Investigation on 3-D-Printing

Technologies for Millimeter-

Wave and Terahertz
Applications
In this paper, the authors discuss several additive manufacturing technologies that
are available for high-frequency applications. An extensive experimental
investigation of 3-D metal printing of waveguide components operating at up to
330 GHz highlights the potential advantages and open technological problems.
By B ing Z hang , Senior Member IEEE, Yong -X in G uo , Senior Member IEEE, H erbert Z ir ath ,
Fellow IEEE, and Yue P ing Z hang , Fellow IEEE

ABSTRACT | Three-dimensional-printing technologies have been hybrid dielectric and metallic 3-D-printing technology for realizing
receiving great attention for a wide variety of applications in recent not only simple mmWave and THz devices but also sophisticated
years for cost effectiveness, eco-friendliness, and process simplic- mmWave and THz systems.
ity in complicated structures. This paper describes the ­challenges
and solutions of applying 3-D-printing technologies in fabricating
passive millimeter-wave (mmWave) and terahertz (THz) devices. KEYWORDS  | Millimeter wave (mmWave); terahertz (THz);
First, we review the state-of-the-art dielectric 3-D-printed passive 3-D printing
mmWave and THz devices. Then, we focus on our novel 3-D-printed
metallic passive mmWave and THz devices such as horn anten- I.   I N T RODUC T ION
nas and waveguides. Next, we analyze the dimensional tolerance The millimeter-wave (mmWave) and terahertz (THz) spec-
and surface roughness of various 3-D-printing technologies in tra are defined in the frequency range of 30–300 GHz and
order to provide a guide for choosing an appropriate technology
0.1–10 THz, respectively. Applications in the mmWave and
for specific applications. Finally, we summarize the current work
THz spectra fall into categories of communication and sens-
and identify the future studies in material powder refinement, sur-
ing. For example, the Ka-band (26.5–40 GHz) is explored
face treatment, optimization of the print process, development of
for satellite communication, close-range targeting radar,
and vehicle speed detection [1]; the Q-band (33–50 GHz)
Manuscript received June 16, 2016; revised November 29, 2016; accepted December
12, 2016. Date of publication March 1, 2017; date of current version March 20, 2017. for terrestrial microwave communication, radio astronomy,
This work was supported by the Swedish Foundation for Strategic Research (SSF)
through the ªRFIC Solutions for Very High Data Rate, Energy and Spectrum Efficient
and automotive radar [2]; the V-band (50–75 GHz) for line-
Wireless THz Communicationº program; by the Swedish Research Council (VR) of-sight unlicensed wireless communication [3]; the E-band
through the ªGigabits at THz Frequenciesº program; and by the Visiting Scientist
Program from T�l�com Bretagne, Institute Mines-T�l�com.
(60–90 GHz) for high-density fixed wireless service [4];
B. Zhang and Y.-X. Guo are with the Department of Electrical and the W-band (75–110 GHz) for automotive cruise control
Computer Engineering, National University of Singapore, Singapore 117576,
Singapore (e-mail: [email protected]). radar, military radar tracking, and targeting [5]; the D-band
H. Zirath is with the Microwave Electronics Laboratory (MEL), Department of (110–170 GHz) for long-distance mobile and backhaul
Microtechnology and Nanoscience (MC2), Chalmers University of Technology,
SE-41296, Gothenburg, Sweden and is also with Ericsson Research, Ericsson AB, communication [6]; and the G-band (140–220 GHz) for
SE-41756, Gothenburg, Sweden.
Y. P. Zhang is with the School of Electrical and Electronic Engineering, Nanyang
remote atmospheric sensing and planetary exploration [7].
Technological University, Singapore 639798, Singapore. Recently, the mmWave spectrum has also been assigned for
Digital Object Identifier: 10.1109/JPROC.2016.2639520
0018-9219 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   723
Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Table 1  Abbreviations of Terminologies II.  T H R EE-DI M E NSIONA L -PR I N T ED

