Melt Spinning of Highly Stretchable, Electrically Conductive Filament Yarns

Electrically conductive fibers are required for various applications in modern textile technology, e.g., the manufacturing of smart textiles and fiber composite systems with textile-based sensor and actuator systems. According to the state of the art, fine copper wires, carbon rovings, or metallized filament yarns, which offer very good electrical conductivity but low mechanical elongation capabilities, are primarily used for this purpose. However, for applications requiring highly flexible textile structures, as, for example, in the case of wearable smart textiles and fiber elastomer composites, the development of electrically conductive, elastic yarns is of great importance. Therefore, highly stretchable thermoplastic polyurethane (TPU) was compounded with electrically conductive carbon nanotubes (CNTs) and subsequently melt spun. The melt spinning technology had to be modified for the processing of highly viscous TPU–CNT compounds with fill levels of up to 6 wt.% CNT. The optimal configuration was achieved at a CNT content of 5 wt.%, providing an electrical resistance of 110 Ωcm and an elongation at break of 400%.


Introduction
For numerous applications, e.g., in the field of smart textiles and textile sensor and actuator technology, electrically conductive fibers and filaments are of great importance. They are essential for the production of textile-processable sensors [1][2][3] and sensor networks [4] as well as for the transmission of information detected in the device. New developments in the smart textiles sector are inconceivable without electrically conductive fibers. For instance, they can transmit the data collected during wound monitoring [5,6] or mechanical structural health monitoring of critical components [7][8][9]. Furthermore, they are essential for the development of novel, wearable devices [10][11][12] and the storage of electrical energy [13].
In terms of actuator technology, the supplied electrical energy can be used to generate mechanical deformation. In shape memory alloys (SMA), the applied electrical energy combined with the intrinsic resistance of SMA causes temperature to increase, which in turn leads to a conversion in the crystal structure from martensite to austenite, thus generating large usable forces and strains [14][15][16][17]. In contrast, shape memory polymers do not have intrinsically conductive components, but they too are able to use electrical energy via the intermediate stage of thermal energy to perform mechanical work [18][19][20]. For this purpose, the entire component can either be exposed to an electric field or a constant temperature, or individual areas can be targeted separately. Electrically conductive yarns are particularly suited for the activation of individual textile parts.
The soft segments have a glass transition point, which is below the usage temperature so that the molecules can be shifted flexibly due to low intermolecular interactions. Moreover, the soft segments exhibit an entropy-elastic behavior; i.e., the polymer chains that are entangled in a stress-free state are stretched under mechanical load with a decrease in entropy [39,40]. If the mechanical load is removed, the soft segments return to their energetically favorable initial state [41]. Thus, the soft segments cause the high elasticity of the polymer, whereas the hard segments determine the solid aggregate state at use temperature as well as the mechanical strength and stiffness [41]. The semi-crystalline hard segments in the TPU assume the function of covalent bonds (network points) in a conventional elastomer, hence preventing the polymer chains from gliding off against each other [37]. When TPU is heated, the intermolecular bonds between the hard segments are broken and the polymer becomes liquid so that it can be melt-spun. Thus, the structure of hard and soft segments enables thermal processing that is unsuitable for standard elastomers since they do not melt when heated but undergo decomposition processes.
In this study, the thermal processability of TPU is exploited to produce highly stretchable and electrically conductive multifilament yarns that can be used for a variety of tasks in the field of smart textiles and fiber elastomer composites. For this purpose, TPU is compounded with CNTs and melt spun. To enable a melt spinning process with this polymer material, which has a very high viscosity, a process modification is necessary to enable a particularly gentle drawing process. At filling levels of 5 wt.% CNT, electrical resistances of 110 Ωcm can be realized in the mechanically unloaded state. Even under relative mechanical strains of up to 100%, the electrical conductivity is maintained, but the electrical resistance increases by up to one order of magnitude.

