e refer to the engineering stress and strain in the elasticity stage according to Hooke's law.
In the testing setup, a high-speed camera was used to monitor the sample deformation during sample extension. The optical observations were performed using a Phantom V7.3 camera from the Photonline company (Marly Le Roi, 78-France) [22,25]. Both low rate and high rate image recordings were used to observe the entire deformation sequence and achieve higher time resolutions at the rupture point. The frame rate was adjusted accordingly between 100 and 5000 fps (frames per second), and the region of interest (ROI) area was modified between the full frame size (800 × 600 pixels) down to (112 × 320 pixels), with a typical pixel size of 130 µm.
Thermal properties of 3D printed PC samples were assessed using the technique of transient hot probe using the equipment FP2C provided by Neotim company (Albi, France) ( Figure 4) according to the ASTM D5930-97 norm [26].  The equipment was used to measure thermal conductivity, effusivity and diffusivity of PC printed samples from the time evolution of the temperature. The principle consists of local heating of the material using a probe and measuring the temperature increase over time. Thermal properties were assessed by processing the heat signal. Three probes were used ( Figure 4). The thermal conductivity was measured using a hot wire probe consisting of a resistive wire and a thermocouple in an insulating Kapton support. The heating power used was 0.14 W. The thermal conductivity was estimated by the equation below: where Φ is the linear heat flux, L is the length of the hot wire and α is the slope of the linear part of the thermogram T(ln(t)). Concerning the thermal effusivity, a hot plane probe was used. The heating power was adjusted to 2.5 W. As for the thermal diffusivity, a hot ring probe was used. The heating time was adjusted to 10 s and the duration of the measurement was, respectively, 60 s and 70 s for PC samples of 10% and 30% of infill rate. The heating power was 3.01 W.
The thermal diffusivity is given by the following expression according to [27] r 6 t t t t log t t t where tm is the time corresponding to the maximum of the temperature by the data of the ring source and r is the radius of the ring. Two samples of (7.5 × 10 × 2) cm were required for each measurement. Figure 4 illustrates the probe geometries used for the measurements.
The thermal conductivity was measured within the range 0.02 à 5 W•m −1 •K −1 . The thermal effusivity was measured in the range (20-10,000 J•m −2 •K −1 •s −1/2 within a temperature interval of −60 °C to 100 °C. The thermal diffusivity measurement range was between 0.1 and 4 mm 2 •s −1 . For all thermal properties, the temperature interval used was between 60 °C and 100 °C.  The equipment was used to measure thermal conductivity, effusivity and diffusivity of PC printed samples from the time evolution of the temperature. The principle consists of local heating of the material using a probe and measuring the temperature increase over time. Thermal properties were assessed by processing the heat signal. Three probes were used ( Figure 4). The thermal conductivity was measured using a hot wire probe consisting of a resistive wire and a thermocouple in an insulating Kapton support. The heating power used was 0.14 W. The thermal conductivity was estimated by the equation below:

Results and Discussion
where Φ is the linear heat flux, L is the length of the hot wire and α is the slope of the linear part of the thermogram T(ln(t)). Concerning the thermal effusivity, a hot plane probe was used. The heating power was adjusted to 2.5 W. As for the thermal diffusivity, a hot ring probe was used. The heating time was adjusted to 10 s and the duration of the measurement was, respectively, 60 s and 70 s for PC samples of 10% and 30% of infill rate. The heating power was 3.01 W.
The thermal diffusivity is given by the following expression according to [27] Di f f = r 2 6 t 0 where t m is the time corresponding to the maximum of the temperature by the data of the ring source and r is the radius of the ring. Two samples of (7.5 × 10 × 2) cm were required for each measurement. Figure 4 illustrates the probe geometries used for the measurements.
