The Use of 3D Printing Technology for Manufacturing Metal Antennas in the 5G/IoT Context

With the rise of 5G, Internet of Things (IoT), and networks operating in the mmWave frequencies, a huge growth of connected sensors will be a reality, and high gain antennas will be desired to compensate for the propagation issues, and with low cost, characteristics inherent to metallic radiating structures. 3D printing technology is a possible solution in this way, as it can print an object with high precision at a reduced cost. This paper presents different methods to fabricate typical metal antennas using 3D printing technology. These techniques were applied as an example to pyramidal horn antennas designed for a central frequency of 28 GHz. Two techniques were used to metallize a structure that was printed with polylactic acid (PLA), one with copper tape and other with a conductive spray-paint. A third method consists of printing an antenna completely using a conductive filament. All prototypes combine good results with low production cost. The antenna printed with the conductive filament achieved a better gain than the other structures and showed a larger bandwidth. The analysis recognizes the vast potential of these 3D-printed structures for IoT applications, as an alternative to producing conventional commercial antennas.


Introduction
The fifth generation of mobile communications (5G) is driven by an unprecedented growth in the number of connected devices and shared data [1]. With the main goal of being a unifying connectivity structure for the next decade and beyond [2], 5G enables the IoT reality, where a device will be able to maintain connectivity, regardless of time or location. Everything will be monitored, measured, or sensed, and to gather that information, the number of devices interacting with the surroundings will increase exponentially.
New scenarios such as the proliferation of sensors to deliver IoT services associated to home appliances, health monitoring, smart offices, efficient navigation systems (autonomous cars), immersive multimedia experiences, either through augmented or virtual reality and cloud computing, will all be combined in a typical 5G network [1]. Given the number of the connected devices, their diversity of nature, sizes, and shapes, the antennas will face multiple challenges with various forms and the combination of several materials.
New techniques to build antennas need to be investigated, to properly create radiation structures in the daily objects. Technology must be able to handle heterogeneous and challenging layouts, always bearing in mind the improvement of both energy and cost efficiencies along with spectrum performance [3].
Wider bandwidths are probably the most effective method to provide the data demands for 5G services [4], thus the migration to the millimeter waves region becomes mandatory. Several concerns arise with these operation frequencies, essentially due to the huge path-loss and consequent fragile link result of the occurring diffractions [1]. To overcome these communication issues, combating the large propagation loss in mmWaves, Also, for the Ka-band, an attractive metallization process was tested, requiring no treatment process before and after 3D printing [16]. The printed structure was metallized using electrodeposition of silver conductive ink and the manufactured prototype achieved a good performance. The drawback of this technique is that the coating process produced a non-uniform layer of paint.
Structures that do not require a post-printing treatment process have received great attention, irrespective of the structures being printed with conductive materials or directly printed on metal [17,18]. In [19], two-horn antennas have been modelled using 3D printing, operating at 12 GHz. One was printed with a regular filament and covered with copper tape while the other was directly printed with ProtoPasta, a conductive PLA. The first prototype obtained very promising results, both in bandwidth and measured gain. On the other hand, the prototype printed with the conductive material obtained a low gain when compared with the typical values of horn antennas and an efficiency of only 35%. Therefore, ProtoPasta does not prove to be the most suitable material for applications within the context of the new mobile communication systems.
To date, there are a few studies on 3D printing with conductive filaments [20][21][22][23]. However, in all these works the operating frequency is low when compared to the 5G demands. Regarding the conductive filaments, an effort was made in the parametric study of 3D printing specifications for use in printing topologies with the Electrifi material [24]. Although a microstrip patch antenna and a pyramidal horn antenna were correctly built with Electrifi [24], respectively for the operating frequencies of 2.5 GHz and 5.8 GHz, there are some challenges in the production of RF devices for higher frequencies using this type of material. In [25][26][27] three different antennas were developed using Electrifi filament with conductive properties, an electric meandered dipole antenna, a 3D-printed dipole, and a 3D-printed conformal patch antenna, for 915 MHz, 900 MHz, and 2.32 GHz respectively.
A prototype of a 3D metal printed polarization reconfigurable horn antenna is presented in [28] for K/Ka band, using titanium material, revealing a reasonable agreement between simulations and measurements. A low-profile all-metal antenna design with resonant cavity is described in [29] to operate in the frequency range of 130-150 GHz. To create the conductive layer, the antennas were coated with a thin layer of platinum using a physical vapor deposition. Another prototype was made using direct metal laser sintering. Both processes present a huge complexity and cost. This paper presents and compares a set of techniques for the development of typically metallic antennas using 3D printing. In this sense, as an example, several horn antennas are developed to operate at 28 GHz, for 5G/IoT communications, using different methods, these methods can be replicated to other different structures.
The article is divided into six sections, starting with an introduction with a state-of-theart report, where possible applications for the prototypes proposed are presented and the objectives of this work are settled. In the Section 2, all construction methods are described, along with the materials used in each prototype. In Section 3 the simulation and measured results of all designed antennas are shown. Next, the Section 4 presents an analysis of the results obtained. Finally, in Section 6, the main conclusions are reported.