Materials and Methods
For the melt spinning trials, the TPU grade Desmopan 9370A from Covestro AG (Leverkusen, Germany) [42] and TPU 1001 from Nanocyl SA (Sambreville, Belgium) [43] were used. TPU 1001 from Nanocyl SA is a masterbatch containing 10 wt.% CNT and 90 wt.% TPU. Before the spinning process was started, the materials were pre-compounded by hand to compounds of 1-6 wt.% CNT. The compounds were dried at a temperature of 80 • C for 24 h.
The tests were carried out on a bicomponent melt spinning plant of Dienes Apparatebau GmbH (Mühlheim am Main, Germany), at ITM, TU Dresden. This plant is equipped with a single-screw extruder, a twin-screw, and several spinning packages to realize different fiber geometries. The following tests were performed with a twin-screw extruder and a 60-filament core-sheath spinning die, although the extruder supplying the sheath component was not taken into operation. The 60-filament die has diameters of 0.6 mm. Each spinning process was performed with particularly coarse-meshed polymer filters and under a nitrogen atmosphere to avoid the oxidation of TPU. A spinning temperature of 180 • C was selected, and the winding speeds were varied between 8 and 650 m/min according to the compound's spinnability.
Extensive modifications to the spinning machine were required to ensure process stability. By means of an additional device, the weight of the solidified filaments was supported shortly below the spinneret so that the melt no longer had to support the entire weight of the filaments. For this purpose, a duo of godets driven by an electric motor was inserted into the spinning shaft 1 m below the spinneret (see Figure 1). Firstly, the spun filaments were guided over the lower cylinder, and, secondly, they ran vertically upwards while being drawn by the upper cylinder. Due to the staggered arrangement of the cylinders, further deflection points could be avoided to minimize potential effects on the yarn path, the geometry of individual filaments, and the arrangement of filaments in the fiber bundle. Once the spun filaments passed this additional device, they were taken off and wound up. Polymers 2021, 13, x FOR PEER REVIEW To determine the melt viscosity, rheometric measurements were performe Haake RheoWin /Thermo Scientific Mars II from Thermo Fisher Scientific Inc. (W MA, USA). The measurements were carried out at a constant temperature of 200 fineness was determined in accordance with DIN EN ISO 2060. For this purpose, ples with a defined length of 1 m each were taken from each spinning specificati mass of the samples was then determined using a precision scale R200D from Sa (Göttingen, Germany). The tensile tests were performed on a Zwicki Junior from Roell GmbH & Co. KG (Ulm, Germany) with a clamping length of 62.5 mm and a speed of 200 mm/min. Tensile testing as well as the determination of fineness we pleted for 5 samples each.
A four wire method was employed for resistance measurements on filament s with a length of 50 mm (see Figure 2). Additionally, a current source Voltcraft LR (Wollerau, Switzerland) and two Keithley DAQ6510-7700 multimeters from Keith struments Corp. (Solon, OH, USA) were used. The current source supplied a ma current of 100 mA. For each sample, four different current values were set at the source, and the multimeters were used to measure current and voltage at the c sample. Thus, four resistance values could be calculated and averaged for each based on the quotients of voltage and current. Of each spinning specification, 7 s were tested.  A four wire method was employed for resistance measurements on filament sections with a length of 50 mm (see Figure 2). Additionally, a current source Voltcraft LRP-1601 (Wollerau, Switzerland) and two Keithley DAQ6510-7700 multimeters from Keithley Instruments Corp. (Solon, OH, USA) were used. The current source supplied a maximum current of 100 mA. For each sample, four different current values were set at the current source, and the multimeters were used to measure current and voltage at the clamped sample. Thus, four resistance values could be calculated and averaged for each sample based on the quotients of voltage and current. Of each spinning specification, 7 samples were tested. To determine the melt viscosity, rheometric measurements were performed on a Haake RheoWin /Thermo Scientific Mars II from Thermo Fisher Scientific Inc. (Waltham, MA, USA). The measurements were carried out at a constant temperature of 200 °C. The fineness was determined in accordance with DIN EN ISO 2060. For this purpose, 5 samples with a defined length of 1 m each were taken from each spinning specification. The mass of the samples was then determined using a precision scale R200D from Sartorius (Göttingen, Germany). The tensile tests were performed on a Zwicki Junior from Zwick-Roell GmbH & Co. KG (Ulm, Germany) with a clamping length of 62.5 mm and a testing speed of 200 mm/min. Tensile testing as well as the determination of fineness were completed for 5 samples each.