The thermal conductivity was measured within the range 0.02 à 5 W·m −1 ·K −1 . The thermal effusivity was measured in the range (20-10,000 J·m −2 ·K −1 ·s −1/2 within a temperature interval of −60 • C to 100 • C. The thermal diffusivity measurement range was between 0.1 and 4 mm 2 ·s −1 . For all thermal properties, the temperature interval used was between 60 • C and 100 • C. Figure 5 compares the tensile response of four as-received PC filaments. The filament exhibits an elastic-plastic behaviour with a large necking behaviour revealed by optical recording in Figure 6.

Results and Discussion
Most of the measured engineering quantities (Table 3) are obtained with a small scatter below 6% with the exception of Young's modulus (20%) and elongation at break (16%). Furthermore, the achieved values for Young's modulus and elongation at break are consistent with the data provided by the supplier. The scatter represents only 3% and 2%, respectively. The filament exhibits an elastic-plastic behaviour with a large necking behaviour revealed by optical recording in Figure 6. Most of the measured engineering quantities (Table 3) are obtained with a small scatter below 6% with the exception of Young's modulus (20%) and elongation at break (16%). Furthermore, the achieved values for Young's modulus and elongation at break are consistent with the data provided by the supplier. The scatter represents only 3% and 2%, respectively. The filament exhibits an elastic-plastic behaviour with a large necking behaviour revealed by optical recording in Figure 6. Most of the measured engineering quantities (Table 3) are obtained with a small scatter below 6% with the exception of Young's modulus (20%) and elongation at break (16%). Furthermore, the achieved values for Young's modulus and elongation at break are consistent with the data provided by the supplier. The scatter represents only 3% and 2%, respectively.  Table 3. Engineering constants for the as-received PC filament and printed specimens as a function of printing temperature. Typical stress-strain response of 3D printed PC is shown in Figure 7 as a function of printing temperature (T p ). 110 639 ± 10 35.5 ± 0.29 42.9 ± 0.49 39.0 ± 0.77 0.08 ± 0.003 290 110 568 ± 15 28.0 ± 0.96 33.9 ± 0.92 31.9 ± 2.61 0.07 ± 0.003 300 110 567 ± 25 31.6 ± 0.60 35.5 ± 1.58 34.8 ± 1.79 0.07 ± 0.003 Typical stress-strain response of 3D printed PC is shown in Figure 7 as a function of printing temperature (Tp). A straightforward ranking of the tensile response with the printing temperature TP is not obtained. The optimal Tp of 280 °C is obtained allowing the 3D printed PC to achieve the highest stiffness and strength. All curves exhibit elastic-plastic behaviour with a minor stretching between strength and ultimate stress. The limited plasticity contrasts with the results obtained for the filament tensile response ( Figure 5). Within the considered range of TP, Table 3 summarises the obtained results for all engineering constants including Young's modulus , yield stress , tensile strength , ultimate stress , and elongation at break . A large reduction in the elongation at break is experienced when printing the PC with the layups considered. This reduction represents, on average, 71% of the elongation potential of PC. Surprisingly, the loss of stiffness is minor compared to other polymers printed using FDM [28]. It represents less than 10%.
For the remaining engineering quantities, namely yield stress, tensile strength, and ultimate stress, the amount of mechanical loss is within the range of variation reported for FDM, with an average scatter of 17%, 28%, and 24%, respectively. Figure 8 illustrates the optical recording of tensile response of 3D printed PC structures as a function of printing temperature TP. With the layups of −45°/+45° oriented within the plane of construction, the stretching of the filament arrangement produces a significant deviation with respect to the opening mode at low printing temperatures below 290 °C. A straightforward ranking of the tensile response with the printing temperature T P is not obtained. The optimal T p of 280 • C is obtained allowing the 3D printed PC to achieve the highest stiffness and strength. All curves exhibit elastic-plastic behaviour with a minor stretching between strength and ultimate stress. The limited plasticity contrasts with the results obtained for the filament tensile response ( Figure 5). Within the considered range of T P , Table 3 summarises the obtained results for all engineering constants including Young's modulus E Y , yield stress σ Y , tensile strength σ S , ultimate stress σ R , and elongation at break ε R . A large reduction in the elongation at break is experienced when printing the PC with the layups considered. This reduction represents, on average, 71% of the elongation potential of PC. Surprisingly, the loss of stiffness is minor compared to other polymers printed using FDM [28]. It represents less than 10%.
