3D Printing of Millimetre Wave and Low-Terahertz Frequency Selective Surfaces Using Aerosol Jet Technology

An investigation of the use of Aerosol jet 3D Printing of frequency selective surface for millimetre and low-THz applications is presented in this article. This 3D printing technique allows the fabrication of intricate details of the designs with high resolution. Band-stop and band-pass FSS are designed and tested. The band stop FSS consisted of a Square loop array that operated in the 26–28 GHz sub-millimetre band. This design is printed on glass substrate and can be used for deployment in windows. The bandpass FSS arrays consisted of simple slot elements arranged in a square lattice and operated at 125 GHz and 280 GHz. The slot arrays were printed on Kapton. Surface profiles demonstrated the uniformity and precision of this printing technique. Simulated and measured results compared well and offered good performances at both the millimetre wave and low-THz bands. The designs find applications in 5G and imminent 6G communications. This printing technique also provides environmentally friendly, rapid, and sustainable alternative for development of highly customised FSS which can be deployed to improve communications in buildings and in future Terahertz applications.

Additive manufacturing (AM), commonly known as 3D Printing is a popular technique that fabricates complex structures directly from their 3-Dimensional digital models [17]- [19]. 3D Printing breaks the 3D digital models into several thin layers and prints layer-by-layer. 3D printing is considered as an alternative to traditional methods and it has witnessed unprecedented growth in the past decade. Notable implementations of 3D printing are widespread and extended from fabricating mechanical components [20] to applications such as integrating electronic circuits within manufacturing [21], [22], antennas [23]- [27] and periodic electromagnetic structures [16], [28]- [38]. VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ In terms of periodic EM structures, a variety of designs and techniques have been reported. Novel 3D structures are described in [28], [29] which use 3D printing of plasters with an additional layer of conductive material. This work was further expanded and utilized with Fused Deposition modelling (FDM) that reduced the resonant frequency by partially metalizing the hand painted conductive elements of the array [30]. 3D FSS developed using carbon reinforced plastic material was introduced in [31] where FDM using plastic polymer combined with Composite Filament Fabrication (CFF) techniques were utilized. 3D bandstop FSS with replaceable unit structures using Fused Filament Fabrication (FFF) was presented in [32]. 3D printed FSS made up entirely of dielectric elements operating at 10 GHz using stereolithography (SLA) is reported in [33]. 3D printing of metamaterial absorbers using AM techniques are reported in [33], [34] where a combination of AM techniques is employed. Broadband absorbers covering the X and K u bands were fabricated using Selective Laser Sintering (SLS) using Nylon and Iron powder in [33]. The metamaterial absorber was printed with Ultimaker using PLA [34].
Inkjet printing using conductive silver nanoparticle ink is yet another lucrative technique which has been tested for producing FSS. Inkjet printing of frame dipole FSS was reported in [16] where a piezo inkjet printer was deployed to develop the dipoles. The technique has been demonstrated on textiles as well [35]. Further, mm-wave FSS filters are described in [36], [37] where inkjet printing of FSS operating at various millimetre wave frequencies was reported.
The AM FSS described above have largely been developed to operate in the microwave bands [16], [28]- [35], although some reported FSS also operate in millimetre wave bands [36]- [37]. With a rapid expansion and roll-out of 5G technologies along with the imminent arrival of 5G beyond and 6G technologies, low-THz designs will be significantly advantageous to help increase the signal strength and decrease the signal losses in near future.
This article evaluates the use of direct Aerosol Jet Printing in developing 3D Frequency Selective Surfaces for millimetre wave and low-THz applications. Aerosol jet printing produces reliable designs on thin, flexible substrates with millimetre to micrometre level precision. Tracks ranging from as low as microns to millimetres can be printed. It also allows for fabrication on curved, 3D and uneven surfaces [38] within the machine's limits. Bandstop and bandpass FSS are examined in this article. The bandstop design operates in millimetre wave bands and is intended for controlling EM waves in 5G communications. The FSS is printed using silver nanoparticle ink on glass for potential window installation. The bandpass FSS designs operate in the millimetre wave and low-THz bands and aimed primarily at future communications systems such as 6G and beyond.
The rest of the paper is arranged as follows: Section II presents the sub-millimetre wave bandstop square loop FSS design and the aerosol jet printing process. Section III presents a sub mm-wave 125 GHz Bandpass FSS design  and section IV presents a low-THz 280 GHz bandpass FSS.

