3D-Printed Conformal Array Patch Antenna Using a Five-Axes Motion Printing System and Flash Light Sintering

Abstract

A conformal microstrip antenna is highly integrated with the carrier profile, giving it the advantage of reduced carrier space. However, the processing technology for fabricating an antenna with a conformal surface is complicated and expensive owing to its irregular shape. Therefore, it is particularly important to find ways to reduce the difficulty in producing conformal conductive patterns. For the production of conformal microstrip antennas, a five-axes motion printing system is proposed to print conductive patterns on nonexpanded surfaces. The conductive pattern uses a new boundary-alignment-optimization algorithm in the printing process to improve the accuracy of the boundary of the conductive pattern by optimizing the ink-dropping points in the motion command. The short-time and low-temperature sintering of the conductive pattern by using a flash light enables it to exhibit conductive properties. This method is practically implemented to fabricate a conformal array microstrip antenna. The main and side lobes of the electric and magnetic fields in the directional pattern measured in the microwave darkroom have the same trend as in the simulation diagram. Thus, the effectiveness of this method was verified through simulation analysis and experimental results.

Introduction

In modern wireless communication systems, the surface of the conformal antenna^1–3 can be fused with the surfaces of high-speed carrier platforms, such as aircrafts, missiles, and satellites, without altering the carrier shape. Thus, this field has become a research hotspot. Photolithography technology is mainly used in the manufacturing of a conventional microstrip antenna. This method involves complex processes, such as etching and electroplating, which require long manufacturing time and high costs.^4,5 With the continuous development of piezoelectric inkjet technology, it is possible to directly print metal nanoparticles into a conductive pattern through inkjet printing.^6–9

Recently, the production of radiofrequency-identification patch antenna by using nano copper ink was realized¹⁰; printed nanosilver solutions on liquid crystal polymer substrates are also available to complete the production of dipole antennas.¹¹ It is technically possible to create an antenna through inkjet printing.

At present, multiaxes printing is used to print graphics on curved surfaces. DCS Corporation has developed a cylindrical three-dimensional (3D) relief UV inkjet printer, which is based on the principle that the substrate moves in a circle around the axis, and the nozzle moves in a straight line to achieve inkjet printing. The inkjet printing process is performed at different positions on the object surface. However, the printing object can only be a rotating surface, such as a cylinder, which was printed across 360°¹²; inkjet printing cannot be achieved on nonexpanded surfaces, because of which its application range is small. The 3D inkjet printing method refers to inkjet printing on a curved surface through a complex motion method. The applications of this technology include 3D printers developed by the Japanese Mimaki Engineering team.¹³ The nozzles perform a linear scanning motion, and the substrates perform spatial motion to achieve inkjet printing on spheres and cylinders. The Gazeau team¹⁴ at the University of Poitiers, France, has developed a five-degree-of-freedom robot, in which five-axes printing is used to perform printing on a nonmovable two-dimensional (2D) plane at different angles; however, the function of conformal surface printing has not yet been realized. Thus, these printers and technologies still cannot meet the inkjet printing requirements on complex nonexpanded surfaces. The use of conductive inks in combination with the described inkjet printing method to fabricate electronic circuits on complex surfaces is still in the early stages of research. The new Aerosol Jet (Aerosol Jet) 5X system developed by Optomec in the United States is capable of five-axes coordinated printing motion. During the printing process, the nozzle device and substrate are used to perform linear, curved, and rotational motions in space through multiaxes linkage control technology to realize the printing of conductive patterns on a complex 3D substrate. However, the use of this technology causes the substrate to perform complex motions. The conductive pattern without drying treatment is likely to be affected by the external force on the curved surface, which seriously affects the print quality, reduces conductivity, and affects circuit performance.