A four wire method was employed for resistance measurements on filament sections with a length of 50 mm (see Figure 2). Additionally, a current source Voltcraft LRP-1601 (Wollerau, Switzerland) and two Keithley DAQ6510-7700 multimeters from Keithley Instruments Corp. (Solon, OH, USA) were used. The current source supplied a maximum current of 100 mA. For each sample, four different current values were set at the current source, and the multimeters were used to measure current and voltage at the clamped sample. Thus, four resistance values could be calculated and averaged for each sample based on the quotients of voltage and current. Of each spinning specification, 7 samples were tested.

Spinning and Stretching Process
The addition of CNTs to TPU led to significant inhomogeneities at the nano level, thus reducing spinnability. Pure TPU showed a slightly shear-thinning material behavior, which was significantly increased by the addition of CNTs. However, increasing the CNT content also resulted in a considerable rise in melt viscosity (see Figure 3) and decrease in stretchability. For example, once the CNT content was increased from 1 wt.% CNT to 5 wt.% CNT, a tenfold increase in viscosity was observed at shear rates of 1-10 s −1 , which are particularly relevant for the melt spinning process.

Spinning and Stretching Process
The addition of CNTs to TPU led to significant inhomogeneities at the nano level, thus reducing spinnability. Pure TPU showed a slightly shear-thinning material behavior, which was significantly increased by the addition of CNTs. However, increasing the CNT content also resulted in a considerable rise in melt viscosity (see Figure 3) and decrease in stretchability. For example, once the CNT content was increased from 1 wt.% CNT to 5 wt.% CNT, a tenfold increase in viscosity was observed at shear rates of 1-10 s −1 , which are particularly relevant for the melt spinning process. As a result, these filaments must be pulled off at a significantly slower pace compared to pure TPU (10 times higher) in order to avoid fiber breakage. TPU-CNT compounds with 0 wt.% or 1 wt.% offer the potential to be spun at high winding speeds of up to 650 m/min, whereas compounds with 2 wt.% or more CNT cannot be spun at speeds exceeding 37.2 m/min.
Especially in the case of low draw ratios and highly elastic melt, draw resonance can occur. This phenomenon causes an unevenness in the filament diameter, which at worst can lead to filament breakage [26]. In the experiments presented in this paper, there was a sharp increase in unevenness at a CNT content above 5 wt.% and a drawing speed of less than 15 m/min; hence, the resulting filaments were considered almost unusable for the desired purpose.
Due to high viscosities, great pressures of over 100 bar occurred at the spinneret, especially when processing compounds with high CNT contents. The reduced draw ratio in combination with constant extruder speed led to increasing filament diameters at increased CNT content. Moreover, the volume and the length-related filament mass increased accordingly. However, greater mass and minimized extensibility caused the polymer melt to emerge from the spinneret, which was no longer able to bear the weight of the solidifying filaments.   As a result, these filaments must be pulled off at a significantly slower pace compared to pure TPU (10 times higher) in order to avoid fiber breakage. TPU-CNT compounds with 0 wt.% or 1 wt.% offer the potential to be spun at high winding speeds of up to 650 m/min, whereas compounds with 2 wt.% or more CNT cannot be spun at speeds exceeding 37.2 m/min.
Especially in the case of low draw ratios and highly elastic melt, draw resonance can occur. This phenomenon causes an unevenness in the filament diameter, which at worst can lead to filament breakage [26]. In the experiments presented in this paper, there was a sharp increase in unevenness at a CNT content above 5 wt.% and a drawing speed of less than 15 m/min; hence, the resulting filaments were considered almost unusable for the desired purpose.