A. SQUARE LOOP FSS DESIGN
Bandstop FSS arrays are capable of operating as screens to safeguard wireless devices from signal interferences. The square loop is a well-known FSS element which was first reported in [39], offers dual polarisation and good angle of incidence behaviours. They typically resonate when the perimeter of the loop is approximately one wavelength. Similar to the other types of filtering structures, the general equation for the resonant frequency of the square loop is expressed as: where L and C relate to the dimensions and electrical properties of the conductive as well as dielectric components of a design. An FSS consisting of an array of square loop elements in a square lattice was designed to operate at around 26 GHz. The configuration a square loop unit cell and corresponding dimensions can be seen in Fig. 1 and Table 1. The unit cell side length is denoted by S while the loop size is denoted by L out . T denoted the thickness of the tracks and L in was the difference between L out and T . The unit cell was simulated using CST Microwave Studio TM with glass as substrate. The thickness of the glass was 0.7 mm. The relative permittivity  (ε r ) for Corning Eagle glass was 5.27 with a loss tangent of 0.001 [40], [41]. Fig. 2 shows the simulated transmission response of the FSS for angle of incidence behaviors at 0 degrees and 45 degrees, respectively. TE denotes the E-plane response whereas TM denotes H-plane response.
The design was not optimized for best angle of incidence performance but still provided a relatively good response. Further improvement in the transmission response at various angles of incidence could be achieved by reducing the distance between elements, using convoluted designs [42] and by selecting substrates with higher permittivity [42], [43]. The glass FSS operated at 26.7 GHz with a shift of approximately 5% at TE45 and 1% at TM45. Transmission response for glass at TE00 was below -10 dB for a range of 23.5 GHz to 29.2 GHz.

B. AEROSOL JET PRINTING
The FSS prototype was fabricated using Aerosol Jet printing that can fabricate designs with high precision [38], [44], [45]. A typical Aerosol Jet Printer layout can be seen in Fig. 3. The printing method uses aerodynamic focusing to deposit silver ink onto substrates in extremely precise and accurate manner. The ink is converted into a mist inside the atomizer. The mist is passed onto the deposition head where the mist is compressed with the help of sheath gas. Compressed air or Nitrogen are typically used as sheath gasses. As the sheath gas and ink aerosol are passed through the tiny nozzle, the aerosol is turned into an accelerated stream of droplets. The accelerated ink droplets and gasses exit the nozzle to travel and deposit onto the substrate. The height gap between the nozzle and the substrate typically varies between 2 to 5 mm, which provides the potential to print on non-uniform and 3D substrates within the limitations of the gap between the deposition head and substrates.

C. DESIGN PROFILES AND MEASUREMENTS
The FSS was printed using Cabot Nanoparticle silver ink [46], with the help of the equipment accessed through the Centre for Process Innovation, Durham, U.K. FSS array of 40 × 40 unit-cell elements was printed on a 200 mm×200 mm Corning Eagle glass of 0.7 mm thickness [40].
The design was cured with the help of dry heating to help the ink dry out and for the design to bind thoroughly across the substrate. The design was cured at a temperature of 180 0 C for approximately 60 to 90 minutes. A magnified photograph of a 4×3 section of the fabricated array can be seen in Fig. 4. The photograph was taken using a Keynce 4K Ultra HD microscope with 20x magnification. The photograph indicated the uniform printing of loops. The fabricated designs were further examined using Talysurf CCI optical interferometer with magnification factor of 50x and are shown in Fig.5. Measured surface profiles are presented in the figure where surface profile 3D map is presented in Fig. 5 (a) and crosssection view of the printed track is presented in Fig. 5 (b).
The surface profiles indicated a uniform deposition of loop elements and the coherence of printed dimensions with simulation values. The cross-section profile explained how the loops were printed as two loops, namely inner and outer loop and the ink spread across to form a single loop of desired dimensions. The calculated value of the resistivity of tracks was roughly 2.5 × 10 −7 -m which was consistent with the expected value for the Cabot nanoparticle silver ink [46]. Parameters for the roughness of the printed surface were obtained from interferometer for the FSS by taking an average VOLUME 8, 2020  of all the measured sampling lengths. The FSS had an average surface roughness (Sa) value of 0.519 µm whereas the root mean square surface roughness (Sq) value was 0.627 µm. Measured average tolerance of ±1% was observed.
RF measurements were conducted using the standard transmission response measurement setup shown in Fig. 6. The setup consisted of two standard high gain log periodic transmitter and receiver antennas which were mounted either side of an absorber screen that can revolve around its axis to obtain different angles of incidence for measurements. The FSS was mounted on a piece of foam in a slot made within the screen. The remaining gap in the slot was filled with absorbers. Transmission responses were measured with the The measured transmission responses of FSS can be seen in Fig. 7 with their corresponding simulation responses. Normal wave incident angle response for FSS on glass resonated at a central frequency of 26.6 GHz. The transmission response was below -10 dB for a range from 22.9 GHz to 28.5 GHz with a bandwidth of 21%. TE45 response exhibited a shift of 8% whereas the TM45 response observed a shift of 1%. The transmission responses for FSS covered a range in the 26 GHz as well as the 28 GHz frequency bands which are highly sought-after bands of the 5G communication spectrum [47].