Unlike the existing rotating-body printing method, the five-axes coordinated motion can realize nonexpanded surface printing; however, the problem of undried conductive pattern flow cannot be avoided. In this article, a new nonexpanded surface conductive pattern printing method is proposed, which transforms the nonexpanded surfaces into a plurality of 2D triangular patches, converts the 3D printing into 2D printing, and avoids the problem of undried conductive pattern flow. A boundary-alignment-optimization algorithm for ink drop is proposed, which is used to improve the uniformity of the graphics boundary.

Materials and Methods

Nonexpanded surface printing method

In this study, a five-axes motion-printing-and-sintering apparatus was developed for printing and sintering conductive patterns on nonexpanded surfaces, and a conformal microstrip antenna with the required antenna performance was successfully developed, as shown in Figure 1a. The apparatus exhibits XYZ axes of motion, the yaw axis, the roll axis of motion, and constant temperature heating. It also comprises a nozzle and flash light for sintering conductive patterns.

FIG. 1.

(a) Five-axes printing and sintering integrated equipment. (b) The conversion of the nonexpanded surface into a plurality of two-dimensional triangles. (c) The triangular patches are rotated onto a horizontal position through the roll and yaw axes.

While printing conductive patterns by using nanosilver ink on a curved surface, if the printing inclination is extremely high, the undried conductive pattern ink will flow and the printing accuracy will be impaired. In this study, we converted the conductive graphic model into an STL format file. The STL file converts the conductive pattern model into a model consisting of multiple spatial triangular patches, as shown in Figure 1b, and records each of the vertex coordinates of the triangular patches. The normal vector of the triangular patch can be calculated from the three vertex coordinates of the triangle by using Equation (1). \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} n ( {n_x} , {n_y} , {n_z} ) = \left\vert { \begin{matrix} i & j & k \\ {{X_2} - {X_1}} & {{Y_2} - {Y_1}} & {{Z_2} - {Z_1}} \\ {{X_3} - {X_1}} & {{Y_3} - {Y_1}} & {{Z_3} - {Z_1}} \\ \end{matrix} } \right\vert , \tag{1} \end{align*} \end{document}

where n is the normal vector of the triangular patch and X₁, X₂, X₃, Y₁, Y₂, Y₃, Z₁, Z₂, and Z₃ are the spatial coordinates of the three vertices of the triangular patch.

The triangular patch is rotated to the horizontal position based on its obtained normal vector by using the quaternion rotation theory. As shown in Figure 1c, the normal vector of a triangular patch in the conductive pattern is rotated on its roll and yaw axes and then coincides with the space Z axis. At this time, the triangular patch is in the horizontal state, and then the triangular patch is planned to generate a print motion command. Subsequently, the conventional 2D printing method is used to print the conductive pattern by moving the XYZ motion axes using the nozzle. Thus, the 3D complex surface printing is converted into a simple 2D plane printing, thereby reducing the impact of unsintered conductive pattern flow on the print accuracy. After completion of printing, the conductive pattern is sintered at low temperature and high speed by using a flash lamp to obtain a triangular patch with conductive properties. The complete conductive pattern can be obtained after all the triangle patches are printed and sintered.

Boundary-alignment algorithm

The orderliness of the borders is one of the important indicators for measuring the quality of printed graphics. A uniform ink-droplet spacing is used in conventional inkjet printing, in which the printer generates the piezoelectric nozzle excitation signal with the same ink-droplet spacing, leaving the ink drops on the substrate. The distance between the ink drops is the same, and eventually, all ink drops form a model pattern. However, because it is difficult to set the print interval length of a single print command as an integral multiple of the ink-droplet spacing when the piezoelectric nozzle performs a complete multiple of the print length, a length of less than one ink-droplet spacing remains, which would not trigger the nozzle to print. The gaps in the overall model boundary produce a jagged pattern. To solve this problem, we propose a new method for correcting the alignment of printed conductive patterns, wherein the ink-droplet spacing of each print command is adjusted and optimized. This improves the uniformity of the conductive pattern boundaries. The stepwise procedure for this method is as follows.

