Manufacturing of Carbon Nanotube-Polystyrene Filament for 3D Printing: Nanoparticle Dispersion and Electromagnetic Properties

: 3D printing is a promising technology for creating polymer objects of a given architecture with speciﬁed functional properties. In fact, the choice of ﬁlaments for 3D printing is quite limited. Here, we report a process for producing polystyrene ﬁlaments with 0.0025–2 wt.% single-walled carbon nanotubes (SWCNTs) by extruding crushed polystyrene composites. The resulting ﬁlaments are characterized by a high uniformity of ﬁller distribution and the absence of air pores. Comparison of microscopy data and electromagnetic properties of base composites and composite materials printed from ﬁlaments showed that extrusion and printing improve SWCNT dispersion. The proposed method can be used to create ﬁlaments for 3D printing of objects from various base polymers containing functional ﬁllers up to the electrical percolation threshold and above.


Introduction
The rapid and cost-effective manufacturing of physical objects with complex geometries and functional properties, tailored to a specific practical application, is one of the most challenging tasks in any process chain. The speed and precision of 3D printers make them a promising tool for prototyping polymer parts. Today, 3D printing is widely used in various fields, including electronics [1][2][3], optics [4][5][6], aerospace [7,8], and medicine [9,10]. Among other polymer-based 3D printing technologies [11][12][13], fused deposition modeling (FDM) looks the most promising for practical applications, including the production of capacitors [14,15], electromagnetic shielding components [16][17][18], sensors [19][20][21], and antistatic coatings [1,22]. FDM starts with the digital design of an element followed by the creation of a 3D object layer by layer from a polymer filament [2,[23][24][25][26]. Although FDM printing is the most popular and economical method [27], it suffers from the limited choice of materials for 3D printing [28] and the associated narrow range of their physical characteristics. Thus, the development of a technique for manufacturing a 3D printing filament for the reproducible production of polymer objects with the required functional properties is an important issue for the industrial implementation of the FDM method.
The most common materials for 3D printing are thermoplastic elastomers such as polylactic acid (PLA), acrylonitrile butadiene styrene (ABS), polyurethane (PU), polyethylene (PE), and polystyrene (PS), etc., and less often, plastics with fillers are used [2,8,11,17,21,29,30]. Examples of the created products are a space radiation protection panels made of PE with the addition of basalt powder [8], a GHz frequency range screen made of PLA with 10% of graphene [17], and a tension sensor made of PU with carbon nanotubes (CNTs) [21]. A rather small number of works devoted to FDM printing from polymer filaments with nanoparticles is primarily due to the lack of commercially available materials for 3D printing and preliminary studies of changes in the structure of the material during filament extrusion and subsequent 3D printing [28]. In addition, working with such materials is associated with the phenomenon of nanoparticle agglomeration, which worsens the functional properties of the created element [27,30] and can block the nozzle, causing material flow instability while printing [31].
The problem of CNT aggregation can be solved by their functionalization [45,46], which is usually undesirable due to the deterioration of nanotube conductivity. Alternatively, these difficulties can be addressed by the right combination of homogenization techniques [41,43,[48][49][50]. In particular, a higher level of CNT dispersion was obtained during post-treatment of PC on a forge-rolling machine [43]. Higher CNT dispersion lowers the percolation threshold, improving PC conductivity. Although forge-rolling reduces aggregates of nanoparticles, nanotubes are also destroyed at the same time [49,51]. Extrusion can also improve the dispersion of CNTs in a polymer but only if the solvent is completely evaporated before [50]. Otherwise, numerous air pores form in the volume of the extruded filament and the conductivity of the PC deteriorates [4,52].
In this work, we use homogenized PCs containing single-walled CNTs (SWCNTs) as the base material for filament extrusion for 3D printing instead of the traditional mixture of polymer and nanotubes. The proposed approach successfully works for a wide range of SWCNTs concentrations (0.0025-2 wt.%), improving the distribution of SWCNTs in PS matrix and lowering the electrical percolation threshold.