M M WAV E A N D T H Z PA S SI V E DE V IC E S
Most of the 3-D-printed mmWave and THz passive devices are
lenses, electromagnetic bandgap (EBG) structures, waveguides,
filters, and horns. Traditionally, those devices are fabricated by
micromachining, electrical discharge machining (EDM), or
injection molding. By 3-D printing, lenses and EBGs can be
fabricated using dielectrics choosing from acrylic polymer, UV-
curable polymer, resin, acrylonitrile butadiene styrene (ABS),
alumina (Al2O3), and zirconia (ZrO2). Waveguides, filters, and
horns can be printed in both dielectric and metallic ways using
Ti-6Al-4V, Cu-15Sn, stainless steel, CoCr alloy, Al alloy, and Ni
alloy. The dielectric 3-D-printed devices are usually built in split
pieces that need electroplating and assembly. They are light-
the 5th-generation (5G) mobile networks applications at weight, while featuring less physical rigidness and more process
frequencies 28, 37, 39, and 64–71 GHz [8]. The THz spec- complexity. The metallic 3-D-printed devices are fabricated in a
trum is used for radio astronomy, passive remote sensing, whole piece in one run, which are more physically robust. The
aeronomy, impulse ranging, spectroscopy, and imaging [9]. conductivity of the metal eliminates the necessity of electro-
In implementing mmWave and THz systems, sophisticated plating, which simplifies the fabrication process. However, the
processes are required to fabricate mmWave and THz devices. increased body mass from metal prohibits the application where
Recently, 3-D-printing technologies have come into vision. Dif- the payload is critical. In the following, we will review represent-
ferent from traditional fabrication technologies that make such ative works on 3-D-printed mmWave and THz passive devices.
a device by a subtracting manner (such as micromachining or
spark erosion [10], [11]), 3-D-printing technologies build up A. Dielectric 3-D-Printed mmWave and THz Passive
the device in an additive way. In this sense, less power is con- Devices
sumed, less waste is generated, or at least most of the material
can be recycled, which endows 3-D-printing technologies eco- The first 3-D-printed mmWave device was from 2005,
friendliness. Moreover, 3-D-printing technologies do not incur shown in Fig. 1(a) [17]. It is an SLA 34–40-GHz dual-
mask-, tooling-, and molding-related cost, which results in fur- polarized horn together with transducers using Emmerson
ther reduced cost and short lead time. and Cumming HiK dielectric powder (​​εr​  ​​ = 3.0​). The dielec-
A rigorous definition of 3-D-printing is “a tooling-free
direct layer-by-layer fabrication process.” It covers sev-
eral different technologies, such as fused deposition mod-
eling (FDM), stereolithography apparatus (SLA), ceramic
­stereolithography apparatus (CSLA), microstereolithography
apparatus (​μ​SLA), binder jetting (BJ), polymer jetting (PJ),
selective laser melting (SLM), and electron beam melting
(EBM) as shown in Table 1. The idea of 3-D printing [or rapid
prototyping (RP), rapid tooling (RT), rapid manufacturing
(RM), and additive manufacturing (AM)] has existed for dec-
ades. In 1980, the first patent on RP was applied by Kodama.
A patent on SLA was issued to Hull in 1986 [12]. In 1992,
Deckard filed a patent on selective laser sintering (SLS)
[13]. The FDM patent was issued to Stratasys in the same
year [14]. In 1997, ARCAM was established to develop the
EBM [15]. In 2000, MCP Technologies introduced the SLM
[16]. For mmWave and THz applications, the 3-D-printing
technology could be categorized as binding or depositing
in terms of processes, active or passive in terms of device
functionalities, and metallic or dielectric in terms of mate-
rials. There has been no report on active mmWave and
THz 3-D-printed devices, since the technology is not well
developed yet to satisfy the requirement for fabrication of
mmWave and THz diodes and transistors, whose dimensions Fig. 1. (a) An SLA Ka-band horn antenna using dielectric powder
are much smaller than the minimum printable dimension. [17]. (b) A CSLA Ka-band Luneberg lens using alumina [18].

724  Proceedings of the IEEE | Vol. 105, No. 4, April 2017

 Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Fig. 2. CSLA Ka-band EBG structure using zirconia [19]. Fig. 4. A 60-GHz lens by CSLA alumina [27].

tric horn is embraced by metal holders, in which profiles of EBG structure was transformed into a W-band antenna, shown
the horn and transducer are machined. The SLM part func- in Fig. 3 [22]. The antenna has the maximum gain of 25 dBi.
tions as dielectric load. The horn has averaged 12-dBi gain A series of polymer jetted devices have been created since
and <¬20-dB x-pol. Two years later, a Ka-band Luneberg 2008. Compared with SLA, CSLA, and ​μ​SLA, the PJ is a sim-
lens implemented by CSLA on photoreactive alumina was plified process that does not need postsintering. The acrylic
shown in Fig. 1(b) [18]. It uses ultraviolet laser to trace the pro- polymer is solidified immediately by a UV lamp after being
file of the lens, and then it is sintered between 350 °C and 450 °C extruded. A 600-GHz EBG woodpile structure (WPS) and a
to solidify. Linear shrinkage is measured as 16.5% ± 0.2%. The 350-GHz Johnson EBG structure were printed [23]. The WPS
lens has an average 23-dBi gain and Gaussian beam. shows fundamental EBG at 180 GHz, secondary EBG at 278
In 2007, the CSLA was used to print a Ka-band EBG- GHz, and tertiary EBG at 372 GHz. The Johnson EBG has
based bandpass filter (BPF), shown in Fig. 2(a) [19]. The the fundamental bandgap at 223 GHz. A G-band hollow-core
EBG is housed in a metal fixture. Frequency shift is observed, electromagnetic crystal (EXMT) waveguide was fabricated by
which is generated by dimensional tolerance in printing and the same technology in 2011 [24]. It has 0.03-dB/mm propa-
shrinkage in postsintering. Manual tuning helps to bring in gation loss at 105 GHz. A D-band hollow-core EXMT antenna
accordance of the BPF’s frequency response with simulation. of the maximum 25-dBi gain was reported in 2012 [25]. The
It has center frequency of 32.94 GHz, − ​ ​3-dB bandwidth of maximum gain of 23 dBi shows up at 200 GHz. A Q-band
1.03%, averaged 3-dB insertion loss, and 0.5-dB ripples. In (33–50 GHz) Luneberg lens was reported in 2014 [26].
2008, a D-band EBG based resonator was printed by the In 2010, a 60-GHz elliptical lens was reported, as shown in
​ ​SLA on alumina [20]. The unloaded Q-factor is 2500. This
μ Fig. 4 [27]. The lens is fabricated by CSLA on alumina. It has
work, whose functional band touches the lower side of the impedance bandwidth of 55–65 GHz and the maximum gain
THz spectrum, moves a long way in 3-D-printed devices. Dif- of 21 dBi at 64 GHz. In 2011, an SLA-printed W-band wave-
ferent from the SLA technology that uses galvanometric mir- guide and a W-band corrugated horn using a UV-curable poly-
rors lighted by a focalized UV laser to polymerize the object, mer were reported in Fig. 5(a) and (b) [28]. The waveguide and
the ​μ​SLA is based on the integral illumination of the desired the corrugated horn are printed in split pieces, electroplated,
patterns using a dynamic mask. As a result, tighter dimen- and then assembled. The performance is comparable to that of
sional tolerance and smoother surface can be achieved, while commercial devices by micromachining or injection modeling.
the dynamic mask used in μ​ ​SLA raises the fabrication cost. Besides low mass, low cost, short lead time, and thermal stabil-
A W-band woodpile EBG structure was printed by the extru- ity, there remains great marketing potential for these devices.
sion freeforming technique, which was CSLA indeed, on alumina In 2013, 3-D-printed THz plasmonic waveguides were reported
in 2007 [21]. It has the 84–118-GHz bandgap. After sintering, a [29]. These waveguides use polymer jetted acrylic resin, which
total linear shrinkage of 21% is observed. In 2008, the woodpile

Fig. 5. SLA UV-curable polymer metal plated devices: (a) a W-band

Fig. 3. CSLA W-band EBG-based antenna using alumina [22]. waveguide; and (b) a W-band corrugated horn [28].

Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   725

Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Fig. 6. SLA photopolymer metal plated devices [31]: (a) an H-band Fig. 8. Ka-band offset stepped-reflector antenna by SLS nylon
waveguide; and (b) an H-band diagonal horn. metal plated [33].

is then sputter deposited with 500-nm Au. The surface rough- B. Metallic 3-D-Printed mmWave and THz Passive
ness of the plated Au is 3-​μ​m root mean square (rms). The THz Devices
EM waves propagate among the surface of the waveguide in
There are both advantages and disadvantages of dielectric
the form of surface plasmon polaritons (SPPs). Because Au is
and metallic 3-D-printed devices. For example, the advantage of
highly conductive, SPPs see very low insertion loss when air
dielectric 3-D-printed device is low body mass, which is a prior-
acts as the adjacent medium. A polymer jetted W-band dielec-
ity in space engineering. However, it is still doubtful whether rig-
tric reflectarray with the maximum gain of nearly 25 dBi was
idness could stand the acceleration of a launch vehicle. Besides,
reported in 2014 [30]. Later in 2014, an H-band (220–325 GHz)
when electroplating is needed in a narrow enclosed surface, the
SLA waveguide and a diagonal horn were introduced, as
device has to be printed in split pieces and then assembled. This
shown in Fig. 6(a) and (b) [31]. The waveguide has on average
adds to process complexity, and consequently is more costly.
0.4-dB insertion loss in the H-band. The diagonal horn has the
Moreover, because of the severe temperature ranges in space,
maximum gain of 26 dBi at 320 GHz. They are first printed
the plated metal may peel off from different coefficients of ther-
and then electroplated. It is amazing to see equal performance
mal expansion when temperate varies drastically. For the porous
of the 3-D-printed devices at such high frequency with tra-
3-D-printed dielectric devices, outgassing in vacuum and elec-
ditionally fabricated commercial counterparts. Recently,
trostatic discharge (ESD) should also be considered. The metal-
a similar SLA dual-reflector antenna has been successfully
lic 3-D-printed device has larger body mass than the dielectric
tested by the European Space Agency (ESA), which might be
device. The physical rigidness of a metallic 3-D-printed device
the first 3-D-printed device truly used in space engineering.
is an advantage, which has been already noted by the automo-
An H-band SLA lens shown in Fig. 7 has also been seen in
bile and prosthesis industries. For the conductivity of metal,
2014 [32]. It has the peak gain of 26.5 dBi at 268 GHz. In
the device can be printed in a whole piece in one run, which
2015, an SLS-printed Ka-band offset stepped-reflector antenna
eliminates the necessity of electroplating and assembly. Moreo-
was reported in Fig. 8 [33]. The design of that antenna has 3300
ver, the thermal conductivity of metal outperforms polymer and
layers. It is first sintered by a laser beam on nylon–polymaide
plastic. It can be seen that there remains a need for the metallic
PA2200, and then silver pasted. The maximum gain is 40.4 dBi
3-D-printed mmWave and THz devices.
at 30 GHz. A W-band waveguide and a W-band sixth-order BPF
Before 2012, the only report on 3-D-printed metallic
in Fig. 9(a) and (b) were reported in 2015 [34]. The devices
mmWave device was on an SLM 50-mm W-band waveguide
are fabricated by SLA in split pieces, and then electroplated.
made of Ti-6Al-4V [35]. Due to low conductivity of the titanium
The performances are comparable with traditionally fabricated
alloy and the unsatisfactory surface finishing, the waveguide is
ones. The waveguide has 11-dB/m attenuation in-band. The
quite lossy. In 2015, Zhang from Chalmers University of Technol-
BPF has the center frequency of 107.2 GHz, 6.8-GHz passband,
and 0.95-dB insertion loss.

Fig. 9. SLA resin metal plated devices: (a) a W-band waveguide; and
Fig. 7. H-band lens [32]. (b) a W-band sixth-order BPF [34].