III. 125 GHz BANDPASS SLOT FSS DESIGN A. SLOT FSS DESIGN
Bandpass FSS with filtering screens are widely used to realize wide-band frequency filtering responses [1]. Wideband transmission response of FSS with slot arrays makes them highly desirable for RF communication systems. Complementary dipole elements are well-known FSS elements that resonate when the slot length is approximately half wavelength in free space [1], [2].
An FSS consisting of an array of dipole slots arranged in a square lattice was designed to operate at a central frequency of 125 GHz. Fig. 8 and Table 2 correspond to the configuration and the dimensions of the unit cell. The unit cell consisted of length L and width W of 1 mm. The slots dimensions are denoted by L s as 0.8 mm and W s as 0.08 mm. The design was simulated using CST Microwave Studio TM using Kapton as substrate.

B. FABRICATION AND MEASUREMENTS
A slot FSS array consisting of a 50×50 grid of elements was printed on Kapton substrate [48] with dimensions of 63.5 mm×63.5 mm and a thickness of 0.05 mm [49]. Kapton was selected for its uniform permittivity value of 3.4 even at millimetre wave and sub-THz frequencies [48] and a loss   tangent value of 0.002. The substrate is also resistant to high temperatures which makes it suitable for various applications in space communication. The slot array was printed by dividing the design into several subparts as per linear toolpath tracks and the printing process printed the subparts as a grid of long horizontal and vertical tracks.
The printed slot can be seen in Fig. 9. Horizontal tracks with gaps in between created the long apertures that constituted the slots while the vertical tracks determined the lengths of the slots and unit cells. A single-polarized FSS design was preferred as creating a toolpath and fabrication process was simpler with fewer fabrication constraints. A series of sample prototype prints were carried out by adjusting the stream of silver ink to produce the required dimensions. The width and length of printed slots deviated slightly around 0.08 mm and 0.8 mm. A width of 80.73 µm and a length 0.79 mm were observed at the points taken in the specific sample. Calculated value of resistivity of the printed surface on Kapton was 2.6 × 10 −7 -m, which was similar to the results obtained for  the glass substrate [46]. Detailed 3D profile of the slot array using a Keynce 4K Ultra HD microscope is seen in Fig. 10. The maximum printed height was measured at approximately 45 µm. The measured Sa value was 0.202 µm while the Sq value was 0.246 µm. Measured tolerance average of ±3% was observed.
FSS was tested at the KU Leuven ESAT facility using the test setup for transmission response in Fig. 11. Absorbers around the FSS and supporting rods were removed for clarity of picture. The setup consisted of two horn antennas connected to the R&S Vector Network Analyzer with the help of frequency extenders. Some substrate area was left around the FSS for accessibility in handling the thin and flexible substrate for the duration of measurements. The FSS was mounted on a rotatable support structure.
The measurement transmission response is illustrated in Fig. 12. This FSS was designed and developed with the purpose of offering and analyzing the transmission response at normal angle of incidence and no further studies were carried out for other angles of incidence. As expected, the singlepolarized FSS presented a passband response when H-plane incident wave was lateral (E-plane incident wave orthogonal) to the slot. A clear signal blocking state was observed when the E-plane was lateral to the slot. Transmission response at normal wave angle of incidence was centered at the resonant frequency of 125 GHz. Wideband transmission response with slow roll-off rate was observed with -10 dB passband ranging VOLUME 8, 2020  from approximately 90 GHz all the way up to 170 GHz with a bandwidth of 64%. Insertion loss of 0.75 dB was observed. Single-band characteristics of FSS are highlighted with a gap of more than 20 dB between passband and signal blocking state. Good agreement was observed between simulated and measured transmission responses. Measured blocking state is 20 dB higher than simulations due to noise floor levels.