For the remaining engineering quantities, namely yield stress, tensile strength, and ultimate stress, the amount of mechanical loss is within the range of variation reported for FDM, with an average scatter of 17%, 28%, and 24%, respectively. Figure 8 illustrates the optical recording of tensile response of 3D printed PC structures as a function of printing temperature T P . With the layups of −45 • /+45 • oriented within the plane of construction, the stretching of the filament arrangement produces a significant deviation with respect to the opening mode at low printing temperatures below 290 • C.
Prior to crack initiation, a change in the grey level of the sample indicates damage initiation, which appears more diffuse with the central area of the specimen.
At higher printing temperatures, the compactness of the filament arrangements restores the opening mode, and unstable cracking takes place leading to a transverse rupture as shown in Figure 8c,d. This tendency toward unstable cracking corresponds to a decrease in the elongation at break (Table 4).
Thermal analysis of 3D printed PC is conducted at the optimal printing temperature (T P = 280 • C) for which the highest tensile ranking is obtained. Under this condition, two main infill rates are considered as 10% and 30%. Thermal properties are summarized in Table 4, where five replicates are used for each printing condition. High confidence in the results is achieved for all measured quantities. The average scatters for thermal conductivity, effusivity and diffusivity are 1.8%, 3.0%, and 2.4%, respectively. Prior to crack initiation, a change in the grey level of the sample indicates damage initiation, which appears more diffuse with the central area of the specimen.
At higher printing temperatures, the compactness of the filament arrangements restores the opening mode, and unstable cracking takes place leading to a transverse rupture as shown in Figure 8c,d. This tendency toward unstable cracking corresponds to a decrease in the elongation at break (Table 4). Thermal analysis of 3D printed PC is conducted at the optimal printing temperature (TP = 280 °C) for which the highest tensile ranking is obtained. Under this condition, two main infill rates are considered as 10% and 30%. Thermal properties are summarized in Table 4, where five replicates are used for each printing condition. High confidence in the results is achieved for all measured quantities. The average scatters for thermal conductivity, effusivity and diffusivity are 1.8%, 3.0%, and 2.4%, respectively.
The 3D printed PC material exhibits promising thermal properties with low thermal conductivity and diffusivity and high effusivity. Thermal conductivity and effusivity tend to increase with the infill rate, whereas the thermal diffusivity decreases. The amount of variation for the three properties is 14%, 40%, and 17%, respectively. The 3D printed PC  The 3D printed PC material exhibits promising thermal properties with low thermal conductivity and diffusivity and high effusivity. Thermal conductivity and effusivity tend to increase with the infill rate, whereas the thermal diffusivity decreases. The amount of variation for the three properties is 14%, 40%, and 17%, respectively. The 3D printed PC exhibits better thermal properties compared to hemp concrete, for which the thermal conductivity is close to 100 mW/(m·K) [29]. In addition, the effusivity and diffusivity of PC are quite similar to those of hemp concrete which are reported to be between 197 and 256 W·s 0.5 /(m 2 ·K) [30], and between 2 × 10 −7 and 6 × 10 −7 m 2 /s [31,32], respectively. These properties allow the 3D printed PC to achieve good thermal inertia and therefore make it a good insulation material for buildings.

Conclusions
This study concludes on the high potential of PC material as a candidate for the new generation of insulators in the building sector. Despite the unclear trend of the mechanical performance of 3D printed PC, an optimal printing temperature of 280 • C is obtained. Thermal analysis of 3D printed PC shows that the infill rate has a logical effect on the Energies 2022, 15, 3686 10 of 11 thermal conductivity, diffusivity and effusivity, which can be related to the amount of insulating air inside the printed PC. The measured properties demonstrate that PC material is a good candidate for thermal energy storage.