Materials
The material containing 75% SWCNTs with a diameter of 1.6-2.2 nm was produced by OCSiAl (Novosibirsk, Russia, LOT # 50-04122014). The catalyst and amorphous carbon by-products were removed according to the procedure described in [53]. SWCNTs (1 g) were heated in a muffle furnace at 500 • C for 1 h to open their tips and to oxidize amorphous carbon. The product was stirred in HCl (200 mL) for 12 h at 60 • C to dissolve the iron catalyst, washed with distilled water to neutral pH, and dried in a muffle furnace at 100 • C for 2 h. The content of amorphous carbon was less than 1% and the content of iron-containing particles did not exceed 2 wt.% [53]. The purified SWCNT sample consists of bundles of nanotubes with a diameter of about 20 nm and their agglomerates with a diameter of about 80-120 nm (see Supplementary Materials for details).
High-impact PS (HIPS-850) with a density of 1.04 g/cm 3 was used as the polymer base.

Filaments Preparation
Base PCs consisting of purified SWCNTs and PS were fabricated by the method described elsewhere [54]. PS granules (20 g), the required amount of SWCNTs, and CHCl 3 (300 mL) were treated with an ultrasonic homogenizer UZTA-0.15/22-O (U-SONIC, Biysk, Russia, 22 kHz) at a power of 70 W for 30 min. The resulting suspension was located in a fume hood at room temperature for a day and then in a desiccator at 40 • C until the solvent had completely evaporated.
The filaments were made from crushed base PCs with a SWCNT content of 0.0025-2 wt.% on a Wellzoom B single screw extruder (Shenzhen Mistar Technology Co., Ltd., Shenzhen, China) at a power of 120 W and a nozzle diameter of 2.85 mm. The extrusion temperatures varied from 195 to 235 • C depending on the viscosity of the composite, the screw speed was fixed to be about 100 rpm. The resulting filaments had the same diameter along the length, uniform color, and did not contain defects such as tears (see Supplementary Materials). That indicates that the polymer was not oxidized during extrusion and SWCNTs were evenly distributed throughout its volume.

3D Printing
3D printed samples were obtained from filaments on an Ultimaker 2 Go 3D printer (Ultimaker B.V., Amsterdam, the Netherlands). The printing temperature (210-260 • C) was selected experimentally for each SWCNT concentration, the platform temperature was about 90 • C, the printing speed was 30 mm/s, and printing was carried out with a 0.6 mm nozzle. The tolerance level of filament diameter was 2.85 ± 0.3 mm. Plates (12 × 12 × 0.6 mm) and rings (outer diameter 7.0 mm, inner diameter 3.0 mm, 2.3 mm) were printed. The upper temperature limit of the apparatus did not allow printing with a filament containing more than 0.5 wt.% SWCNTs due to clogging of a printer nozzle caused by the insufficient viscosity of the filled PC. The schematic representation of all stages of printed samples fabrication is shown in Figure 1.
Nanomanufacturing 2022, 2,3 Biysk, Russia, 22 kHz) at a power of 70 W for 30 min. The resulting suspension was located in a fume hood at room temperature for a day and then in a desiccator at 40 °C until the solvent had completely evaporated. The filaments were made from crushed base PCs with a SWCNT content of 0.0025-2 wt.% on a Wellzoom B single screw extruder (Shenzhen Mistar Technology Co., Ltd., Shenzhen, China) at a power of 120 W and a nozzle diameter of 2.85 mm. The extrusion temperatures varied from 195 to 235 °C depending on the viscosity of the composite, the screw speed was fixed to be about 100 rpm. The resulting filaments had the same diameter along the length, uniform color, and did not contain defects such as tears (see Supplementary Materials). That indicates that the polymer was not oxidized during extrusion and SWCNTs were evenly distributed throughout its volume.

3D Printing
3D printed samples were obtained from filaments on an Ultimaker 2 Go 3D printer (Ultimaker B.V., Amsterdam, the Netherlands). The printing temperature (210-260 °C) was selected experimentally for each SWCNT concentration, the platform temperature was about 90 °C, the printing speed was 30 mm/s, and printing was carried out with a 0.6 mm nozzle. The tolerance level of filament diameter was 2.85 ± 0.3 mm. Plates (12 × 12 × 0.6 mm) and rings (outer diameter 7.0 mm, inner diameter 3.0 mm, 2.3 mm) were printed. The upper temperature limit of the apparatus did not allow printing with a filament containing more than 0.5 wt.% SWCNTs due to clogging of a printer nozzle caused by the insufficient viscosity of the filled PC. The schematic representation of all stages of printed samples fabrication is shown in Figure 1.