Basis Antenna Structure
In this work, the analysis and description of the various techniques for the construction of typically metallic radiating elements using the potential of 3D printing will be supported by the construction of a typical structure which offers reasonable high bandwidths and gain values, which are the horn antennas. However, it can be extended to other types of metal structures.
Horn antennas are one of the simplest antennas in the microwave frequency bands. Traditionally, these antennas are made of metallic materials and are typically fed by a section of a waveguide. Their structure forms a smooth transition between the waveguide and the free space, directing the radio waves into a beam [30]. Horn antennas may adopt several forms and the most common types are the rectangular, or pyramidal, and the conical. Rectangular horns start from a rectangular waveguide propagating the TE10 mode, and in these cases, one face is enlarged with a pyramidal shape, as can be seen in Figure 1a,b. Due to its ease of construction, wide bandwidth, versatility, ease of excitation, and high gain, this type of antenna is used in multiple applications.
tion of typically metallic radiating elements using the potential of 3D printing will be supported by the construction of a typical structure which offers reasonable high bandwidths and gain values, which are the horn antennas. However, it can be extended to other types of metal structures.
Horn antennas are one of the simplest antennas in the microwave frequency bands. Traditionally, these antennas are made of metallic materials and are typically fed by a section of a waveguide. Their structure forms a smooth transition between the waveguide and the free space, directing the radio waves into a beam [30]. Horn antennas may adopt several forms and the most common types are the rectangular, or pyramidal, and the conical. Rectangular horns start from a rectangular waveguide propagating the TE10 mode, and in these cases, one face is enlarged with a pyramidal shape, as can be seen in Figure  1a,b. Due to its ease of construction, wide bandwidth, versatility, ease of excitation, and high gain, this type of antenna is used in multiple applications. In this work, the pyramidal structure was selected, mainly due to the associated simplicity of construction. To properly design the antenna, some aspects were considered, regarding gain, operating frequency, and the feeding waveguide dimensions (wg_a × wg_b).
The waveguide's dimensions were found based on the standard rectangular waveguide WR34 (8.636 mm × 4.318 mm, which operates from 22 GHz up to 33 GHz). Regarding the size of the horn, it was calculated to operate at 28 GHz and perform a gain of G = 12 dBi. In horn antennas the gain is the starting point to calculate its aperture dimensions. Bearing in mind that it is important to obtain a compact prototype, this gain was chosen as a tradeoff for a value that would not represent a prototype too large. Considering the design parameters shown in Figure 1, the theoretical models [31] were used to estimate the initial dimensions of the horn antenna.
The preliminary values of the horn were submitted to an optimization process using the CST simulator. Although the waveguide was based on a standard, the versatility of the 3D printing allows to optimize its dimensions for this specific application.
Horn antennas are typically fed by a radiation element that is placed at a λ/4 distance from the waveguide's closed face. There are a variety of possibilities to excite a horn antenna, however a slot or a monopole element are the most common methods. In this work, and for simplicity, RS-Pro-27GHz SMA connector (two hole flange mount) operating as a monopole antenna was used, as it is shown in Figure 2. Its inner pin has a length of La and acts as a monopole at 28 GHz, properly feeding the structure. In this work, the pyramidal structure was selected, mainly due to the associated simplicity of construction. To properly design the antenna, some aspects were considered, regarding gain, operating frequency, and the feeding waveguide dimensions (wg_a × wg_b).
The waveguide's dimensions were found based on the standard rectangular waveguide WR34 (8.636 mm × 4.318 mm, which operates from 22 GHz up to 33 GHz). Regarding the size of the horn, it was calculated to operate at 28 GHz and perform a gain of G = 12 dBi. In horn antennas the gain is the starting point to calculate its aperture dimensions. Bearing in mind that it is important to obtain a compact prototype, this gain was chosen as a tradeoff for a value that would not represent a prototype too large. Considering the design parameters shown in Figure 1, the theoretical models [31] were used to estimate the initial dimensions of the horn antenna.
The preliminary values of the horn were submitted to an optimization process using the CST simulator. Although the waveguide was based on a standard, the versatility of the 3D printing allows to optimize its dimensions for this specific application.
Horn antennas are typically fed by a radiation element that is placed at a λ/4 distance from the waveguide's closed face. There are a variety of possibilities to excite a horn antenna, however a slot or a monopole element are the most common methods. In this work, and for simplicity, RS-Pro-27GHz SMA connector (two hole flange mount) operating as a monopole antenna was used, as it is shown in Figure 2. Its inner pin has a length of La and acts as a monopole at 28 GHz, properly feeding the structure.