Due to high viscosities, great pressures of over 100 bar occurred at the spinneret, especially when processing compounds with high CNT contents. The reduced draw ratio in combination with constant extruder speed led to increasing filament diameters at increased CNT content. Moreover, the volume and the length-related filament mass increased accordingly. However, greater mass and minimized extensibility caused the polymer melt to emerge from the spinneret, which was no longer able to bear the weight of the solidifying filaments. the cross-section, whereas the CNT distribution becomes more inhomogeneous at a CNT content of 4 wt.% (Figure 4b). In the case of high CNT contents of 6 wt.% (Figure 4c,d), an outer sheath layer of CNT-poor TPU was formed during the spinning process. This means that although the TPU-CNT compounds were spun as monocomponents, a core-sheath structure was obtained. It can be assumed that the melt separated into components of high viscosity (high CNT content) and low viscosity (low CNT content) as it passed through the spinneret. During filament formation, the core was formed by high-viscosity melt, while low-viscosity melt formed into the sheath.

Microscopic Analyses
Polymers 2021, 13, x FOR PEER REVIEW 6 of 13 Figure 4 shows cross-sections of the melt spun filaments and the distribution of the CNTs. At a low CNT content of 2 wt.% (Figure 4a) CNTs are evenly distributed throughout the cross-section, whereas the CNT distribution becomes more inhomogeneous at a CNT content of 4 wt.% (Figure 4b). In the case of high CNT contents of 6 wt.% ( Figure  4c,d), an outer sheath layer of CNT-poor TPU was formed during the spinning process. This means that although the TPU-CNT compounds were spun as monocomponents, a core-sheath structure was obtained. It can be assumed that the melt separated into components of high viscosity (high CNT content) and low viscosity (low CNT content) as it passed through the spinneret. During filament formation, the core was formed by highviscosity melt, while low-viscosity melt formed into the sheath.

Stress-Strain Tests
All spun TPU-CNT filaments exhibited elongations at break of over 170 % and low Young's moduli of less than 80 kPa. Table 1 lists the fineness, elongation at break, Young's modulus, and electrical resistance of all filaments as a function of the CNT content and the winding speed.

Stress-Strain Tests
All spun TPU-CNT filaments exhibited elongations at break of over 170% and low Young's moduli of less than 80 kPa. Table 1 lists the fineness, elongation at break, Young's modulus, and electrical resistance of all filaments as a function of the CNT content and the winding speed.
The compounding of CNTs and TPU created a percolative system [44,45] with a percolation threshold between 3 and 4 wt.% CNT. The lowest achieved specific resistance was 110 ± 39 Ωcm. This value was obtained at an unstretched multifilament yarn with a CNT content of 5 wt.% and a spinning speed of 10 m/min. Under stretching load, the electrical resistances increase by up to one order of magnitude. This can be explained by the fact that the electrically conducting CNT particles are moved away from each other, so that percolation paths are interrupted. The lowest measured resistance value at 50% relative elongation is 662 ± 221 Ωcm and was achieved at filaments with 5 wt.% CNT and 8 m/min spinning speed. At a relative elongation of 100%, an electrical resistance of 2185 ± 608 Ωcm was recorded at 6 wt.% CNT and 10 m/min spinning speed. In general, electrical resistance increased in the case of faster spinning speeds. This behavior can be explained by the fact that at high spinning speeds, the CNTs within the solidifying filament were pulled away from each other, thus interrupting percolation paths. However, electrical conductivity in the unstretched filaments was not further improved by adding more CNTs beyond 5 wt.%. In unloaded yarns, higher electrical resistances were measured at 6 wt.% CNT than at 5 wt.% CNT. It can be assumed that due to the increased CNT content, the tendency of the CNTs to agglomerate was also more pronounced. Hence, more clusters were formed within the filament without improving its electrical conductivity as a result of insufficient distribution of the CNTs within the TPU. However, at relative elongations of up to 100%, some of these clusters may contribute to improving the filament's electrical conductivity. The particles are pulled away from each other under mechanical load so that agglomeration are broken up, and more particles are available to build percolation paths. Therefore, filaments with 6 wt.% CNT offer lower electrical resistances at 100% relative elongation than filaments with 5 wt.% CNT. Figure 5 provides examples for stress-strain diagrams of multifilament yarns with 3 wt.% CNT and multifilament yarns with 6 wt.% CNT. Both compounds were spun at a take-off speed of 15 m/min. Figure 5 provides examples for stress-strain diagrams of multifilament yarns with 3 wt.% CNT and multifilament yarns with 6 wt.% CNT. Both compounds were spun at a take-off speed of 15 m/min. It became evident that at higher CNT contents, the material failure of individual filaments occurred in a staggered manner. This suggests that the textile-physical properties of the filaments vary more widely among themselves than in compounds with a lower CNT content. At a CNT content of 6 wt.%, filaments began to break at an elongation of approx. 320 %, whereas other individual filaments did not fail even at an elongation exceeding 440 %. In contrast, at a CNT content of 3 wt.%, all filaments broke within a much smaller tensile force range. This confirms the assumption that the probability of CNT agglomeration increased significantly with increasing CNT content. Thus, the polymer network was more affected, causing local weak points to be generated in individual filaments and a staggered material failure to occur.
