Metal-only additive manufacturing of V-band lightweight waveguide and horn components

IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES 1 Metal-Only Additive Manufacturing of V-Band Lightweight Waveguide and Horn Components Mingzheng Chen , Graduate Student Member, IEEE, José Rico-Fernández , Graduate Student Member, IEEE, Francisco Mesa , Fellow, IEEE, and Oscar Quevedo-Teruel , Fellow, IEEE Abstract— Additive manufacturing (AM) is growing as a key technology for the miniaturization and integration of microwave components. Among several AM processes, laser powder-bed fusion (LPBF) is especially convenient for waveguides and horns because it allows for 3-D printing of metal-only parts with high accuracy and low surface roughness. During the 3-D printing process, metallic powder materials are selectively consolidated by melting layer by layer together using a heat source, that is, a laser, allowing for adaptation to complex and shaped structures. The present study investigates the feasibility of using LPBF-AM to fabricate waveguides and geodesic H-plane horns in V-band. Geodesic H-plane horns comprise two parallel curved metallic plates to reduce phase errors and achieve high gain and aperture efficiency, and are particularly suited to be fabricated using LPBF. The monolithic waveguides demonstrate significantly better performance than their two-piece counterparts, achieving an average attenuation coefficient of 5.3 dB/m for the straight waveguide and 8 dB/m for the 90◦E-plane waveguide bend at 50–70 GHz. Similarly, the monolithic geodesic H-plane horn has higher realized gains and radiation efficiencies with only a sixth of the weight compared to the two-piece version at 52–68 GHz. The results demonstrate that the LPBF-AM technique is a promising candidate to produce monolithic metal-only microwave components in V-band. Index Terms— 3-D printing, additive manufacturing (AM), geodesic H-plane horns, laser powder-bed fusion (LPBF), metalonly, V-band, waveguides. I. INTRODUCTION ADDITIVE manufacturing (AM) techniques, widely referred to as 3-D printing, have transformed the fabrication of everyday items and, more recently, many microwave Manuscript received 25 July 2024; accepted 24 August 2024. The work of Francisco Mesa was supported in part by MICIU/AEI/10.13039/501100011033 under Grant PID2020-116739GB- I00. The work of Oscar Quevedo-Teruel was supported by Vetenskapsrådet (VR) Project under the call “Research Project Grant within Natural and Engineering Sciences” under Grant 2022-03865. (Corresponding author: Mingzheng Chen.) Mingzheng Chen and Oscar Quevedo-Teruel are with the Division of Electromagnetic Engineering and Fusion Science, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden (e-mail: [email protected]; [email protected]). José Rico-Fernández is with Northern Waves AB, 115 43 Stockholm, Sweden (e-mail: [email protected]). Francisco Mesa is with the Department of Applied Physics 1, ETS Ingeniería Informática, Universidad de Sevilla, 41012 Seville, Spain (e-mail: [email protected]). Color versions of one or more figures in this article are available at https://doi.org/10.1109/TMTT.2024.3451065. Digital Object Identifier 10.1109/TMTT.2024.3451065 components. In this manufacturing technique, layers of material that are deposited in a controlled manner are used to construct prototypes. AM is applicable to several materials, such as polymers, ceramics, and even metals. Compared to conventional subtractive manufacturing methods, such as computerized numerical control (CNC) machining, AM offers a cost-effective alternative for the fabrication of microwave components featuring complex and intrinsic shapes with reduced volume, weight, and lead time [1]. Thus, this technique has recently been used to fabricate waveguides [2],[3],[4],[5], [6], filters [7],[8], orthomode transducers [9],[10],[11], horns [12],[13],[14],[15],[16], metasurfaces [17], and array antennas [18],[19],[20]. When applied to microwave component manufacturing, there are generally two major categories of AM techniques, depending on whether metal is the only material used. In the first category of AM techniques, components are fabricated by first 3-D printing the base body using polymer (or equivalent materials) and then applying surface metallization using either conductive spray or electroless plating. The most commonly used two-step AM techniques are fused deposition modeling (FDM) [2], stereolithography (SLA) [7],[9],[13],[21], and selective laser sintering (SLS) [6],[8]. The main advantage of this category is that the manufactured prototypes are lighter than their all-metal versions. However, plastic skeletons raise concerns about mechanical