726  Proceedings of the IEEE | Vol. 105, No. 4, April 2017

 Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

ogy conducted a series of experiments on metallic 3-D-printed

technologies for mmWave and THz applications, with the pur-
pose of finding the appropriate material and process for device
fabrication as well as exploring the limit of the metallic 3-D-print-
ing technology. The chosen materials and related technologies
were “316L stainless steel + binder jetting” and “Cu-15Sn + selec-
tive laser melting.” Postprocesses such as manual polishing, gold
electroplating, and the micromachined process (MMP) were
adopted to treat the surface [36]. Manual polishing can treat
every open facet of the horn. Because of the narrow space, the
tool used in manual polishing cannot reach the inner surface of
the feeding waveguide. It is more cost effective than the other
two. In gold electroplating, the deposited gold particles help to
smooth the surface as well as to improve the surface conductivity.
The MMP process is the most expensive; its cost equals that of
device printing. It uses abrasive-particle-contained high-pressure
fluid to wash over the object. However, since the fluid pressure
decreases dramatically along a narrow enclosed channel, the
MMP does not sufficiently ensure well-controlled roughness,
especially for small dimensions of mmWave and THz devices.
In the first phase, six V-band horns are fabricated
[37]: binder jetted 316L stainless steel without treatment Fig. 11. Roughness of 3-D-printed V-band horns' inner surfaces:
(a) binder jetted 316L stainless steel without treatment; (b) gold-
[Fig.  10(a)], gold-electroplated binder jetted 316L stainless
electroplated binder jetted 316L stainless steel; (c) MMP-treated
steel [Fig. 10(b)], MMP-treated binder jetted 316L stainless binder jetted 316L stainless steel; (d) manually polished SLM
steel [Fig. 10(c)], manually polished SLM Cu-15Sn [Fig. 10(d)], Cu-15Sn; (e) gold-electroplated SLM Cu-15Sn; and (f) MMP-treated
gold-electroplated SLM Cu-15Sn [Fig. 10(e)], MMP-treated SLM Cu-15Sn.
and SLM Cu-15Sn [Fig. 10(f)]. They are designed with the
same dimensions and aperture efficiency for a fair compari-
son. Screw threads and guiding holes are drilled after print- ing. The flange interface is milled for better connection with
the PNA waveguide extender. For a quantitative analysis, a
ZeGage optical surface profilometer is used to scan over an
800-​μ​m ×​ ​800-​μ​m area on the inner surface of each horn. The
surface roughness is illustrated in Fig. 11. Table 2 compares
the measured surface roughness. Spv is the peak-to-valley
distance over the sampling surface. Sq is the mean square root
of the surface roughness. Sa is the averaged roughness over
the sampling area. IsoFlatness reflects the overall profile flat-
ness on the sampling area. All treatments proved their effec-
tiveness in surface quality improvement. For the 316L stainless
steel, Sa is 12.89 ​μ​m on the raw surface, electroplating helps to
smooth it to Sa ​=​4.64 μ​ ​m, while MMP smooths the surface to
Sa =
​ ​7.82 ​μ​m. Cu-15Sn is intrinsically softer than the 316L stain-
less steel, and hence easy to polish. MMP outperforms electro-
plating in smoothing for Cu-15Sn. Sa is 2.79 ​μ​m for manually

Table 2  Measured Inner Surface Roughness of V-Band Horns in Fig. 10

Fig. 10. Three-dimensional-printed V-band horns: (a) binder jetted

316L stainless steel without treatment; (b) gold-electroplated
binder jetted 316L stainless steel; (c) MMP-treated binder jetted
316L stainless steel; (d) manually polished SLM Cu-15Sn; (e) gold-
electroplated SLM Cu-15Sn; and (f) MMP-treated SLM Cu-15Sn.

Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   727

Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Fig. 12. Loss of the 3-D-printed V-band horn antennas against

surface roughness.

polished Cu-15Sn, 1.29 ​μ​m for the electroplated and 0.25 ​μ​m
for MMP treatment. In terms of surface roughness, the MMP-
treated Cu-15Sn is the optimum choice for antenna fabrica-
tion. However, the cost of MMP treatment equals the antenna
printing cost, which is not preferred in volume production. The
manually polished SLM Cu-15Sn is selected as a compromise
between cost and performance. Fig. 12 describes the V-band
surface-roughness-related antenna loss. The “antenna loss” is
calculated as the difference between the measured and simu-
lated antenna gain, which is similar but not exactly the same as
the conductor loss, since other parameters might also be influ-
ential, such as the loose dimensional control, the chamfered
right-angle corner of the waveguide, and the misalignment
on the flange in the measurement. It gives a quantitative view
of the surface-roughness-related loss. It is reasonable that the
antenna loss increases with surface roughness.
In the second phase, E-, D-, and H-band horns are
printed by SLM Cu-15Sn with manual polishing to explore
the frequency limit of the metallic 3-D printing for horn
fabrication in Fig. 13 [38]. The E-, D-, and H-band horns
are designed to have the same aperture efficiency; to have
​| ​S​11​​  |  <  −​20-dB 60–90 GHz, 110–170 GHz, 220–325 GHz,
respectively; and to have in-band gains >20 dBi. The
Fig. 13. SLM Cu-15Sn conical horns: (a) overall view; (b) zoomed-in
antennas are simulated by CST Microwave Studio [39] and view at the flange and taper; and (c) photograph from the top: E-,
tested in the far-field setup at the Karlsruhe Institute of D-, and H-band horns.
Technology [40]. As shown in Fig. 13(a) and (b), the ­antennas are
composed of a standard flange; a section of a rectangular wave-
guide with cross section a​ . × .b​and length v​ ​; a taper (length ​t​) r­ espectively. In Fig. 14 (a), the behavior of the simulated and
from the rectangular waveguide to the circular waveguide measured ​| ​S​11​​ |​agrees well. It has a ​| ​S​11​​  |  <  −​20-dB impedance
(diameter g​ ​); a circular waveguide (diameter ​g​ and length h​ ​ bandwidth across the E-band. A 0.5-dB difference between the
); and the conical horn (slant length ​l​, aperture diameter d​ ​, simulated and measured gain is observed in Fig. 14 (b). The
and wall thickness ​s​). The designed dimensions of the anten- agreement of radiation patterns on E- and H-planes is good at
nas are listed in Table 3. Table 4 lists the measured dimen- 75 GHz in Fig. 14(c) and (d). For the D-band horn, ​| ​S​11​​ |​agrees
sions of the antennas. Most of the measured dimensions are well in simulation and measurement, as shown in Fig. 15(a).
within 0.05-mm dimensional tolerance. However, significant The average discrepancy between simulated and measured gain
errors are found on d​ ​and f​​of the printed horn antennas. This is 0.75 dB, as shown in Fig. 15(b). Good agreement of radia-
indicates the inconsistency of the SLM technology, which is tion patterns on both E-and H-planes is observed at 145 GHz in
a major limitation for mmWave and THz device fabrication. Fig. 15(c) and (d). ​| ​S​11​​ |​of the H-band horn remains at < –20 dB
Figs. 14–16 compare the simulated and measured across the H-band in Fig. 16(a). The gain discrepancy between
­performances of the antennas in E-, D-, and H-bands, ­simulation and measurement also drops to 1 dB in Fig. 16(b).