Furthermore, it was observed that the maximum tensile strength of the filaments increased, whereas the elongation at break decreased with increasing CNT content. For a multifilament with 3 wt.% CNT, the average tensile strength was 8.65 N, and the average elongation at break was 641%. If the CNT content increased to 6 wt.%, the average tensile strength almost doubled to 17.4 N; simultaneously, the elongation at break almost halved to 326 %. Figure 6a represents the elongation at break as a function of the CNT content at a constant spinning speed of 15 m/min. It can also be seen that Young's modulus ( Figure  6b) and fineness (Figure 6c) increase with increasing CNT content, while elongation at break decreases. The data collected for electrical resistance (Figure 6d) suggest large standard deviations, especially at a CNT content of 4 wt.% (specific electrical resistance: 1777 ± 756 Ωcm); thus, further investigations are needed to increase reliability. For multifilament yarns containing less than 4 wt.% CNT, no electrical resistance could be measured, because it is beyond the measurable range. This indicates that the percolation threshold must lie between 3 wt.% CNT and 4 wt.% CNT. It became evident that at higher CNT contents, the material failure of individual filaments occurred in a staggered manner. This suggests that the textile-physical properties of the filaments vary more widely among themselves than in compounds with a lower CNT content. At a CNT content of 6 wt.%, filaments began to break at an elongation of approx. 320%, whereas other individual filaments did not fail even at an elongation exceeding 440%. In contrast, at a CNT content of 3 wt.%, all filaments broke within a much smaller tensile force range. This confirms the assumption that the probability of CNT agglomeration increased significantly with increasing CNT content. Thus, the polymer network was more affected, causing local weak points to be generated in individual filaments and a staggered material failure to occur.
Furthermore, it was observed that the maximum tensile strength of the filaments increased, whereas the elongation at break decreased with increasing CNT content. For a multifilament with 3 wt.% CNT, the average tensile strength was 8.65 N, and the average elongation at break was 641%. If the CNT content increased to 6 wt.%, the average tensile strength almost doubled to 17.4 N; simultaneously, the elongation at break almost halved to 326%. Figure 6a represents the elongation at break as a function of the CNT content at a constant spinning speed of 15 m/min. It can also be seen that Young's modulus ( Figure 6b) and fineness ( Figure 6c) increase with increasing CNT content, while elongation at break decreases. The data collected for electrical resistance (Figure 6d) suggest large standard deviations, especially at a CNT content of 4 wt.% (specific electrical resistance: 1777 ± 756 Ωcm); thus, further investigations are needed to increase reliability. For multifilament yarns containing less than 4 wt.% CNT, no electrical resistance could be measured, because it is beyond the measurable range. This indicates that the percolation threshold must lie between 3 wt.% CNT and 4 wt.% CNT.