Proposed Metallization Techniques
To analyze the different techniques of producing an antenna using the 3D printing technology, three approaches were adopted, using copper tape, conductive paint, and conductive filament. All antennas were printed with an Ultimaker 3 Extended printer [32], which is based on the fused deposition modelling (FDM) technique. Their design process

Proposed Metallization Techniques
To analyze the different techniques of producing an antenna using the 3D printing technology, three approaches were adopted, using copper tape, conductive paint, and conductive filament. All antennas were printed with an Ultimaker 3 Extended printer [32], which is based on the fused deposition modelling (FDM) technique. Their design process and construction methods are presented in the following sections.

Horn Antenna with Copper Tape Metallization
The process started by printing the optimized horn antenna structure of Figure 1 with regular PLA, which is the most common material within the 3D printing filaments. After printing, the antenna was covered with a single layer of copper tape.
The PLA used in this process was from the Ultimaker supplier and it possesses a dielectric constant ε r of 2.7 and a dissipation factor tg(δ) of 0.008 @ 1 MHz [19]. The copper tape was from Würth Elektronik and has a thickness of 0.04 mm. It is important to highlight that in the case of this particular tape, its glue has electrical conductivity properties. The final parameters of the antenna's dimensions are presented in Table 1 and the produced prototype is shown in Figure 2. The wall thickness of the antenna is 0.8 mm.

Horn Antennas with Copper Conductive Paint
The construction process of these antennas is similar to that previously described, except that instead of covering the structure with copper tape, a conductive spray-paint is applied on the top of the PLA structure.
The commercial spray used for coating these antennas was the RS EMI/RFI Shielding Aerosol 400 mL [33]. This is a spray that is easy to apply and with a drying time of approximately 5 min. However, it reaches its maximum conductivity after 24 h of its application.
Several prototypes were assembled, with small modifications, regarding the number of conductive layers, and the roughness of the structure walls, to analyze their impact in the performance of the antenna and therefore, to study the best technique to fabricate this type of antenna. Five prototypes were printed, where in three of them the spray was directly applied to the antenna without any preparation. The other two prototypes were previously sanded to achieve a smoother surface, diminishing the roughness among printed layers, where the paint would be applied.
With the first three models it was intended to analyze the impact of the variation of the number of layers of conductive spray-paint applied, and thus, two, three, and four layers of the above-mentioned spray coating were applied. In the sanded prototypes two and three layers of the same spray-paint were applied. Figure 3 shows two of the prototypes made with conductive paint using each technique, without the sanding process, Figure 3a, and with the sanding process, Figure 3b.

Horn Antennas Built from Conductive Filament
Typically, antennas produced with 3D printing technology require treatment after printing, such as the metallization of their structure. Given the increasing advancement in technology and the widespread use of mobile communications, it becomes necessary to develop antennas that are easy and fast to produce in mass. Therefore, a possible solution would be to produce an antenna entirely with a conductive filament.