728  Proceedings of the IEEE | Vol. 105, No. 4, April 2017

 Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Table 3  Designed Dimensions of SLM Horn Antennas in Fig. 13

The radiation patterns agree well on E- and H-planes at 245

GHz in Fig. 16(c) and (d). Though there remain problems in
dimensional control of the SLM horns, performances of the SLM
horns are already acceptable. Compared with traditional ways to
fabricatehornantennas,suchascomputernumericalcontrol(CNC)
machining and EDM, the “Cu-15Sn + SLM” printed horns fea- Fig. 15. Simulated and measured performances of the D-band SLM
Cu-15Sn horn: (a) ​| ​S1​ 1​​ |​; (b) gain; (c) copolarized radiation patterns at
ture much reduced cost. For example, the fabrication cost of the
145 GHz on E-plane; and (d) copolarized radiation patterns at
metallic 3-D-printed H-band horn is 90% less than the price of a 145 GHz on H-plane.
commercial H-band horn on the European market.
The experiments are also conducted to use SLM to fab-
ricate E- and D-band waveguides, as shown in Fig. 17 [41]. Dimensions of the waveguide aperture and the length of
Fig. 17 (a)–(c) shows the designed profile of the wave- the straight waveguides have ±5% tolerance, while dimen-
guides. A ​ ​and b​ ​are the waveguide aperture dimensions; l​1​, l​2​, sional tolerance of the waveguide bends is quite large,
​l31​, and ​l32​ are the related length of the waveguides; r​ ​is the which originates from the printing process. For the straight
radius of the 90° bend. Table 5 lists the designed dimensions. waveguide that is printed along one direction, the dimensional
Table 6 lists the measured dimensions of SLM waveguides. tolerance is relatively easy to control. When printing along

Fig. 14. Simulated and measured performances of the E-band SLM Fig. 16. Simulated and measured performances of the H-band SLM
Cu-15Sn horn: (a) ​| ​S1​ 1​​ ​​|​; (b) gain; (c) copolarized radiation patterns at Cu-15Sn horn: (a) ​| ​S1​ 1​​ |​; (b) gain; (c) copolarized radiation patterns at
75 GHz on E-plane; and (d) copolarized radiation patterns at 75 GHz 245 GHz on E-plane; and (d) copolarized radiation patterns at
on H-plane. 245 GHz on H-plane.

Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   729

Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Table 4  Measured Dimensions of SLM Horn Antennas in Fig. 13

Table 5  Designed Dimensions of SLM Waveguides in Fig. 17

α​  D​​​is the Ohmic loss, which is

associated with the roughness of
the waveguide inner surface [42]

​αR​  ​​  = − 10 ⋅ ​log​10​​​(1 − ​​|​S​11​​|​​​  2​)​[ dB ] (1)

⎧ 10
​  ⎪ l ( 1 − ​​|​S​11​​|​​​  2​) ​
__ ​  ⋅ ​log​ ​​​ ______​​|​S​21​​|​​​  2​
− ​  10 ​    ​ ​[ dB/m ]
   
​​ ​​ (2)
​ D​  ​​  = ⎨
⎪ − ​  ____
α ​ ​     ​ ​​

( 1 − ​​|​S​11​​|​​​  2​)
10 ​ λ​  g​​ 2
​​|​S​ ​​|​​​  ​
​  21   ​ ​[ dB / ​λ​  g​​ ]
 ​   ⋅ ​log​10​​​ ______
⎩ l

where ​​λg​  ​​​ is the guided wavelength; and ​​S​11​​​ and ​​S​21​​​ are the
measured reflection coefficient and the transmission coeffi-
cient. In this case, ​S​-parameters of the waveguides are meas-
ured using a PNA. The Ansoft HFSS is used for simulation [43].
The Ohmic loss of the SLM waveguide is compared with that
of the commercial waveguide. In Fig. 17(e), the E-band 50-mm
SLM waveguide has the largest Ohmic loss of 9.05 dB/m at
67.7 GHz with an average of 7.51 dB/m in band. The 100-mm
SLM waveguide has the largest Ohmic loss of 10.32 dB/m at
88 GHz with an average of 7.76 dB/m. The D-band 50- and
100-mm SLM waveguides are compared with commercial 127-
and 254-mm waveguides in Fig. 17(f). For the D-band 50-mm
SLM waveguide, the largest Ohmic loss is 23.09 dB/m at
116 GHz (averaged 18.96 dB/m in band). The 100-mm SLM
waveguide has the largest Ohmic loss of 28.37 dB/m at 170 GHz
(averaged 21.47 dB/m in band). The behavioral agreement of the
Fig. 17. SLM Cu-15Sn D-band waveguides: (a) 50-mm waveguide; Ohmic loss between the 50- and 100-mm SLM waveguides indi-
(b) 100-mm waveguide; (c) waveguide bend; (d) photograph from left: cates the consistency of the printing process. For the 127- and
50-mm, 100-mm, and waveguide bend; (e) measured Ohmic loss of
commercial and SLM Cu-15Sn waveguides in E-band; and (f) measured Table 6  Fabricated Dimensions of SLM Waveguides in Fig. 17
Ohmic loss of commercial and SLM Cu-15Sn waveguides in D-band.

two directions, for example the waveguide bend, the dimen-

sional tolerance becomes unacceptably large. Due to the small
waveguide opening, no postprocess can be applied to the inner
surface treatment. The total power loss of a waveguide for a
​​ ​  T​​  = ​α​  R​​  + l ​α​  D.​​​ ​​α​  R​​​is the reflection loss, while​​
given length l​​ is α

730  Proceedings of the IEEE | Vol. 105, No. 4, April 2017

 Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Table 7  Comparison of SLM Cu-15Sn Waveguides with Dielectric 3-D-Printed and Commercial Waveguides