robustness, power handling capability, and suitability for space environments due to outgassing. In the other category of AM techniques, only metal is used as the primary material for fabrication, which is normally referred to as selective laser melting (SLM) [3],[12] or direct metal laser sintering (DMLS) [4],[18],[19]. The metal-only AM techniques enable microwave components with complex geometries to be monolithically 3-D printed using metal and do not need a further coating procedure. Compared to CNC machining, the metal-only AM techniques relax interfaces and assembly designs, resulting in compact and lightweight monolithic prototypes without any screws. Moreover, metalonly components possess superior mechanical and thermal properties, making them highly desirable for space and highpower applications. The metal-only AM technique used in this work is laser powder-bed fusion (LPBF). This technique involves the use of a recoater to apply uniform metal powder layers with thicknesses of approximately 30 µm on the printing bed. A laser is then used to selectively melt the metal powder © 2024 The Authors. This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 2 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES in each layer until the printing is complete. As reported in [22],[23], and [24], the application of the LPBF-AM technique to lens antennas with curved profiles results in a substantial decrease in antenna weight while maintaining strong mechanical and RF capabilities. However, the performance of this technique has not been thoroughly investigated using standard rectangular waveguides, especially the attenuation performance. This information is very important for engineers to select a proper manufacturing technique for their prototypes, especially at high frequencies, such as V-band. In the present work, we evaluate the applicability of the LPBF-AM technique to the manufacture of lightweight waveguide and geodesic H-plane horn components in V-band. First and for comparison purposes, we manufactured four standard WR-15 waveguides in the frequency band 50–70 GHz, including monolithic straight waveguide, monolithic 90◦E-plane waveguide bend, two-piece straight waveguide without electromagnetic band gap (EBG) structures, and two-piece straight waveguide with EBG structures. The measured results demonstrate that the monolithic waveguides have much better performance in terms of attenuation coefficients, reflection coefficients, and weight. To demonstrate the advantage of LPBF-AM for more complex components, we further apply this technique to manufacture V-band geodesic H-plane horns, which have complex and curved profiles that are difficult for traditional CNC processes. Geodesic H-plane horns use the height profile of the horn to control the path lengths of waves that reach the aperture, thus correcting for phase errors and increasing directivity [25]. Similarly, we designed and manufactured two horns, monolithic and two-piece, that operate in the frequency band 52–68 GHz. The performance of the two designs is compared, including operating bandwidth, realized gain, radiation efficiency, robustness, and antenna weight. We show that the monolithic geodesic H-plane horn is a better design in all aspects in V-band. The article is organized as follows. In Section II, the LPBF-AM process is described and the application to waveguides in V-band is discussed, including detailed comparisons between the prototypes and with other AM techniques. In Section III, we discuss the manufacturing and compare the performance of two geodesic H-plane horns. Finally, some conclusions are drawn in Section IV. II. AM OF WAVEGUIDE COMPONENTS A. LPBF-AM Process In this section, we outline the manufacturing process of waveguide and horn components using the LPBF-AM technique. A basic scheme of the AM process is illustrated in Fig. 1. Taking into account the electromagnetic and mechanical properties, we selected AlSi10Mg aluminum alloy as the printing material in the LPBF-AM process. This material has an electrical conductivity of σ=1.68 ×107S/m [26] and an average surface roughness of Ra =8−10 µm without any postprocessing. First, a uniform layer of AlSi10Mg aluminum alloy powder with a thickness of approximately 30 µm is spread on the building plate using a recoater. A laser is then used to fuse the first cross section of the model. In the next Fig. 1. Illustration of the AM process for a prototype using the LPBF technique. Fig. 2. Photographs of (a) monolithic straight waveguide, (b) monolithic 90◦ E-plane waveguide bend, (c) two-piece straight waveguide without EBG, and (d) two-piece straight waveguide with EBG. step, a new layer of powder is spread over the