Horn Antennas Built from Conductive Filament
Typically, antennas produced with 3D printing technology require treatment after printing, such as the metallization of their structure. Given the increasing advancement in technology and the widespread use of mobile communications, it becomes necessary to develop antennas that are easy and fast to produce in mass. Therefore, a possible solution would be to produce an antenna entirely with a conductive filament.
The material used to produce the designed horn structure exclusively with the 3D printing process was the Electrifi, a conductive filament from Multi3D [34]. As mentioned in [34], this material is considered as the most conductive filament available on the market and has an electrical conductivity of σ = 1.67 × 10 4 S/m, making its usage highly promising for the design of radiating structures for mmWaves.
To confirm the proper conductivity of Electrifi, it was characterized. A line with dimensions 0.2 cm × 10 cm × 0.2 cm was printed, as shown in Figure 4. Epoxy EPO-TEK H20E from Epoxy technology [35] was used to ensure the electrical contact between the multimeter terminals and the printed part.

Horn Antennas Built from Conductive Filament
Typically, antennas produced with 3D printing technology require treatment after printing, such as the metallization of their structure. Given the increasing advancement in technology and the widespread use of mobile communications, it becomes necessary to develop antennas that are easy and fast to produce in mass. Therefore, a possible solution would be to produce an antenna entirely with a conductive filament.
The material used to produce the designed horn structure exclusively with the 3D printing process was the Electrifi, a conductive filament from Multi3D [34]. As mentioned in [34], this material is considered as the most conductive filament available on the market and has an electrical conductivity of σ = 1.67 × 10 4 S/m, making its usage highly promising for the design of radiating structures for mmWaves.
To confirm the proper conductivity of Electrifi, it was characterized. A line with dimensions 0.2 cm × 10 cm × 0.2 cm was printed, as shown in Figure 4. Epoxy EPO-TEK H20E from Epoxy technology [35] was used to ensure the electrical contact between the multimeter terminals and the printed part. Several measurements of the material DC resistance were performed, and the values obtained are shown in Table 2. Using (1) and (2), where R is the resistance of the material, ρ is the resistivity, σ is the conductivity, l represents the line length, and A is the respective section, it was possible to obtain an estimation of the conductivity value of σ = 2.22 × 10 3 S/m. The difference between this value and the one mentioned by the manufacturer may be because this resistance is of a printed piece and not of the material itself, as mentioned in [34]. Table 3 holds all the estimated properties of the referred material, which were considered on the antenna design process.   Several measurements of the material DC resistance were performed, and the values obtained are shown in Table 2. Using (1) and (2), where R is the resistance of the material, ρ is the resistivity, σ is the conductivity, l represents the line length, and A is the respective section, it was possible to obtain an estimation of the conductivity value of σ = 2.22 × 10 3 S/m. The difference between this value and the one mentioned by the manufacturer may be because this resistance is of a printed piece and not of the material itself, as mentioned in [34]. Table 3 holds all the estimated properties of the referred material, which were considered on the antenna design process. Despite the Electrifi supplier suggests the most suitable printing settings for this material, for high frequencies, the printing process can be a challenging task since the structure to be printed has a small size. After some manufacturing tests, the printing parameters have been set, and are presented in Table 4. It should be noted that this prototype was printed in vase mode. This mode consists of a constant printing, in which the nozzle goes through layer by layer in a spiral form as if forming a vase. This solution not only allowed the use of less material but significantly optimized the printing time, with the prototype being printed in approximately 12 min.
The prototype printed with Electrifi has a wall thickness of 0.4 mm. The final dimensions, listed in Table 5 and Figure 5, show the antenna built using the conductive filament. Despite the Electrifi supplier suggests the most suitable printing settings for terial, for high frequencies, the printing process can be a challenging task since t ture to be printed has a small size. After some manufacturing tests, the printing ters have been set, and are presented in Table 4. It should be noted that this prototype was printed in vase mode. This mode of a constant printing, in which the nozzle goes through layer by layer in a spira if forming a vase. This solution not only allowed the use of less material but sign optimized the printing time, with the prototype being printed in approximately The prototype printed with Electrifi has a wall thickness of 0.4 mm. The fina sions, listed in Table 5 and Figure 5, show the antenna built using the conductive

Results
The constructed prototypes using the different techniques to metallize the s the antenna were tested and measured, and their results were compared with obtained from the simulated models. Measurements were performed with a V operates up to 67 GHz (E8361C by Agilent Technologies, USA) and two referen nas (LB-180400-KF 18-40 GHz Broadband Horn Antenna), for radiation pattern a ensuring the antennas in the far-field. Figure 6 shows the comparison between the simulated and measured reflect ficient of the copper-taped horn antenna. It is possible to observe that the measu simulated results are clearly in agreement, and the antenna is properly matched, minimum value of S11 appearing at 28 GHz.