Step 1: A single triangular patch requires route planning and generates print commands to control the printer, as shown in Figure 2a. Length L of the printing space can be provided in the printing command.

FIG. 2.

(a) Path planning for triangular patches. (b) Schematic of ink-drop radius compensation. (c) Comparison of the boundary-alignment-optimization method and conventional uniform ink-droplet-spacing method. (d) Flowchart of boundary-alignment algorithm judgment.

Step 2: As the ink droplet is not a dot but a circle, the length of the printing space requires an ink-droplet-radius compensation to prevent the printing pattern from increasing the radius of the ink droplet. Compensation distance h is determined by ink-droplet radius R and preset ink-droplet spacing L_onestep. As shown in the geometric relationship in Figure 2b, the midpoint between the two ink droplets, ink droplet center point, and focus of the two ink drops can be assumed to constitute a right-angled triangle, which can be compensated by the Pythagorean theorem: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} h = \frac { { R - \sqrt { { R^2 } - \frac { { { L_ { { \rm { onestep } } } } } } { 2 } } } } { 2 } , \tag { 2 } \end{align*} \end{document}

where h is the droplet radius compensation length, R is the ink drop radius, and L_onestep is the preset distance between the ink drops.

Step 3: After compensating for the radius of the ink droplet using Equation (3), the final printing space length L_new is obtained. The mechanism of the boundary-alignment method is shown in Figure 2c. Length L_rest of the less than one ink-droplet spacing is evenly dispersed in the length of the printing space to set the distance between the ink droplets in this printing command so that the droplets can fill the entire length uniformly. In Equation (4), the value obtained by dividing printing section length L_new by the given ink-droplet step length L_onestep is rounded off to one decimal place, and integer N is obtained (the printing space is divided into N segments). At this time, the evaluation process is initiated, as shown in Figure 2d, to obtain the optimized distance between the ink droplets. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} {L_{{ \rm{new}}}} = L - 2h \tag{3} \end{align*} \end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} N = { \rm { round } } ( { \frac { { L_ { { \rm { new } } } } } { { L_ { { \rm { onestep } } } } } } ) , \tag { 4 } \end{align*} \end{document}

where L_new is the print length after ink-drop-radius compensation and L_onestep is the given ink-droplet step length.

Step 4: Each print command is optimized for feedback to the printer, and triangular patch printing is performed. The printing of all triangular patches must be completed to obtain a complete conductive pattern.

Figure 3a shows the result of using the boundary-alignment-optimization algorithm. This can be compared with the result of the conventional uniform ink-droplet-spacing printing method shown in Figure 3b. As shown, the uniformity of the boundary is significantly improved, and all the ink droplets are evenly distributed within the printing space. The boundary of the print obtained using uniform ink-droplet-spacing printing method clearly shows gaps, which reduce the accuracy of the boundary. We used two methods to print triangular patches and compared them with respect to boundary precision. The ink-droplet spacing is 70 μm, with resolution 363 dpi and droplet spread radius R of 50 μm, calculated according to Equation (2). A droplet radius compensation length of 7.15 μm was obtained.

FIG. 3.

Illustration of triangular patches using (a) the boundary-alignment-optimization algorithm and (b) the uniform ink-droplet spacing method. Triangular patches, as seen under a metallographic microscope and the associated boundary accuracy calculations by using (c) the border alignment method and (d) the uniform ink-droplet spacing method. Color images are available online.