previous layer using the recoater, and further cross sections are fused and added. Finally, the process repeats until the entire model is created, with the unfused powder removed during postprocessing. Postprocessing includes removing manufacturing supports, clearing the prototype, and sandblasting with glass microspheres to reduce the root mean squared (rms) surface roughness Ra. B. LPBF-AM Waveguides To evaluate the potential benefits of the LPBF-AM process in a straightforward way, we begin our study with the four standard WR-15 waveguides shown in Fig. 2. The monolithic straight waveguide in Fig. 2(a) and the monolithic 90◦E-plane waveguide bend in Fig. 2(b) are printed as a single piece. The two straight waveguides in Fig. 2(c) and (d) comprise two separate bottom and top pieces, which are individually 3-D printed and then assembled together using screws. The advantage of the two-piece prototypes lies in that we can polish the internal surfaces of the waveguides to reduce the surface roughness, hence reducing insertion losses. Due to the presence of manufacturing and assembly errors, achieving perfect contact between the two waveguide pieces is challenging. Therefore, it is essential to incorporate an EBG structure to effectively minimize leakage in the side sections of the waveguides [27],[28]. In contrast, the monolithic prototypes avoid misalignment errors and leakage between the two plates, thus ensuring robust RF performance. The manufacturing and measurement of the four waveguides are subjected to the following analysis. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. CHEN et al.: METAL-ONLY AM OF V-BAND LIGHTWEIGHT WAVEGUIDE AND HORN COMPONENTS 3 Fig. 3. Photographs of the opened two-piece straight waveguides. (a) Without EBG and (b) With EBG. Fig. 4. Dispersion diagram of the Brillouin zone 0X M0for the glide-symmetric EBG unit cell with the stopband and operational band (shaded in colors) indicated. 1) Attenuation performance of the waveguide components manufactured by the LPBF-AM technique in the V- band considering both the straight and bent waveguide scenario. 2) Effect of leakage mitigation using glide-symmetric EBG holes in the two-piece waveguides. 3) Overall performance of the monolithic and two-piece waveguides in terms of attenuation and reflection coefficient, mechanical robustness, and weight. The three straight waveguides have a length of 33 mm while the waveguide bend is 31.8 mm long. The cross section of the waveguides is 3.76 ×1.88 mm2, corresponding to the dimensions of the standard waveguide WR-15. The photographs of the inner parts for the two-piece waveguides are presented in Fig. 3, where it is evident that the EBG waveguide illustrated in Fig. 3(b) is broader compared to the non-EBG waveguide to accommodate the glide-symmetric holes responsible for the EBG effect. To characterize the stopband performance of the EBG structure, we show in Fig. 4 the complete dispersion diagram of the Brillouin zone 0X M0 of the corresponding unit cell assuming a 30 µm gap between the two pieces, with the Brillouin zone’s boundaries illustrated in the inset. The periodicity of the EBG is p=2.9 mm, the diameter of the hole d=2.3 mm, and the depth of the hole h=1 mm. We can observe in Fig. 4that the stopband (colored blue) expands from 44 to 93 GHz, fully covering the operational band (colored orange) of 50–70 GHz. The measured reflection coefficients of the four waveguides are presented in Fig. 5. It can be seen that the reflection coefficient for the two monolithic waveguides remain below −30 dB across nearly the whole frequency range. In contrast, the Fig. 5. Measured reflection coefficients of the monolithic straight waveguide (red curve), the monolithic 90◦E-plane waveguide bend (orange curve), the two-piece straight waveguide with EBG (dark blue curve), and the two-piece straight waveguide without EBG (light blue curve). Fig. 6. (a) Measured and (b) simulated transmission coefficients of the monolithic straight waveguide (red curves), the monolithic 90◦E-plane waveguide bend (orange curves), the two-piece straight waveguide with EBG (dark blue curves), and the two-piece straight waveguide without EBG (light blue curves). coefficients for the two-piece waveguides are barely better than −25 dB. This is attributed to the small air gap between the plates and the inevitable assembly errors that slightly worsen the impedance matching between the waveguides and the standard WR-15 adapter used for measurement. The measured and simulated transmission coefficients of the four waveguides are depicted in Fig. 6(a) and (b), respectively. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 4 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES Fig. 7. Measured attenuation coefficients of the monolithic straight waveguide (red curve), the monolithic 90◦E-plane waveguide bend (orange curve), the two-piece straight waveguide with EBG (dark blue curve), and the two-piece straight waveguide without EBG (light blue curve). The average measured transmission coefficient (from which we can infer the insertion loss) for the monolithic straight waveguide and the waveguide bend are −0.17 and −0.28 dB, respectively, from 50 to 70 GHz. The measured data are in good agreement with the simulated results when Ra is assumed to be 1 µm for the straight waveguide and 2 µm for the waveguide bend in CST. However, it has been reported that the surface roughness model implemented in CST is only valid for normally distributed surfaces at low frequencies [6]. For more complex surfaces, the comparison with CST data could not reflect the real surface roughness of the prototypes, especially in the V-band and above. Therefore, the comparison with CST regarding surface roughness is provided only as a reference point. In Fig. 6(a), it can be seen that the implementation of glide-symmetric holey EBG structures improves the measured transmission coefficient of the two-piece waveguides from below −1 dB to an average of −0.4 dB across the entire frequency band. This improvement can also be seen from the simulated results in Fig. 6(b) when the gap between pieces is assumed to be 30 µm. The differences between the simulated and measured data in the two-piece waveguides are mainly attributed to the difficulty of maintaining a consistent air gap during the assembly process. As the straight waveguides and the waveguide bend have a slightly different length, a fairer comparison is presented in Fig. 7for the attenuation coefficient per unit length. This coefficient αconsiders both the measured S-parameters and the length of the waveguides laccording to [29] α(dB/m) = −10 llog10|S21S12| 1− |S11S22|.(1) The curves plotted in Fig. 7show that the attenuation coefficients of the four waveguides in general decrease as the frequency increases, which is typical for rectangular waveguides. The monolithic straight waveguide has an average α≈5.3 dB/m while this average value for the waveguide bend is approximately 8 dB/m. This difference is due to the fact that the monolithic straight waveguide is 3-D printed vertically along the waveguide direction, but the waveguide bend needs to be printed with a tilted angle, and this tilt TABLE I PERFORMANCE COMPARISON OF LPBF-AM WAVEGUIDES TABLE II COMPARISON OF ATTENUATION COEFFICIENT FOR ADDITIVE MANUFACTURED WAVEGUIDES increases the surface roughness. Moreover, it is clear that even though the EBG structures could significantly improve the attenuation coefficient of the two-piece waveguides from an average 36.8 to 12 dB/m, it is still twice worse than the monolithic one. As a final analysis, the overall performance of the four manufactured waveguides are compared in Table Iin terms of the attenuation and reflection coefficient and their weight. In general, the monolithic prototypes are much better than the two-piece designs in all aspects. Even with better microwave performance of attenuation and reflection coefficient, the monolithic prototypes have almost half the weight of the two-piece waveguides because less material and no screws for assembly are needed. The waveguide bend is lighter because their metallic walls are thinner than those of the monolithic straight waveguide, which are kept the same as in the two-piece cases for a fair comparison. Additionally, the monolithic prototypes exhibit greater mechanical robustness in extreme environments since they eliminate the risk of screw breakage. In those cases where two-piece designs are necessary, employing EBG structures can considerably lower the attenuation coefficient, although with a marginal increase in weight and volume. The measured attenuation performance for the monolithic waveguides are compared to other AM technologies in Table II. In [3], SLM is used to print straight waveguides in E-band with an average αof 7.5 dB/m. Compared to the Cu-15Sn powder used in [3], the AlSi10Mg aluminum alloy powder used in our LPBF technology is cheaper and lighter. Furthermore, compared to SLA technology in [2] and SLS technology in [6], both of which require printing the nonconductive part initially followed by a metallic coating process, the metal-only waveguides produced with LPBF technology are mechanically more robust and suitable for space and high-power applications. Finally, compared to a recently reported Ka-band 3-D printed waveguide realized using highly conductive fused filament fabrication (FFF) technology, the