Results
The constructed prototypes using the different techniques to metallize the surface of the antenna were tested and measured, and their results were compared with the ones obtained from the simulated models. Measurements were performed with a VNA that operates up to 67 GHz (E8361C by Agilent Technologies, Santa Clara, CA, USA) and two reference antennas (LB-180400-KF 18-40 GHz Broadband Horn Antenna), for radiation pattern and gain, ensuring the antennas in the far-field. Figure 6 shows the comparison between the simulated and measured reflection coefficient of the copper-taped horn antenna. It is possible to observe that the measured and simulated results are clearly in agreement, and the antenna is properly matched, with the minimum value of S 11 appearing at 28 GHz.  Considering the S11 < −10 dB criteria, the antenna exhibits a measured bandw 7.19 GHz (26.0%). Comparing this result with those simulated, 2.93 GHz (10.7% provement is observed. Figure 7 shows the comparison between the simulated and measured radiat terns of the horn antenna using copper tape, for the two main radiation planes and ϕ = 90°. It is possible to observe that the measurement results are close to the simu This antenna exhibits a measured half power beamwidth (HPBW) of 38° in the p 90°, while in the orthogonal plane (ϕ = 90°) is about 33°. The copper-taped horn has a simulated gain of 12 dBi and a measured gain of 11.2 dBi, at 28 GHz, wh clearly satisfying result.

Horn Antennas with Copper Paint
Using this method, two approaches were tested with effect mainly on the which are with or without sanding of the 3D structure surface.

•
Without sanding process: Three prototypes were built using a different number of layers of conductive Considering the S 11 < −10 dB criteria, the antenna exhibits a measured bandwidth of 7.19 GHz (26.0%). Comparing this result with those simulated, 2.93 GHz (10.7%), an improvement is observed. Figure 7 shows the comparison between the simulated and measured radiation patterns of the horn antenna using copper tape, for the two main radiation planes, θ = 90 • and φ = 90 • . Considering the S11 < −10 dB criteria, the antenna exhibits a measured bandwidth of 7.19 GHz (26.0%). Comparing this result with those simulated, 2.93 GHz (10.7%), an improvement is observed. Figure 7 shows the comparison between the simulated and measured radiation patterns of the horn antenna using copper tape, for the two main radiation planes, θ = 90° and ϕ = 90°. It is possible to observe that the measurement results are close to the simulations. This antenna exhibits a measured half power beamwidth (HPBW) of 38° in the plane θ = 90°, while in the orthogonal plane (ϕ = 90°) is about 33°. The copper-taped horn antenna has a simulated gain of 12 dBi and a measured gain of 11.2 dBi, at 28 GHz, which is a clearly satisfying result.

Horn Antennas with Copper Paint
Using this method, two approaches were tested with effect mainly on the finishes, which are with or without sanding of the 3D structure surface.

•
Without sanding process: Three prototypes were built using a different number of layers of conductive ink and characterized by their main parameters. Figure 8 represents the comparison of the simulated and measured reflection coefficient of the painted antennas without the sanding process. It is possible to observe that the measurement results are close to the simulations. This antenna exhibits a measured half power beamwidth (HPBW) of 38 • in the plane θ = 90 • , while in the orthogonal plane (φ = 90 • ) is about 33 • . The copper-taped horn antenna has a simulated gain of 12 dBi and a measured gain of 11.2 dBi, at 28 GHz, which is a clearly satisfying result.

Horn Antennas with Copper Paint
Using this method, two approaches were tested with effect mainly on the finishes, which are with or without sanding of the 3D structure surface.
• Without sanding process: Three prototypes were built using a different number of layers of conductive ink and characterized by their main parameters. Figure 8 represents the comparison of the simulated and measured reflection coefficient of the painted antennas without the sanding process. Analyzing these results, it can be observed that all prototypes witho process are well matched, and the minimum value of S11 happens at 28 possible to observe that by increasing the number of layers, the antenna ado closer to the simulations, improving even more its matching. Moreover, with four layers presents roughly 7.13 GHz of bandwidth, from 24.84 GH GHz, making this a very promising result. Figure 9 shows the comparison of the simulated and the measured no ation pattern of the antenna coated with four layers of spray-paint without According to Figure 9, a good agreement between both curves is notic tenna has a HPBW of 34.2° and 29° in the planes θ = 90° and ϕ = 90°, respe measured gain of 10.8 dBi at 28 GHz, and reminding that the simulated val which is a satisfactory result.