A metallographic microscope (SAIKEDIGITAL SK2008H) was used to observe the boundary of the triangular patch, as shown in Figure 3c and d. The original boundary coordinates of the conductive pattern were extracted, and the reference boundary was fitted using the least square method. The absolute value of the difference between the boundary of the conductive pattern and the reference boundary was used to obtain y_i, and Equation (5) was used to calculate boundary accuracy Ra. The higher the boundary precision, the lower is the Ra value and the better is the boundary uniformity. As shown in Figure 3c, for the triangular patch optimized using the boundary-alignment-optimization algorithm, Ra = 9.17 μm, which is a significant improvement compared with the Ra = 24.47 μm of uniform ink-droplet spacing (Fig. 3d). The boundaries in Figure 3d and b show that jagged gaps are formed with a certain periodicity. Thus, the problem in the principle of uniform ink-droplet-spacing printing method is confirmed, and it is verified that the boundary-alignment-optimization method can effectively solve the problem of boundary unevenness. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} Ra \approx \frac { 1 } { n } \mathop \sum \limits_ { i = 1 } ^n { \left\vert { { y_i } } \right\vert } \tag { 5 } \end{align*} \end{document}

where y_i is the absolute value of the radial distance between the original boundary coordinates and the fitting boundary.

Flash sintering

Conductive patterns must be sintered before they can attain conductive properties. To improve sintering efficiency and reduce sintering temperature^15–18, we used a pulsed xenon lamp as a sintering source; it can release 1500 J of energy in 1–2 ms for sintering the conductive patterns. The sintering process is shown in Figure 4d. The printed conductive pattern was then dried. The temperature of the conductive pattern in the drying process was 40–50°C. At this time, the low-boiling solvent volatilizes first, and the silver concentration of the conductive pattern increases; however, the overall performance is nonconductive. Then, the flash lamp was turned on for sintering to complete the evaporation of the medium and high boiling point solvents (80–90°C). Simultaneously, the energy of the flash lamp exacerbates the thermal motion of the silver atom of the conductive pattern, destroys the bond between the atoms, and the original crystal grains gradually disintegrate into a small atomic group, transforming the metal from the solid state to flow capacity. Moreover, the adjacent nanosilver particles are connected to form a sintered neck, forming a silver film. Finally, the entire conductive pattern is assumed to have conductive properties.

FIG. 4.

FESEM images of (a) incompletely sintered silver films, (b) an optimized sintered silver film, and (c) a damaged surface of an oversintered silver film. (d) Flash sintering mechanism. (e) Sintering versus conductivity for different power densities and different number of flash pulses. (f) Temperature versus number of flash pulses at power density of 17.9 J/cm². FESEM, field emission scanning electron microscope. Color images are available online.

To determine the appropriate sintering parameters, experiments were conducted to determine the effects of different sintering durations and power densities on the conductivity of the conductive patterns. Conductivity measurements were performed on the conductive patterns by using a four-probe instrument (RTS-9). Figure 4e shows the variation in the flash sintering energy density from 13.6 to 23.4 J/cm² under the application of 1–8 flash pulses per energy density. Table 1 presents certain important sintering data. When the energy density was 13.6 J/cm², the best conductivity of 64,500 S/cm was obtained under five flash pulses; the low conductivity is due to the fact that the very low flash power cannot fully drive the nanosilver particles to form a sintered neck. The conductive pattern was observed through a scanning electron microscope (FESEM HITACHI s4800), as shown in Figure 4a. As shown, the nanosilver particles are independent of each other, and the entire conductive pattern cannot be connected into a sheet. When the energy density was 17.9 J/cm² and the number of flash pulses was 3, the highest conductivity of 69,870 S/cm was obtained. With the increasing number of flash pulses or energy density, the conductivity begins to decrease, as shown in Figure 4e. The best conductivities for medium energy densities of 20.3 and 23.4 J/cm² were 68,100 and 66,980 S/cm, respectively, which are <69,870 S/cm. This is because of the excessive energy, which destroys the sintered neck formed between the nanosilver particles. As clearly shown in Figure 4c, some of the nanosilver particles are again isolated from the other nanosilver particles because of continued sintering. The atoms around the pores are first separated from the silver layer because the energy required to achieve jump diffusion is minimized. As the flash sintering time is short, these atoms do not continue to maintain the free state after leaving the silver layer, and then the atoms begin to approach each other. This causes the occurrence of aggregation recrystallization. The overall silver layer was thus destroyed and the conductivity was reduced. Finally, 17.9 J/cm² was determined as the best sintering energy density because of the corresponding highest conductivity. Figure 4b shows that the sintering neck between the nanoparticles is larger and the silver nanoparticles are connected to each other, thus exhibiting good electrical conductivity. Simultaneously, the temperature of the sintering process was monitored using a thermal imager (FLUKE Ti300). Figure 4f shows that the maximum temperature after performing sintering eight times is 52.6°C; therefore, many materials that do not have high temperature resistance can be used as substrates.