This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. CHEN et al.: METAL-ONLY AM OF V-BAND LIGHTWEIGHT WAVEGUIDE AND HORN COMPONENTS 5 Fig. 8. Illustration of the two-piece LPBF-AM geodesic H-plane horn. proposed LPBF-AM straight waveguide has a lower attenuation coefficient in a higher frequency band. III. AM OF GEODESIC H-PLANE HORNS It is well established that H-plane sectoral horns utilize a tapering in the H-plane to match electromagnetic waves from a waveguide to open space. To correct the phase error caused by the tapering and enhance antenna directivity, Chen et al. [25] proposed the concept of geodesic H-plane horns and realized them in the Ka-band using CNC machining. In this section, we outline the manufacturing of V-band geodesic H-plane horns using the LPBF-AM technique. The focus of this section is on the AM considerations; hence, details of the horn design are given in the Appendix. Similar to the waveguides studied in Section II, for comparison purposes, we manufactured two prototypes, two-piece, and monolithic. Particular adaptations are considered here to significantly reduce performance impairment resulting from manufacturing and assembly errors for the two prototypes. A. Two-Piece LPBF-AM Geodesic H-Plane Horn The two-piece LPBF-AM prototype is illustrated in Fig. 8, where the transparency of the upper plate is intended for illustrative purposes. To prevent potential leakage caused by imperfect electrical contact in the considered high-frequency range, glide-symmetric EBG holes are positioned on both sides of the horn. The dimensions of the EBG holes are the same as those shown in Fig. 3(b) used in the two-piece straight waveguide. Outside of the EBG holes, extra space is needed to drill holes for the assembly of M2 screws. To ensure good electrical contact in the critical flange region, where the electric field is concentrated, the screw holes in that area are positioned closer together. In addition, to ensure robustness and prevent distortion during the printing procedure, the thickness of the two plates must be increased. Taking into account all of these factors, the ultimate prototype has dimensions of W1=93 mm (width) and H1=17 mm (height). To demonstrate the necessity of using glide-symmetric EBG holes, in Fig. 9, we compare the simulated realized gain at 60 GHz in the beamforming plane (θ=π/2) for the prototype with (w.) and without (w.o.) the EBG structure, assuming a 30 µm gap between the two plates. In the simulation, the antenna material is set as aluminum with a surface roughness of Ra =3µm. We can observe from the figure that the Fig. 9. Simulated realized gain at 60 GHz in the beamforming plane for the two-piece geodesic H-plane horn with (w.) and without (w.o.) the EBG structure, assuming a 30 µm gap between the pieces. It is assumed that the metal is aluminum with a surface roughness of Ra =3µm. Fig. 10. Simulated realized gain versus frequency for the two-piece geodesic H-plane horn with (w.) and without (w.o.) the EBG structure, assuming a 30 µm gap between the pieces. The metal is assumed to be aluminum with a surface roughness Ra =3µm. realized gain deteriorates by around 1 dB without using the EBG holes. Furthermore, to show the wideband performance of the EBG holes, the frequency behaviors of the realized gain for the prototype with and without the EBG structure are compared in Fig. 10. It is clear that the EBG structure improves the realized gain of the prototype by approximately 1 dB throughout the operational band. Therefore, the use of the glide-symmetric holey EBG structure ensures a robust twopiece LPBF-AM geodesic H-plane horn prototype. B. Monolithic LPBF-AM Geodesic H-Plane Horn In addition to the two-piece LPBF-AM design, we propose a monolithic prototype of the geodesic H-plane horn. One major disadvantage of the two-piece design is the requirement for additional space to prevent leakage and facilitate assembly, leading to a bulkier design. This extra space can be minimized in a monolithic prototype, as illustrated in Fig. 11. This prototype requires less processing time in the LPBFAM process, resulting in lower manufacturing costs. Here, the surrounding block is conformal to the geodesic shape of the horn, with a thickness of 1.5 mm to ensure mechanical robustness. A similar thickness was applied to the feed and This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 6 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES Fig. 11. Illustration of the monolithic LPBF-AM geodesic H-plane horn. Fig. 12. Simulated realized gain at 60 GHz in the beamforming plane for the monolithic and two-piece LPBF-AM geodesic H-plane horn prototype, assuming aluminum with a surface roughness