•
With sanding process: To evaluate the roughness impact of this manufacturing process on Analyzing these results, it can be observed that all prototypes without the sanding process are well matched, and the minimum value of S 11 happens at 28 GHz. It is also possible to observe that by increasing the number of layers, the antenna adopts a behavior closer to the simulations, improving even more its matching. Moreover, the prototype with four layers presents roughly 7.13 GHz of bandwidth, from 24.84 GHz up to 31.97 GHz, making this a very promising result. Figure 9 shows the comparison of the simulated and the measured normalized radiation pattern of the antenna coated with four layers of spray-paint without sanding. Analyzing these results, it can be observed that all prototypes without the sanding process are well matched, and the minimum value of S11 happens at 28 GHz. It is also possible to observe that by increasing the number of layers, the antenna adopts a behavior closer to the simulations, improving even more its matching. Moreover, the prototype with four layers presents roughly 7.13 GHz of bandwidth, from 24.84 GHz up to 31.97 GHz, making this a very promising result. Figure 9 shows the comparison of the simulated and the measured normalized radiation pattern of the antenna coated with four layers of spray-paint without sanding. According to Figure 9, a good agreement between both curves is noticeable. The antenna has a HPBW of 34.2° and 29° in the planes θ = 90° and ϕ = 90°, respectively, with a measured gain of 10.8 dBi at 28 GHz, and reminding that the simulated value was 12 dBi, which is a satisfactory result.

•
With sanding process: To evaluate the roughness impact of this manufacturing process on the antenna's characteristics, the two painted prototypes were compared applying the sanding process in the structure. Figure 10 presents the comparison between the simulated and measured reflection coefficient of the two and three-layer painted antenna after a hand-made sanding process, which is not accurate.
Although both prototypes that were submitted to the sanding process exhibit similar behavior to the simulated antenna, there is a slight deviation from the resonant frequency. This may be due to the sanding process, since small variations in the antenna's structure influence its performance. To compensate this effect, it is suggested to superficially increase the external antenna's dimensions. According to Figure 9, a good agreement between both curves is noticeable. The antenna has a HPBW of 34.2 • and 29 • in the planes θ = 90 • and φ = 90 • , respectively, with a measured gain of 10.8 dBi at 28 GHz, and reminding that the simulated value was 12 dBi, which is a satisfactory result.
• With sanding process: To evaluate the roughness impact of this manufacturing process on the antenna's characteristics, the two painted prototypes were compared applying the sanding process in the structure. Figure 10 presents the comparison between the simulated and measured reflection coefficient of the two and three-layer painted antenna after a hand-made sanding process, which is not accurate. Figure 10. Comparison between simulated and measured reflection coefficient of the fabr horn antennas using conductive paint, after applying sanding process.
The sanded antenna covered with three layers has a bandwidth of 2.86 GHz clearly close to the 2.93 GHz of simulations. The measurements of the two main r planes of this prototype were performed and are shown in Figure 11. It can be observed, and in agreement with the other antennas, both radiation are similar. Based on the high directivity observed and the gain of 11.8 dBi, these isfactory results for a 3D-printed horn antenna. The measured HPBW was 26.7° in the planes θ = 90° and ϕ = 90°, respectively.

Horn Antennas with Conductive Filament
Regarding the 3D-printed horn antenna using the conductive material, the c son between the simulated and measured reflection coefficient is shown in Figur   Figure 10. Comparison between simulated and measured reflection coefficient of the fabricated horn antennas using conductive paint, after applying sanding process.
Although both prototypes that were submitted to the sanding process exhibit similar behavior to the simulated antenna, there is a slight deviation from the resonant frequency. This may be due to the sanding process, since small variations in the antenna's structure influence its performance. To compensate this effect, it is suggested to superficially increase the external antenna's dimensions.
The sanded antenna covered with three layers has a bandwidth of 2.86 GHz (10.2%) clearly close to the 2.93 GHz of simulations. The measurements of the two main radiation planes of this prototype were performed and are shown in Figure 11. The sanded antenna covered with three layers has a bandwidth of 2.86 GHz (10.2%) clearly close to the 2.93 GHz of simulations. The measurements of the two main radiation planes of this prototype were performed and are shown in Figure 11. It can be observed, and in agreement with the other antennas, both radiation patterns are similar. Based on the high directivity observed and the gain of 11.8 dBi, these are satisfactory results for a 3D-printed horn antenna. The measured HPBW was 26.7° and 27° in the planes θ = 90° and ϕ = 90°, respectively.