Table 1.

Conductivity of Sample Under Different Flash Sintering Energy Densities and Number of Sintering Pulses

Energy density (J/cm²)	Number of flash pulse	Conductivity (S/cm)
13.6	2	50,690 ± 1378
	3	59,020 ± 3046
	4	63,360 ± 3812
17.9	2	60,290 ± 4052
	3	69,870 ± 2711
	4	68,500 ± 5276
20.3	2	61,780 ± 3457
	3	67,680 ± 5252
	4	68,100 ± 4083
23.4	2	66,210 ± 2839
	3	66,980 ± 3234
	4	65,740 ± 2577

Results

Printed conformal microstrip antenna

A 2 × 4 array microstrip antenna is used in this article. The dielectric substrate is a photocurable resin printed using stereo lithography appearance technology. The parameters are shown in Table 2. The roughness of the dielectric substrate after grinding is Ra = 0.094 μm. The array microstrip antenna is composed of an array of radiation layers, a substrate layer, and a ground plane layer, and is connected to a signal generator using a sub-miniature-A (SMA) interface. Figure 5a shows the structure of the microstrip antenna. Microstrip lines were used to feed the antenna in the antenna design. The simulation software HFSS was used for antenna simulation. After optimization, the size of the optimized array was 6 × 12.2 mm. The optimized printed sintering parameters were used to print and sinter the array antenna and obtain an antenna with conductive properties. The SMA connector was soldered to the wave port, and a 2 × 4 array microstrip antenna was finally obtained, as shown in Figure 5b.

FIG. 5.

(a) Schematic of a conformal array microstrip antenna. (b) Conformal array microstrip antennas. (c) Microwave anechoic chamber test. Comparisons of the simulated and actual measurements of (d) the return loss, (e) electric field pattern, and (f) magnetic field pattern.

Table 2.

Resin Substrate Parameters

Permittivity	Dielectric loss angle	Thickness (mm)	Roughness (Ra)
2.8	0.0262	1.3	0.094

A vector network analyzer (R&S ZNC) was used to test the array microstrip antenna, and the simulated and measured frequency responses are shown in Figure 5d. As shown, the return loss at the designed antenna center frequency of 13 GHz was −23.03 dB, and the bandwidth of the antenna corresponding to −15 dB was 12.37–13.11 GHz; the frequency bandwidth was 0.74 GHz. The antenna was placed in a microwave darkroom for the pattern test (Fig. 5c), and the simulation results were compared with the measured results, as shown in Figure 5e and f, where the main measured trends of the flap and side lobes are similar to the simulation results. The measured gain was 3.652 dB.

Conclusion

The proposed nonexpanded surface printing method could reduce a 3D surface into 2D triangular patches for 2D printing, which reduces the difficulty of surface printing and avoids the problem of unsintered conductive pattern ink flow caused by printing 3D surfaces. The new boundary-alignment-optimization algorithm avoids uneven pattern boundaries by optimizing the ink-droplet distance of each print command. Flash sintering was used in the sintering process, and optimum sintering parameters with an energy density of 17.9 J/cm² and sintering frequency of 3 were obtained through testing. The entire process was applied to the fabrication of a 13-GHz 2 × 4 conformal microstrip array antenna, and a 3D printed photosensitive resin arched substrate was used as the dielectric plate. The results show that the return loss and gain at 23 GHz were 23.03 and 3.65 dB, respectively. The trends of the main and side lobes are consistent with those in the simulation results, verifying the feasibility of the microstrip antenna method. Compared with the existing electroplating, lithography, and other technologies, the proposed method saves considerable production time and material costs.