254-mm commercial D-band waveguides, the largest Ohmic

loss is 12.37 dB/m at 117 GHz (averaged 9.67 dB/m in band) and
16.79 dB/m at 110 GHz (averaged 11.62 dB/m in band), respec-
tively. Though the D-band 100-mm SLM waveguide features
14.5-dB larger attenuation than the 127-mm commercial one,
it is promising that equal performance will become possible as
the technology improves. Table 7 compares the reported die-
lectric and SLM Cu-15Sn waveguides with commercial ones in
the mmWave and THz frequency. It can be concluded that the
SLM technology could be used for straight waveguide fabrica-
tion up to the E-band, while for the high-frequency waveguides
of small cross sections and complex structures, there remains
work to be done to make this technology suitable.
The SLM Cu-15Sn gives acceptable waveguide performance
at the E-band. Next, E-band Chebyshev bandpass filters are
attempted [44]. The BPFs are designed for high-density fixed
wireless services of passband 71–76 GHz and 81–86 GHz, Fig. 19. SLM Cu-15Sn E-band Chebyshev 11th-order 81Ð86-GHz BPF:
respectively. Contrary to the rule of thumb for filter design, (a) schematic view; (b) photograph; (c) flange view; (d) simulated
orders of the BPFs are not minimized to explore the limit of and measured frequency response ​| ​S1​ 1​​ |​; and (e) simulated and
3-D-printing technology for filter fabrication. Fig. 18 shows measured frequency response ​| ​S2​ 1​​  |​.
the 15th-order 71–76-GHz BPF. The measured frequency
responses in Fig. 18(d) and (e) show 2.5-GHz frequency shift
and average 6-dB insertion loss increase. For the 11th-order
81–86-GHz BPF in Fig. 19, the measured frequency responses
shift 3 GHz upward, and the insertion loss increases by 1.5 dB.
The increased insertion loss is caused by the surface roughness
Sa =
​ ​6 ​μ​m in this work. The 2%–3% variation of the cavity size
caused by the dimensional tolerance in printing is accountable
for the shifted frequency response. Minimizing the orders of
BPFs can help to alleviate the discrepancies between measure-
ment and simulation. Post-tuning can be applied to compensate
for the frequency shift. Electroplating can improve the surface
roughness that results in the controlled insertion loss. It can be
seen that as the complexity of the structure increases, the inca-
pability of the SLM technology becomes obvious. Tough behav-
ioral agreements are observed in the SLM BPFs, SLM is not yet
competitive with traditional technologies for filter fabrication.

III.   FA BR IC AT ION T OL ER A NC E
A NA LYSIS
Fig. 18. SLM Cu-15Sn E-band Chebyshev 15th-order 71Ð76-GHz BPF:
(a) schematic view; (b) photograp; (c) flange view; (d) simulated
There are two factors that dominate the fabrication toler-
and measured frequency response ​| ​S1​ 1​​ |​; and (e) simulated and ance in 3-D printing: dimensional tolerance and surface
measured frequency response ​| ​S2​ 1​​  |​. roughness. A tolerance analysis is helpful for choosing the

Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   731

Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Table 8  Comparison of Dielectric 3-D-Printing Technologies postsintering. For EBM and SLM, no postsintering is needed
which leads to tighter dimensional tolerance. In EBM, inert
gas is needed to prevent the metal from being oxidized, which
brings in extra cost. SLM overcomes this problem by using
a high-power laser beam to directly melt the metal powder.
SLM is the optimum for metallic part printing, considering
dimensional tolerance and cost.

B. Surface Roughness
Surface roughness is of more concern in 3-D print-
ing. It is related with material particle size, laser/electron
beam size, density of the material powder, Gaussianity of
appropriate technology for specific application. Tables 8 the laser beam, and movement control of the laser/electron
and 9 list the dielectric and metallic 3-D-printing tech- beam. Existing techniques to improve surface quality of
nologies that could be used for mmWave and THz device 3-D-printed devices include mechanical polishing, chemical
­fabrications. polishing, MMP, and electroplating.
For dielectric 3-D printing, the “scallop effect” is quite
A. Dimensional Tolerance
obvious in FDM. Surface roughness, which is in the scale
The dimensional tolerance is decided by the material of millimeters, is determined by the nozzle size and mate-
particle size, thermal shrinkage in sintering and postsinter- rial filament diameters. The SLS has better surface rough-
ing, laser/electron beam size, and movement control of the ness than FDM. The powder size, Gaussianity of the laser
laser/electron beam and nozzle. The thermal shrinkage of beam, and speed of the beam movement determine surface
the 3-D-printed devices, either dielectric or metallic, could roughness in SLS. The liquid-state-material-based 3-D print-
be compensated for by preliminary correction of geometry ing, such as SLA and PJ, improves roughness by the surface
to be manufactured. Though there is no related study in such tension of the liquid. The minimum extruded droplet size is
sense, according to the authors’ experience, this possibility ​~​16 ​μ​m in PJ, and after surface tension smoothing, it
is viable. Further improvement of the dimensional tolerance becomes ​~​1 μ​ ​m. SLA cures a sink of resin by a laser beam,
is a multidiscipline involved subject that requires coopera- and the macroscopic surface tension of the resin sink results
tion in the field of material science, optics, and mechanics. in the best roughness of ​~​0.15 μ​ ​m. SLA is the optimum
Table 8 lists the popular dielectric 3-D-printing technol- choice for mmWave and THz dielectric 3-D printing, con-
ogies. FDM is not considered because the dimensional tol- sidering surface roughness.For metallic 3-D printing, the
erance is too large for mmWave and THz applications. It is EBM technology has the worst surface roughness of ​~​5 ​μ​m.
often used in household 3-D printers. SLA on liquid material It is believed that there remains space for improvement up to​
has tight dimensional tolerance. The dimensional tolerance ~​1 μ​ ​m. BJ has the second worst surface roughness of ​~​4 μ​ ​m.
occurs in the form of thermal shrinkage in both sintering Because the metal powder is not completely melted but sin-
and postsintering. In PJ, the polymer is cured immediately tered, the roughness is basically dependent upon the particle
by a UV lamp after being extruded. Thus, it avoids the risk of size. SLS and SLM have the best roughness of ​~​1.5 ​μ​m and​
thermal shrinkage in postsintering. Trading off between pro- ~​2 μ​ ​m, respectively. SLM is the best choice for metallic 3-D
cess complexity and dimensional tolerance, PJ is the opti- printing according to surface roughness. It is worthy of men-
mum to print the mmWave and THz dielectric parts. tioning that surface roughness in Table 9 does not practi-
Table 9 lists the metallic 3-D-printing technologies. BJ and cally reflect the optimum roughness that could be achieved
SLS have medium dimensional tolerance, since the devices by different techniques. For example, the melting-based
may go through thermal shrinkage in both printing and technique should have better roughness than the sintering-
Table 9  Comparison of Metallic 3-D-Printing Technologies based technique for the surface tension of the melted metal.
In the long run, EBM and SLM are expected to outperform
the other two methods in surface finishing, while the tech-
nology develops. Fig. 20 shows the X-ray inspection of the
inner surface of BPF in Fig. 19 by Carl Zeiss Metrotom 800.
Highlighted color indicates worse surface roughness. It can
be seen that the largest roughness is 162 μ​ ​m, which is more
like a fabrication abnormality caused by the residual metal
powder. Residual metal powders are also spotted at cavity
corners. This explains the frequency shift and increased
insertion loss in Figs. 18 and 19. The deteriorated surface
roughness is observed within such a narrow waveguide of