of Ra =5 and 3 µm. The gap between the two plates of the two-piece prototype is assumed to be 30 µm. flare area, with two reinforcing ribs added on both sides of the flare to alleviate thermal deformation during the LPBFAM process. After all of the above considerations, the final prototype has a width of W2=78 mm and a flare height of H2=8 mm. Despite the mentioned advantages, the monolithic design has one drawback compared to the two-piece design. It is not possible to open the antenna to polish the inner surfaces (to reduce the surface roughness), resulting in increased insertion losses. In Fig. 12, a comparison is made between the simulated realized gain at 60 GHz in the beamforming plane for the monolithic and the two-piece horn prototype, considering different surface roughness of aluminum. It is observed that even with a higher surface roughness (Ra =5µm compared to 3 µm), the monolithic design achieves a slightly higher realized gain due to the absence of leakage. Note that in the measurements, the performance of the two-piece prototype is expected to degrade further as a result of inevitable misalignment errors in assembly. C. Simulation and Measurement Results To further characterize the performance of the two-piece and monolithic LPBF-AM geodesic H-plane horns, we manufacture, measure, and compare the two prototypes. The results presented in this section include reflection coefficients, radiation patterns, realized gains, aperture efficiencies, radiation efficiencies, and weights. The photographs in Fig. 13(a) and (b) show the prototypes in detail, with the Fig. 13. Photographs of the (a) two-piece and (b) monolithic geodesic H-plane horn prototype. two-piece version opened in two plates to show the internal surfaces. For the manufacture of the upper and lower plates of the two-piece prototype, the chosen printing direction followed the z-axis orientation, as indicated in Fig. 13(a). This means that the flat part of the plates was in contact with the building plate and the geodesic shape was at the top. This direction offers the advantages of avoiding postprocessing on the geodesic surfaces and minimizing the required supports for fabricating the component. On the contrary, for the manufacture of the monolithic prototype, the chosen printing direction followed the x-direction, as indicated in Fig. 13(b) and Fig. 1, positioning the input port flange in contact with the building plate and the flare at the top. This printing direction offers the advantages of maintaining the symmetry of the design with respect to the printing bed and minimizing the required supports for the fabrication of the component. Furthermore, given the profile of the geodesic horn, there is no area that poses an overhanging issue in the inner cavity. After printing the prototypes, we first removed the manufacturing supports and cleaned the horns, minimizing possible printing errors caused by sagging or manufacturing imperfections. In the next step, both prototypes were sandblasted with glass microspheres to uniformly reduce surface roughness inside and outside to Ra =4−6µm. For the two-piece prototype, we finer polished the inner surfaces to further reduce the roughness of the surface to Ra ≈3µm and the EBG areas were machined 500 µm to ensure a good mechanical contact. Subsequently, to properly connect with the WR-15 standard UG385/U of the measurement equipment, the flange areas on both prototype were reduced 500 µm, added in the manufacturing CAD files. Finally, the prototypes were measured in a far-field setup in the anechoic chamber of KTH, as illustrated in Fig. 14(a) and (b). This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. CHEN et al.: METAL-ONLY AM OF V-BAND LIGHTWEIGHT WAVEGUIDE AND HORN COMPONENTS 7 Fig. 14. Measurement setup in the anechoic chamber for the (a) two-piece and (b) monolithic geodesic H-plane horn prototype. Fig. 15. Simulated and measured reflection coefficients of the (a) two-piece and (b) monolithic geodesic H-plane horn prototype. The simulated and measured reflection coefficients for the two-piece and monolithic horns are reported in Fig. 15(a) and (b). To eliminate reflections caused by the adapter, the measurements used the time-gating method. Consequently, the measured data can be directly compared with the simulated results. We can observe that the simulated and measured reflection coefficients of both prototypes are smaller than −15 dB in almost the entire operating frequency range from 52 to 68 GHz. The measured data of the monolithic prototype show better agreement with the simulation results in comparison to the two-piece prototype. This can be attributed to the monolithic AM process, which eliminates the