Footnotes

Acknowledgments

This work was supported by the National Natural Science Foundation of China under grant 51575419 and the Joint Funds of Ministry of Education of China under grant 6141A02022107.

Author Disclosure Statement

No competing financial interests exist.

References

Taha

, Elwi

, Abbas

, et al. Conformal antenna array for MIMO applications. J Electromagnet Anal Appl, 2014; 6:43–50.

Fernandes

HCC

, Neto

, Da Silva Barreto

. Conformal antennas array for wireless communications at 2.5GHZ. J Commun Comput, 2013; 13:266–269.

Adams

, Slimmer

, Malkowski

, et al. Comparison of spherical antennas fabricated via conformal printing: Helix, meanderline, and hybrid designs. IEEE Antennas Wirel Propag Lett, 2011; 10:1425–1428.

Park

, Lee

, Park

, et al. Development of hybrid process for double-side flexible printed circuit boards using roll-to-roll gravure printing, via-hole printing, and electroless plating. Int J Adv Manuf Technol, 2016; 82:1921–1931.

Keränen

, Korhonen

, Rekilä

, et al. Roll-to-roll printed and assembled large area LED. Int J Adv Manuf Technol, 2015; 81:529–536.

Kim

, Moonz

. Highly conductive ink jet printed films of nanosilver particles for printable electronics. Electrochem Solid-State Lett, 2005; 8:30–33.

Lee

, Kim

, et al. Reliable and uniform thin-film transistor arrays based on inkjet-printed polymer. Adv Mater, 2013; 25:5886–5892.

Han

, Wei

, Dong

. Droplet formation and settlement of phase-change ink in high resolution electrohydrodynamic (EHD) 3D printing. J Manuf Processes, 2015; 342.

Smith

, Shin

, Stringer

, et al. Direct ink-jet printing and low temperature conversion of conductive silver patterns. Mater Sci, 2006; 41:4153–4158.

10.

Kim

, Lee

, Yang

, et al. Use of copper ink for fabricating conductive electrodes and RFID antenna tags by screen printing. Curr Appl Phys, 2013; 101:2038–2048.

11.

Kao

, Yeh

, Zhang

, et al. Inkjet printed series-fed two-dipole antenna comprising a balun filter on liquid crystal polymer substrate. IEEE Trans Compon Packag Manuf Technol, 2014; 7:1228–1236.

12.

Moon

, Grau

, Knezevic

. Ink-jet printing of binders for ceramic components. J Am Ceram Soc, 2002; 4:755–762.

13.

Liu

Research on the motion control system of curve inkjet. Dissertation. Beijing Institute of Graphic Communication, 2015.

14.

Gazeau

, Zeghloul

, Ramirez-Torres

A novel 5-axis robot for printing high resolution pictures from media on 3D wide surface. Int Conf Ind Technol, 2009:1–6.

15.

Sarkar

, Gupta

. Flash light sintering of silver nanoink for inkjet-printed thin-film transistor on flexible substrate. IEEE Trans Nanotechnol, 2017; 16:1649–1652.

16.

Joo1

, Hwang

, Kim

. Highly conductive copper nanomicroparticles ink via flash light sintering for printed electronics. Nanotechnology, 2009; 25.

17.

Par

, Kim

. Flash light sintering of nickel nanoparticles for printed electronics. Thin Solid Films, 2014; 550:575–581.

18.

Kang

, Ryu

, Kim

, et al. Sintering of inkjet-printed silver nanoparticles at room temperature using intense pulsed light. J Electron Mater, 2011; 40:2268–2277.