Microscopic Characterization
The morphology of the base PCs with SWCNTs (films 200 µm thick), printed plates, and cross-sections of the obtained filaments was studied by optical microscopy using an Olympus BX51 microscope (Olimpus Corp., Tokyo, Japan) in transmission and reflection modes, respectively.
To examine the distribution of SWCNTs in PS in more details, the filaments with 0.005 and 0.1 wt.% filler loading were studied using scanning electron microscopy (SEM) on a JEOL-6700F microscope (JEOL Ltd., Tokyo, Japan) at an accelerating voltage of 15 kV. To do this, the filaments frozen with liquid nitrogen were broken perpendicular to their axes. The resulting cross-sections were covered with a 10 nm thick gold layer using a conventional sputtering tool.

Electromagnetic Measurements
The measurements were carried out for the base PC films and 3D printed samples (plates and rings). Plates were used to measure impedance at low frequencies, while rings were used to study GHz characteristics.
The real and imaginary components of the impedance were measured by the twocontact method on a Z-2000 impedance meter (Elins, Chernogolovka, Russia) in the frequency range of 1 kHz-2 MHz. The electrical conductivity σ AC and permittivity ε AC were reconstructed from the impedance data taking into account the following model [61]. The measurement cell was considered as four parallel elements: R 1 and C 1 are resistance and capacitance of the cell, and R 2 and C 2 are resistance and capacitance for studied PC film. R 1 and C 1 parameters were determined from the impedance data with empty cell while maintaining the same interplate distance as for the corresponding PC film. Further, the material parameters, namely the AC electrical conductivity σ AC and permittivity ε AC , were found from the calculated C 2 and R 2 values in assumption that the measurement cell can be considered as flat capacitor with leakage currents.
Microwave measurements were carried out on a Mikran P2M-04A scalar network analyzer (Mikran, Tomsk, Russia) in a coaxial cell 12.9 mm long in the range of 100 MHz-4 GHz using a voltage standing wave ratio (VSWR) sensor DK1-04-11R-11R (Mikran, Tomsk, Russia). As a result, the relative amplitudes of the wave reflected from the sample surface (S 11 ) and the wave transmitted through the sample (S 21 ) were obtained. The measured S-parameters were used to calculate the high-frequency values of ε AC and σ AC of the investigated materials in accordance with the referred works [62][63][64].

Optimization of Extrusion Conditions
The extrusion of PS granules without SWCNT additives was carried out at a temperature of 220 • C, which corresponds to the extrusion temperature of high-impact PS (190-230 • C). As a result, a uniform white filament with a diameter of 2.8 mm was obtained ( Figure 2).
Nanomanufacturing 2022, 2,5 released at the PS/SWCNT interface ensures the transition of the material to the plastic state at a lower temperature. It can be seen that the filament diameter decreases when SWCNTs are added to PS (Figure 2c). However, the dependence of the PC filament diameter on the SWCNT content shows a rise at a filler content of 0.1 wt.% followed by a gradual decrease. We assume that this indicates the formation of a network of nanotubes, which affects the flow of the polymer melt [65]. It should be noted that the reported temperatures are optimal for the PS filaments with SWCNTs at the screw speed of about 100 rpm using a Wellzoom B extruder. Changing any process parameter, while maintaining the found optimal temperature, will immediately affect the filament diameter. Figure 3a,b shows transmission optical images of base PC films containing 0.005 and 0.1 wt.% SWCNTs. Even at lower nanotube concentrations, the base PC films appear opaque, however, areas of unfilled PS remain, indicating the formation of micron-sized SWCNT agglomerates, consistent with the previous work [69]. Cross-section images of filaments with the same SWCNT loading have a uniform black color without visible air pores and the unfilled areas of the polymer (Figure 3c,d). That distinguishes our fila- The addition of SWCNTs affects the rheological properties of polymers [60,65]. The reorientation of nanotubes in the flow field during extrusion and the elimination of interpenetration of soft polymeric matrix with highly anisotropic nanoparticles change viscoelastic properties of melt matrix [65] and the plastic transition temperature [66]. As a result, the optimal extrusion temperature for obtaining filaments of an acceptable diameter (2.85 ± 0.3 mm) varies depending on the nanotube content (Figure 2b,c). With increasing concentration from 0.0025 to 0.05 wt.%, the processing temperature becomes 5-20 • C higher than for clear PS. The maximum temperature of about 235 • C was required for the extrusion of PC with 0.005 wt.% SWCNTs. With a larger load of SWCNTs, the extrusion temperature decreases, reaching 195 • C for a filament with 2 wt.% SWCNTs. This behavior is associated with the high thermal conductivity coefficient for SWCNT bundles (6.6-126.6 W/(m K)) [67] versus 0.14 W/(m K) for PS [68]). Therefore, the heat released at the PS/SWCNT interface ensures the transition of the material to the plastic state at a lower temperature. It can be seen that the filament diameter decreases when SWCNTs are added to PS (Figure 2c). However, the dependence of the PC filament diameter on the SWCNT content shows a rise at a filler content of 0.1 wt.% followed by a gradual decrease. We assume that this indicates the formation of a network of nanotubes, which affects the flow of the polymer melt [65].