uneven air gap between the antenna pieces and reduces assembling errors. Next, we present the results of the radiation characteristics of the prototypes through measurements, simulations, and numerical analysis. The contour maps of the radiation patterns in the uv-plane at 60 GHz are displayed in Fig. 16. In Fig. 16(a) and (b), the measured and simulated results of the two-piece prototype are plotted, while Fig. 16(c) and (d) show the results of the monolithic prototype. Fig. 16(e) presents Fig. 16. Contour maps of the normalized radiation patterns in the uv-plane at 60 GHz for (a) measurement and (b) simulation of the two-piece geodesic H-plane horn prototype, (c) measurement and (d) simulation of the monolithic geodesic H-plane horn prototype, and (e) RT-PO algorithm, on the same scale. Fig. 17. Comparison of normalized radiation patterns at 60 GHz obtained with various methods for (a) two-piece prototype in the H-plane, (b) monolithic prototype in the H-plane, (c) two-piece prototype in the E-plane, and (d) monolithic prototype in the E-plane. the results obtained using a combined ray-tracing (RT) and physical-optics (POs) algorithm. The RT-PO algorithm utilizes the amplitude and phase of the incident electric field in the horn aperture, obtained from RT, to compute the electric far fields applying the field equivalence principle [30],[31]. Due to the thin aperture shape, the resulting patterns feature a fan-shaped beam, with a narrow beam in the beamforming H- plane and a wide beam in the E-plane. We can observe a good agreement between the results, which confirms the successful design and fabrication of the horn prototypes. A slightly higher sidelobe level (SLL) can be seen in the measured result of the two-piece prototype due to inevitable manufacturing and assembly errors. Also, we can observe that the measured beamwidth in the E-plane of the monolithic prototype is somewhat wider because of the thermal deformation of the horn aperture. However, these deviations are insignificant and the prototypes exhibit excellent performance considering the high operating frequency of 60 GHz. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 8 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES TABLE III COMPARISON WITH SIMILAR HORN AND METALENS DESIGNS Fig. 18. Realized gain patterns of prototypes over frequency in the beamforming H-plane. (a) Measured results of the two-piece prototype. (b) Simulated results of the two-piece prototype with Ra =3µm and a 30 µm gap between the plates. (c) Measured results of the monolithic prototype. (d) Simulated results of the monolithic prototype with Ra =5µm. A more detailed comparison of radiation patterns in the principal H- and E-planes at 60 GHz for both prototypes is shown in Fig. 17. Radiation patterns are provided by the RT-PO model, the CST model, and the measurement. In the beamforming H-plane (θ=π/2), the measured main beams of both prototypes agree very well with the results of the CST simulation and the RT-PO model. The SLL of the two-piece horn prototype deteriorates by approximately 3 dB, while that of the monolithic one aligns well with the full-wave simulation and the numerical calculation. In the E-plane (φ=0), the measured and simulated beamwidths of the two-piece design agree well, both slightly narrower than the numerical prediction due to a slightly enlarged aperture height hfcaused by the small air gap between the two plates (see Fig. 21(c) for an illustration of hf). On the contrary, the measured beamwidth of the monolithic horn is slightly wider due to a reduced hfarising from the slight thermal deformation of the horn aperture. The realized gain patterns of the two prototypes are compared in Fig. 18, including CST simulated and measured data Fig. 19. Realized gain versus frequency of the (a) two-piece and (b) monolithic geodesic H-plane horn prototype. in the operating frequency range 52–68 GHz. The simulated and measured patterns agree well with each other for both horns, demonstrating the wideband properties of the LPBFAM prototypes. Furthermore, both prototypes maintain an SLL lower than −20 dB within the whole working frequency range. Note that the realized gain values of the monolithic design are slightly higher than those of the two-piece design. This fact can be appreciated again in the detailed frequency behaviors of the realized gain of both prototypes in Fig. 19. In this figure, the gray-dashed lines represent the simulated results using a perfect electric conductor (PEC) as the metal material; and the green dashed lines represent the simulated results assuming This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. CHEN et al.: METAL-ONLY AM OF V-BAND LIGHTWEIGHT