Even the lowest value of electrical resistance was several orders of magnitude higher than the resistivity of fine copper wires (1.7 × 10 −6 Ωcm [46]); however, it was in the same range as electrically conductive liquid rubber (30-75 Ωcm [47,48]). Thus, CNT-filled TPU is suitable for a wide range of sensors and actuators, as its combination of high elasticity, electrical conductivity, and spinnability results in a completely new property profile. Figure 7 provides a first impression of the sensorial behavior of the melt spun fibers. The diagram shows the correlation between mechanical elongation and electrical resistance for three different filament yarns, each containing 5 wt.% CNT but spun at different winding speeds. It can be seen that the electrical resistance increases with increasing winding speed in the spinning process. Furthermore, the intermediate peaks (also known as shoulder phenomenon) are less pronounced with increasing winding speed. Nevertheless, there is no unambiguous correlation between mechanical elongation and electrical resistance in any specification. Further investigations will follow to determine the extent to which Polymers 2021, 13, 590 9 of 12 pretreatment of the filaments by means of mechanical pre-stretching or annealing has a positive influence on the sensorial behavior.
Due to the very high viscosity gradient in the liquid compound, a CNT-rich region was formed in the filament core, while a sheath of almost pure TPU surrounded this core (see Figure 4c,d). Thus, without the need for an additional work step or a bicomponent melt spinning process, a core-sheath filament with an electrically conductive filament core and an insulating sheath layer was created. This insulating layer offers advantages for many applications, for example, by minimizing the probability of undesirable short circuits in sensor networks. It is also worth mentioning that the sheath established a strong physical and chemical bond with the electrically conductive filament core, hence encouraging the assumption of high fatigue strength. Even the lowest value of electrical resistance was several orders of magnitud than the resistivity of fine copper wires (1.7 10 Ωcm [46]); however, it was in t range as electrically conductive liquid rubber  Ωcm [47,48]). Thus, CNT-fil is suitable for a wide range of sensors and actuators, as its combination of high el electrical conductivity, and spinnability results in a completely new property pro ure 7 provides a first impression of the sensorial behavior of the melt spun fib diagram shows the correlation between mechanical elongation and electrical re Due to the very high viscosity gradient in the liquid compound, a CNT-rich region was formed in the filament core, while a sheath of almost pure TPU surrounded this core (see Figure 4c,d). Thus, without the need for an additional work step or a bicomponent melt spinning process, a core-sheath filament with an electrically conductive filament core and an insulating sheath layer was created. This insulating layer offers advantages for many applications, for example, by minimizing the probability of undesirable short circuits in sensor networks. It is also worth mentioning that the sheath established a strong physical and chemical bond with the electrically conductive filament core, hence encouraging the assumption of high fatigue strength.

Conclusions
By adding CNTs to TPU, fibers that are elastic and electrically conductive can be melt spun. The resulting filaments exhibited a very high elongation at break while providing mechanical properties in the range of conventional elastic fibers and electrical conductivities in the range of electrically conductive liquid rubbers. Additionally, the fiber core established a highly favorable bond with the surrounding insulating layer of pure TPU. In future research projects, the insulating properties have to be determined more specifically and the surrounding sheath has to be thoroughly investigated in terms of potential conducting flaws. Thus, the newly developed TPU-CNT filaments are durable and highly stress resistant. This new class of electrically conductive, highly stretchable yarns offers a great potential for sensors (for example, as strain sensors, pressure sensors, and electrochemical sensors) and actuators (for example, in dielectric elastomer actuators). Furthermore, these yarns could be used in wearable smart textiles for energy harvesting, computing, and communication.

Conclusions
By adding CNTs to TPU, fibers that are elastic and electrically conductive can be melt spun. The resulting filaments exhibited a very high elongation at break while providing mechanical properties in the range of conventional elastic fibers and electrical conductivities in the range of electrically conductive liquid rubbers. Additionally, the fiber core established a highly favorable bond with the surrounding insulating layer of pure TPU. In future research projects, the insulating properties have to be determined more specifically and the surrounding sheath has to be thoroughly investigated in terms of potential conducting flaws. Thus, the newly developed TPU-CNT filaments are durable and highly stress resistant. This new class of electrically conductive, highly stretchable yarns offers a great potential for sensors (for example, as strain sensors, pressure sensors, and electrochemical sensors) and actuators (for example, in dielectric elastomer actuators). Furthermore, these yarns could be used in wearable smart textiles for energy harvesting, computing, and communication.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.