Morphology of Base PCs and Filaments
It should be noted that the reported temperatures are optimal for the PS filaments with SWCNTs at the screw speed of about 100 rpm using a Wellzoom B extruder. Changing any process parameter, while maintaining the found optimal temperature, will immediately affect the filament diameter. Figure 3a,b shows transmission optical images of base PC films containing 0.005 and 0.1 wt.% SWCNTs. Even at lower nanotube concentrations, the base PC films appear opaque, however, areas of unfilled PS remain, indicating the formation of micron-sized SWCNT agglomerates, consistent with the previous work [69]. Cross-section images of filaments with the same SWCNT loading have a uniform black color without visible air pores and the unfilled areas of the polymer (Figure 3c,d). That distinguishes our filaments from those obtained in [4,52]. In the work [52], filaments were obtained using a mixture of ABS and multiwalled CNT solutions. The pores in those filaments were optically detected even after 8 extrusion cycles, while the electrical properties of the materials deteriorated after the third cycle. SEM analysis of fracture surfaces of filaments with SWCNT concentrations of 0.005 and 0.1 wt.% shows uniform nanotube networks coated with PS (Figure 3e,f). Bundles of nanotubes with a diameter of 15-20 nm and a length of about 2 µm can be distinguished (inset in Figure 3f). The better dispersion of SWCNTs in filaments as compared to base PCs could be explained by the splitting of SWCNT bundle agglomerates and improved distribution of nanotubes in the polymer matrix after extrusion.  Figure 4 shows the frequency and concentration dependencies of AC electrical conductivity and permittivity for base PCs and plates printed from the filaments. At a frequency of 1 MHz, the AC conductivity of base PCs increases from 10 −6 S/m to 10 −5 S/m with an increase in the SWCNT content from 0.0025 to 0.5 wt.%. With a further increase in the nanotube loading, the increase in conductivity slows down (σAC is 10 −4 S/m for base   Figure 4 shows the frequency and concentration dependencies of AC electrical conductivity and permittivity for base PCs and plates printed from the filaments. At a frequency of 1 MHz, the AC conductivity of base PCs increases from 10 −6 S/m to 10 −5 S/m with an increase in the SWCNT content from 0.0025 to 0.5 wt.%. With a further increase in the nanotube loading, the increase in conductivity slows down (σ AC is 10 −4 S/m for base PC with 2.0 wt.% loading). This trend is less pronounced in microwaves (Figure 4c) due to the higher conductivities of both PS (black curves in Figure 4a,b) and SWCNTs [70]. The permittivity of base PCs at 1 MHz increases from 2.8 to 18.5 when the SWCNT loading changes from 0.025 to 0.5 wt.% (Figure 4d,f). A further increase in the SWCNT content reduces the permittivity, which drops to 4.2 at 2.0 wt.% SWCNTs. This phenomenon can be understood within the micro capacitor model [71]. At low loading, nanotubes and their bundles act as highly polarizable domains isolated from each other. The polymer layer separating adjacent nanoparticles becomes thinner with concentration, creating micro capacitors, causing a significant increase in permittivity, observed as a peak at 0.5 wt.% SWCNTs. Above this critical concentration, the thickness of the polymer decreases so much that the interparticle current increases the losses, deteriorating the permittivity to the value of the matrix.