Horn Antennas with Conductive Filament
Regarding the 3D-printed horn antenna using the conductive material, the comparison between the simulated and measured reflection coefficient is shown in Figure 12. It can be observed, and in agreement with the other antennas, both radiation patterns are similar. Based on the high directivity observed and the gain of 11.8 dBi, these are satisfactory results for a 3D-printed horn antenna. The measured HPBW was 26.7 • and 27 • in the planes θ = 90 • and φ = 90 • , respectively.

Horn Antennas with Conductive Filament
Regarding the 3D-printed horn antenna using the conductive material, the comparison between the simulated and measured reflection coefficient is shown in Figure 12.

Horn Antennas with Conductive Filament
Regarding the 3D-printed horn antenna using the conductive materia son between the simulated and measured reflection coefficient is shown in Figure 12. Comparison between simulated and measured reflection coefficient of th horn antenna using conductive filament. Although the minimum value of S 11 occurs at 29.76 GHz, a reasonable level of impedance matching was obtained since for the desired operating frequency, the practical S 11 is 16.02 dB. Regarding the obtained bandwidth, a significant improvement was verified. While in simulation a bandwidth of 3.55 GHz was found, the measured bandwidth of 16.52 GHz (58.6%) was accomplished.
Lastly, the normalized radiation pattern of this prototype is presented in Figure 13.
Although the minimum value of S11 occurs at 29.76 GHz, a reasonable level of impedance matching was obtained since for the desired operating frequency, the practical S11 is 16.02 dB. Regarding the obtained bandwidth, a significant improvement was verified. While in simulation a bandwidth of 3.55 GHz was found, the measured bandwidth of 16.52 GHz (58.6%) was accomplished.
Lastly, the normalized radiation pattern of this prototype is presented in Figure 13. It is possible to notice that the measured radiation pattern presents an agreement compared with the obtained through simulation, mainly in boresight. Furthermore, the 12 dBi of measured gain proves that this material is suitable for this operating frequency, allowing its implementation in antenna structures for emerging communications systems. In the plane θ = 90°, the simulated HPBW was 27° while its measured value was 38.4°. Regarding the plane ϕ = 90°, the simulated HPBW was 32° while 31.5° was measured.

Additional Results
A comparison of the simulated and measured gain over the frequency was performed, for the PLA structure covered with copper and for the Electrifi structure. These results are presented in Figures 14 and 15, respectively. It is possible to notice that the measured radiation pattern presents an agreement compared with the obtained through simulation, mainly in boresight. Furthermore, the 12 dBi of measured gain proves that this material is suitable for this operating frequency, allowing its implementation in antenna structures for emerging communications systems. In the plane θ = 90 • , the simulated HPBW was 27 • while its measured value was 38.4 • . Regarding the plane φ = 90 • , the simulated HPBW was 32 • while 31.5 • was measured.

Additional Results
A comparison of the simulated and measured gain over the frequency was performed, for the PLA structure covered with copper and for the Electrifi structure. These results are presented in Figures 14 and 15, respectively.

Additional Results
A comparison of the simulated and measured gain over the frequency was per formed, for the PLA structure covered with copper and for the Electrifi structure. Thes results are presented in Figures 14 and 15, respectively.   According to Figure 14 it is possible to confirm that the gain as a function of the fre quency of the three metallized antennas is similar and agrees with the simulated results Observing the Figure 15 it is possible to evidence the electrical conductivity of the Electrif filament and the proper performance for this operating frequency. Figure 16 shows the simulated total efficiency over the frequency of both antennas. It is possible to observe that the horn antenna printed with PLA and covered wit copper has a maximum efficiency of 95.6% at the desired frequency, 28 GHz. On the othe hand, the prototype fabricated with the Electrifi material presents 90.0% of total efficienc and although this result was obtained in a simulation environment, it is a promising one especially given the novelty associated with the conductive filaments.