IV. 280 GHz BANDPASS SLOT FSS DESIGN A. FSS DESIGN
A further analysis of the printing technique was carried out to develop low-THz bandpass FSS and the results are presented here. The design was analogous to the design demonstrated in Fig. 8. Dimensions of the low-THz design are presented in Table 3. The unit cell size was designed 40% smaller than the previous design and the slot size was roughly halved with L and W as 0.6 mm with slot length L s of 0.43 mm and width W s of 0.04 mm.

B. FABRICATION AND MEASUREMENTS
This FSS consisted of an array of 83×83 elements printed on Kapton [49] of an area of 6.35 cm×6.35 cm with 0.05 mm thickness. The tool path and deposition head speed for this FSS were drastically and regularly altered and modified throughout the fabrication process due to the sheer complexity and the extremity of this design. A photo of the printed FSS can be seen in Fig. 13. Dimensions of 0.43 mm×0.045 mm for one sample of the printed slot were observed. The dimensions of the printed slots were mostly consistent and on average varied close to the intended dimensions. 3D surface profile of one specific single slot using Keyence 4K Ultra HD microscope is shown in Fig. 14.   Surface roughness and height were comparable with the 125 GHz design. Average tolerance of ±5% was measured.
Measured setup for this FSS was same as seen in Fig. 11. Standard R&S transmitter and receiver horn antenna were mounted with frequency extenders to operate in the low-THZ frequency band.
The measured and simulated transmission response are presented in Fig. 15. The passband operated at a resonant frequency of about 280 GHz. A wideband transmission response was observed with -10 dB passband ranging from 240 GHz all the way up to 310 GHz with a passband bandwidth of approximately 28%. The slight shift in the response was caused by scattering of the deposition of ink marginally beyond the desired dimensions during the fabrication process. -10 dB bandwidth for the low-THz design was narrower in comparison to the 125 GHz design discussed earlier.
Measured and simulated transmission responses demonstrated a gap of about 20 dB between passband and signal blocking states. A measured insertion loss of approximately 2.4 dB was observed which was higher than the previous design as expected. Exceedingly high blocking response was observed in simulation but owing to the noise floor levels and the measurement constraints, a lower blocking state response is observed which is less than half of the simulated response. The response is still below -30 dB level and is a common occurrence for measurements of this sort. Overall, simulated, and measured transmission responses showed good relationship with each other.
Finally, a comparison table is presented in Table 4, comparing the advantages and disadvantages of various FSS fabricated using four popular techniques, namely etching, screen printing, low-cost and professional inkjet printing and Aerosol Jet printing. The table presents a list of essential characteristics of the various fabrication techniques used in developing an FSS and highlights the pros and cons of the fabrication methods. Aerosol Jet Printing presented an extremely precise, environment friendly fabrication alternative to the traditional methods with the potential to fabricate on nonplanar substrates within the system constraints. VOLUME 8, 2020

V. CONCLUSION
Aerosol Jet printing of Frequency Selective Surfaces ranging from sub mm-wave to low-THz frequency regions was demonstrated. The FSS were printed using an Optomec Aerosol Jet machine and nanoparticle silver ink. The fabrication of FSS was demonstrated on glass and Kapton substrates. Kapton was the preferred substrate at low-THz frequencies owing to the lower thickness of the substrate. Smooth surfaces and uniform conductivity of the printed layers of the designs were achieved. A bandstop FSS design for sub mmwave and bandpass FSS designs for mm-wave and low-THz frequencies were proposed and studied. The bandstop FSS consisted of an array of square loops printed on glass and resonated at around 26.5 GHz. The FSS operated in the range from 24 GHz to 28 GHz which is extremely sought after in 5G and can be installed in windows for frequency shielding or to improve communications. The bandpass FSS were made up of slot arrays which were printed on Kapton and were tested at two frequencies: 125 GHz and 280 GHz.
The two designs tested the extremity and precision of aerosol jet printers and the insertion losses at the two frequencies. FSS design at 125 GHz offered a significantly wide −10 dB passband ranging from almost 90 GHz to 170 GHz with a bandwidth of 64%. Passband for 280 GHz design ranged from 240 GHz to 310 GHz with a bandwidth of 25%. More importantly, there was difference of approximately 20 dB between the passing states and the blocking states of the singly polarised designs. Developing dual-polarized bandstop design for sub mm-wave applications and singlepolarized bandpass FSS structures for mm-wave and low-THz applications provided economical and uncomplicated solutions as a validation for their development. The manufacturing challenges in the dispersion of ink and in the development of the toolpath were addressed within this study. Kapton demonstrates extremely high temperature tolerance and finds applications in space and satellite communications. The initial studies for the sub-THz design can be developed further towards developing intricate and advance designs for ongoing 5G beyond and imminent 6G networks.