Electrical Properties of Base PCs and Printed Samples
Nanomanufacturing 2022, 2,7 characterized by the transverse conductivity of the nanotubes, while the response of the base PCs contains both less-and more-conductive components. On the other hand, microscopy images show that the surface of the printed plate is not flat (Figure 3g,h). This is due to the peculiarities of FDM technology. The product is grown layer by layer with several parallel print head passes. The deposited strips of diluted polymer cool down before the layer creation ends. The result is a relief surface. When the next polymer layer covers the previous one, the obtained air gaps go into volume of the printed product [11,72]. At a sufficiently high SWCNT concentration, these structural features do not affect the flow of electrical current through the sample. However, when the fraction of SWCNTs is small, inhomogeneity in the volume of the plate worsen its conductivity. The maxima of the permittivity appeared in the printed plates at 0.1 wt.% SWCNTs (Figure 4e,f). Below this concentration, the permittivity of printed plates is higher than that of base PCMs containing the same SWCNT portion. This indicates that the extrusion and printing processes improve the dispersion of SWCNTs in PS as compared to base PCs.
Electromagnetic data show that highly conductive 3D printed composite materials can be used to create antistatic coatings and microwave screens. At low concentrations of SWCNTs, the conductivity of printed samples somewhat decreases compared to the values for base PCs (Figure 4b,c). Transport degradation reflects changes in the network structure and composite architecture. SWCNTs are oriented in the flows of the extruder and printer [65], while nanotubes in base PCs are randomly distributed. Since the measurements are carried out for the in-plane orientation of the electric field relative to the PC layer, the conductivity of the printed plates is mainly characterized by the transverse conductivity of the nanotubes, while the response of the base PCs contains both less-and more-conductive components. On the other hand, microscopy images show that the surface of the printed plate is not flat (Figure 3g,h). This is due to the peculiarities of FDM technology. The product is grown layer by layer with several parallel print head passes. The deposited strips of diluted polymer cool down before the layer creation ends. The result is a relief surface. When the next polymer layer covers the previous one, the obtained air gaps go into volume of the printed product [11,72]. At a sufficiently high SWCNT concentration, these structural features do not affect the flow of electrical current through the sample. However, when the fraction of SWCNTs is small, inhomogeneity in the volume of the plate worsen its conductivity.
The maxima of the permittivity appeared in the printed plates at 0.1 wt.% SWCNTs (Figure 4e,f). Below this concentration, the permittivity of printed plates is higher than that of base PCMs containing the same SWCNT portion. This indicates that the extrusion and printing processes improve the dispersion of SWCNTs in PS as compared to base PCs.
Electromagnetic data show that highly conductive 3D printed composite materials can be used to create antistatic coatings and microwave screens.

Conclusions
Polymers with nanoparticles are in high demands for a range of practical applications such as flexible conductive screens, electromagnetic shielding, microwave and terahertz optics, etc. Each one requires materials with special properties formed to a certain architecture, which can be obtained using the modern technologies of 3D printing. In this work, we described a simple process for manufacturing of PS filaments with permittivity and conductivity determined by the SWCNT concentration for FDM 3D printer and evaluated the nanotube dispersion and electromagnetic properties of printed plates. The process is based on the extrusion of milled PCs and can be transferred to the production of highly homogeneous filaments from other thermoplastic polymer matrices and reinforcing materials. The absence of air pores in the filaments is achieved by careful evaporation of the solvent during the production of PC and optimization of extrusion parameters. Extrusion and 3D printing serve as additional steps for splitting the nanotube agglomerate inherited from base PCs and further improving the dispersion of SWCNTs. A comparison of the electromagnetic properties of base PCs and PCs printed from fabricated filaments revealed an improvement in the dielectric properties in the latter case. This suggests the practical application of the obtained material for 3D printing of protection skins for electronic components from damage caused by electrostatic discharge, impeding the output of any signal and preventing electromagnetic interference from affecting sensitive electronics.