WAVEGUIDE AND HORN COMPONENTS 9 aluminum with a certain surface roughness. In Fig. 19(a), there is no air gap for the simulation with PEC due to the software convergence problem, while with aluminum it is assumed that it has an air gap of 30 µm between the plates. The measured realized gains are on average higher than those of the simulations assuming the roughness of the inner surface Ra to be 3 µm. A comparable conclusion can be drawn from the examination of both the measurement and the simulation of the monolithic prototype, assuming the surface roughness to be 5 µm in the simulation. The monolithic prototype generally has higher gain values than the two-piece prototype throughout the entire frequency range. Additionally, it is apparent that the measured realized gain of the monolithic horn increases more smoothly as the frequency increases compared to that of the two-piece prototype, indicating a more robust RF performance. Finally, we summarize in Table III the main achievements of the developed monolithic geodesic H-plane horn and compare them with the two-piece design and other horn and metalens antennas. Both geodesic H-plane horns realize a high gain with high aperture and radiation efficiencies at a high frequency compared to other similar designs realized using different fabrication techniques. In addition, the weight of the monolithic horn is only 32 g (a sixth of the two-piece version), which means significant savings in manufacturing time and cost; this is because the main cost of the LPBF-AM process comes from the time needed for the machine to print the prototypes, the power consumed by the machine and the raw powder material. IV. CONCLUSION This paper experimentally demonstrates the applicability of the metal-only LPBF-AM technology to microwave waveguide and horn components in V-band. Compared to two-piece components, monolithic prototypes feature lower attenuation coefficients, lower weight, and better mechanical performance, both in rectangular waveguides and geodesic H-plane horns. The metal-only LPBF-AM technology shows great potential for both industrial production and fast prototyping of high-frequency microwave components. Future work includes applying the technology to other complex components and moving to even higher frequencies. APPENDIX The proposed V-band geodesic H-plane horn antennas include a geodesic H-plane horn, a cascaded feed, and a sinusoidal flare. The design of these components is explained in detail in the following. A. Geodesic H-plane Horn A geodesic H-plane horn has a well-defined triangular geodesic surface that is used to adapt the lengths of the paths followed by the rays emitted from the feed and that arrive at various positions within the aperture [25]. The geodesic surface is defined as the mean surface between the upper and lower parallel conductive plates of the horn. By designing the height profile of the geodesic surface in a suitable manner, we can decrease the differences in path length experienced by Fig. 20. (a) Geometrical parameters and electric field distribution of the geodesic H-plane horn. (b) Height profile along the ρ-axis when φ=0◦. (c) Height profile in terms of φwhen ρ=ρ0. rays within the horn. This, in turn, helps mitigate phase errors in the aperture of the horn. To ensure that only the fundamental mode propagates in the H-plane horn, the separation distance hbetween the two parallel conducting plates should be less than half of the wavelength λ(ideally around λ/4 to allow some margin). In this work, the distance his set to 1 mm for a central frequency of 60 GHz. The geometry of the designed geodesic H-plane horn is shown in Fig. 20(a), where R=75 mm (i.e., 15λat 60 GHz) and φ0=30◦determine the basic triangular area of the horn. Due to this triangular geometry, the path length of the rays is expected to increase monotonically from the center (φ=0◦) to the sides (φ= ±φ0) of the horn aperture. Therefore, to compensate for this variation in path length, we use a ray-tracing (RT) algorithm that provides us with the following optimal height profile [25]: z(ρ, φ) =AhRexp"−ρ−ρ0 η2# ×    cosπφ φ0+1+δ 2  p −δ 2p  (2) where Ah=0.184, ρ0=0.7, η=0.079, p=0.157, and δ=0.01 (δis introduced to avoid a numerical singularity as φ→φ0). The height profile has a Gaussian distribution along the radial direction and a quasi-cosine distribution along the angular direction, as illustrated by the blue and red dashed curves in Fig. 20(a). Specifically, the profile along the ρ-coordinate is shown in Fig. 20(b) when φ=0◦, and the profile along the φ-coordinate is shown in Fig. 20(c) when ρ=ρ0(at this position the height of the horn is This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.