Discussion
3D printing is a common technique to all antennas produced, nevertheless, differen According to Figure 14 it is possible to confirm that the gain as a function of the frequency of the three metallized antennas is similar and agrees with the simulated results. Observing the Figure 15 it is possible to evidence the electrical conductivity of the Electrifi filament and the proper performance for this operating frequency. Figure 16 shows the simulated total efficiency over the frequency of both antennas. According to Figure 14 it is possible to confirm that the gain as a function of the frequency of the three metallized antennas is similar and agrees with the simulated results. Observing the Figure 15 it is possible to evidence the electrical conductivity of the Electrifi filament and the proper performance for this operating frequency. Figure 16 shows the simulated total efficiency over the frequency of both antennas. It is possible to observe that the horn antenna printed with PLA and covered with copper has a maximum efficiency of 95.6% at the desired frequency, 28 GHz. On the other hand, the prototype fabricated with the Electrifi material presents 90.0% of total efficiency and although this result was obtained in a simulation environment, it is a promising one, especially given the novelty associated with the conductive filaments.

Discussion
3D printing is a common technique to all antennas produced, nevertheless, different approaches were followed in this work. In all prototypes, the results obtained were satisfactory and quite promising bearing in mind the requirements of the next generation of mobile communications. It is possible to observe that the horn antenna printed with PLA and covered with copper has a maximum efficiency of 95.6% at the desired frequency, 28 GHz. On the other hand, the prototype fabricated with the Electrifi material presents 90.0% of total efficiency and although this result was obtained in a simulation environment, it is a promising one, especially given the novelty associated with the conductive filaments.

Discussion
3D printing is a common technique to all antennas produced, nevertheless, different approaches were followed in this work. In all prototypes, the results obtained were satisfactory and quite promising bearing in mind the requirements of the next generation of mobile communications.
The production cost was estimated considering the percentage of material used in the production of the prototype relative to the total cost of the materials used in the structure of each prototype, without taking into account the cost of the connector, confirming the advantage of 3D printing in the reduction of manufacturing cost when compared with conventional techniques. Table 6 summarizes the main measured results of the developed prototypes, along with their estimated production cost and total weight. It is possible to verify that the first two prototypes present very similar values, both in bandwidth and in measured gain.
The minor differences between both results, we believe are due to the behavior of both coating layers, their losses, and their interaction with the base PLA structure. While the copper strip has a homogeneous structure on the PLA, being unaffected by the possible imperfections in the PLA, such as pores or small roughness, the PLA structure covered with paint does not guarantee a completely homogeneous structure, without pores or with a uniform layer.
On the other hand, and despite the acceptable gain value, the antenna that was submitted to the sanding process presents a bandwidth lower than all the other prototypes, but it is the closest to the simulation. Finally, the antenna produced with the conductive filament has a considerably higher bandwidth than the other prototypes.
Concerning production costs, the antenna with the highest cost is the last prototype made of conductive filament due to the initial cost of the material used. In the context of the available 3D filaments, and more specially in the electrically conductive ones, Electrifi is truly revolutionary and thus its higher price. Nevertheless, the overall cost of the antenna production remains low.
All prototypes built have reduced weight and costs compared to a metal-based antenna with similar characteristics, weighing at least ten times more and being sold with cost of approximately 1500 euros.

Conclusions
3D printing is considered to be a technology with greater change for development and integration in emerging communication systems. This technology has several advantages over conventional part-making techniques, with rapid prototyping and low cost being the most notable. In this article, different techniques for creating metal antennas with the aid of 3D printing were presented and applied to the typical case of a horn antenna. A total of seven samples were produced, and their main characteristics were analyzed, combining the low-cost manufacturing, the simplicity of the structure with the positive results achieved. The sanded antenna coated with conductive paint has a lower bandwidth despite having a higher gain, when compared to the prototype using copper tape and that without the sanding process. On the other hand, although the antenna using conductive filament is the one with higher production costs, it has in fact not only a better gain, but also a significantly improved bandwidth. It also has the advantage that no processing is required after printing.
Overall, the results are promising and point to 3D printing technology as a good choice for communication applications to provide a possible cost-effective solution for commercial applications. Funding: This work is funded by FCT/MCTES through national funds and when applicable cofunded EU funds under the project UIDB/50008/2020-UIDP/50008/2020.