732  Proceedings of the IEEE | Vol. 105, No. 4, April 2017

 Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

Fig. 20. X-ray inspection of the SLM Cu-15Sn Chebyshev 11th-order

81Ð86-GHz BPF.

3.1 mm × ​ ​ 1.5 mm that manual polishing is impossible. The

MMP process does not show effectiveness in an aperture less
than 30-mm diameter, when the fluid pressure decreases
dramatically along the narrow tunnel. The ultrasonic sink
can help to “shake off” the residual metal powder, which
results on average in 0.1-dB insertion loss improvement. The
microscopic flange view of SLM waveguides compared with
commercial ones reveals surface roughness in a different
way. For the commercial WR-06 flange in Fig. 21(a), sharp
right-angle corners and smooth waveguide surfaces along
the short and long sides are observed. In Fig. 21(b), corners
are chamfered and the waveguide surface looks bumpy. In
Fig. 21(c), the commercial WR-03 flange is still neat, while
the 3-D-printed flange in Fig. 21(d) is rough.

I V.   CONCLUSION
This paper has presented the state-of-the-art 3-D-printed
passive devices for mmWave and THz applications. Typical Fig. 21. Comparison of the commercial and SLM Cu-15Sn flanges: (a)
commercial WR-06; (b) SLM Cu-15Sn WR-06; (c) commercial WR-03;
examples of passive devices are lenses, waveguides, filters,
and (d) SLM Cu-15Sn WR-03.
EBGs, and horns. Some of them are printed in split pieces
and then assembled, which does not demonstrate the supe-
riority of this technology, while introduces more freedom
is achieved. Though performances of the 3-D-printed
to the surface treatment and dimensional tolerance con-
mmWave and THz devices is not yet comparable with that
trol. Others are printed in one piece in one run, which fea-
of traditional fabrication technologies, it is believed that
tures process simplicity while the surface roughness and
with the development of 3-D-printing technology as well
dimensional tolerance are of concern. A series of experi-
as related subjects such as material science and surface
ments are conducted to explore the potential and limit of
treatment technology, the 3-D printing will become one of
metallic 3-D printing to fabricate mmWave and THz pas-
the mainstream technologies for mmWave and THz device
sive devices. Compared with traditional techniques such
fabrication.
as micromachining, EDM, and injection molding, cur-
Major problems of the 3-D-printed devices originate
rently the state-of-the-art metallic 3-D printing is not yet
from the surface roughness and dimensional tolerance. The
competitive in terms of dimensional tolerance and surface
current state of 3-D-printed mmWave and THz devices sug-
roughness. For horn antenna fabrication, the 3-D-printing
gests future trends in the following aspects:
technology has achieved comparable antenna performance
to that of traditional technologies. To fabricate waveguide, 1) Material powder refinement. The particle size of the
the 3-D-printing technology becomes incapable when the material determines the dimensional tolerance and
waveguide cross section decreases and the structure com- surface finishing of the printed product. Refined
plexity increases. For filter fabrication, only behavioral powder size leads to tight dimensional tolerance
agreement between designed and measured performance and improved surface roughness.

Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   733

Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

2) Development of surface treatment technology. The dielectric+metallic printing. For example, a dielec-
existing technologies have their limit in narrow tric loaded metallic horn or a fully packaged inte-
enclosed surfaces. Hopefully, the development of grated circuit (IC) can be expected by the hybrid
surface treatment technology will break the limit as dielectric+metallic 3-D printing. 
a substantial amendment for the quality improve-
ment of 3-D-printed devices. Acknowledgement
3) Print process optimization. Existing print technolo- The authors would like to thank Prof. P. Linnér, Prof. V.
gies can ensure controlled dimensional tolerance Vassilev, Prof. P. Starski, Prof. J. Stake, Prof. Y. Cao, Dr.
only along one direction. When the design has mul- L. Hammar, Prof. J. Sun, Dr. P. Nilsson, Prof. S. Chered-
tiple extensions toward different orientations, the nichenko, Dr. A. Pavolotskiy, Prof. V. Desmaris, Dr. P. L.
dimensional tolerance increases sharply to an unac- Tam, and Prof. U. Södervall from Chalmers University
ceptable level. Further optimization of the printer of Technology; Dr. Y. Li, Dr. M. Bao, Dr. O. Tageman,
or the print process needs to be conducted for bet- Dr. J. Hansryd, and Dr. T. Emanuelsson from Ericsson
ter dimensional control, especially for the complex AB, Sweden; Dr. S. Thundal from ARCAM AB, Sweden;
structure printing. Dr. H. Kimblad from Höganäs AB, Sweden; Dr. F. Bajard
4) Hybrid dielectric 3-D-printed devices. Existing tech- from BINC; Prof. C. Kärnfelt and R. Jezequel from Télécom
nologies allow the dielectric hybrid printing of dif- Bretagne-Institute Mines-Télécom; Dr. H. Gulan and Prof.
ferent materials. For the different permittivity of T. Zwick from Karlsruhe Institute of Technology; Prof.
various dielectric devices, devices such as a hybrid S. Lucyszyn from Imperial College London; Dr. Z. Zhan
printed dielectric lens seem possible. from Nanyang Technological University; Dr. T. Deng from
5) Hybrid dielectric and metallic 3-D-printing tech- the National University of Singapore; and Dr. Z. Chen
nology. Further development of 3-D-printing from the Agency for Science, Technology and Research,
technology is expected to evolve into a hybrid Singapore, for the discussions, comments, and assistance.

R EFER ENCES security and surveillance,” Proc. IEEE, vol. manufacturing of 3-D millimeter-wave
[1] S. Shambayati, “Ka-band telemetry operations 95, no. 8, pp. 1683–1690, Aug. 2007. filters,” IEEE Trans. Microw. Theory Techn.,
concept: A statistical approach,” Proc. IEEE, [10] X. Shang, M. Ke, Y. Wang, and M. J. vol. 55, no. 3, pp. 548–554, Mar. 2007.
vol. 95, no. 11, pp. 2171–2179, Nov. 2007. Lancaster, “WR-3 band waveguides and [20] T. Chartier, C. Duterte, N. Delhote, D.
[2] V. B. Atanassov et al., “Experimental study filters fabricated using SU8 photoresist Baillargeat, and S. Verdeyme, “Fabrication of
of nonstationary X- and Q-band radar back- micromachining technology,” IEEE Trans. millimeter wave components via ceramic
scattering from the sea surface,” Inst. Electr. THz Sci. Technol., vol. 2, no. 6, pp. 629–637, stereo- and microstereolithography
Eng. Proc. F—Radar Signal Process., vol. 137, Nov. 2012. processes,” J. Amer. Ceram. Soc., vol. 91, no. 8,
no. 2, pp. 118–124, Apr. 1990. [11] F. Caspers and E. Neumann, “Optical-fibre pp. 2469–2474, Jul. 2008.
[3] B. Zhang, D. Titz, F. Ferrero, C. Luxey, and end preparation by spark erosion,” Electron. [21] Y. Lee et al., “Rapid prototyping of ceramic
Y. P. Zhang, “Integration of quadruple linear- Lett., vol. 12, no. 17, Jul. 1976. millimeterwave metamaterials: Simulations
ly-polarized microstrip grid array antennas for [12] C. Hull, “Apparatus for production of three- and experiment,” Microw. Opt. Technol. Lett.,
60-GHz antenna-in-package applications,” dimensional objects by stereolithography,” vol. 49, no. 9, pp. 2090–2093, Sep. 2007.
IEEE Trans. Compon. Packag. Manuf. Technol., U.S. Patent 4575330, Mar. 11, 1986. [22] Y. Lee, X. Lu, Y. Hao, S. Yang, J. Evans, and
vol. 3, no. 8, pp. 1293–1300, Aug. 2013. [13] C. Deckard, J. Beaman, and J. Darrah, C. Parini, “Directive millimeterwave
[4] H. Mehrpouyan, M. R. Khanzadi, M. Mat- “Selective laser sintering with layerwise antennas using freeformed ceramic
thaiou, A. M. Sayeed, R. Schober, and Y. cross-scanning,” U.S. Patent 5155324, metamaterials in planar and cylindrical
Hua, “Improving bandwidth efficiency in Oct. 13, 1992. forms,” in Proc. IEEE APS Int. Symp., San
E-band communication systems,” IEEE Com- [14] S. Crump, “Apparatus and method for Diego, CA, USA, Jul. 2008, pp. 1–4.
mun. Mag., vol. 52, no. 3, pp. 121–128, Mar. creating three-dimensional objects,” [23] Z. Wu, J. Kinast, M. Gehm, and H. Xin,
2014. U.S. Patent 5121329, Jun. 9, 1992. “Rapid and inexpensive fabrication of
[5] M. Lucente et al., “Experimental missions in [15] Electron Beam Melting. ARCAM, accessed on terahertz electromagnetic bandgap
W-band: A small LEO satellite approach,” May 10, 2016. [Online]. Available: http:// structures,” Opt. Exp., vol. 16, no. 21,
IEEE Syst. J., vol. 2, no. 1, pp. 90–103, www.arcam.com/technology/electron-beam- pp. 16442–16451, Oct. 2008.
Mar. 2008. melting/ [24] Z. Wu, W. R. Ng, M. Gehm, and H. Xin,
[6] B. Zhang et al., “Integration of a 140 GHz [16] SLM Solutions. SLM, accessed on May 10, “Terahertz electromagnetic crystal
packaged LTCC grid array antenna with an 2016. [Online]. Available: http://slm- waveguide fabricated by polymer jetting
InP detector,” IEEE Trans. Compon. Packag. solutions.com/about-slm rapid prototyping,” Opt. Exp., vol. 19, no. 5,
Manuf. Technol., vol. 5, no. 8, pp. 1060–1068, [17] L. Schulwitz and A. Mortazawi, “A compact pp. 3962–3972, Feb. 2011.
Aug. 2015. dual-polarized multibeam phased-array [25] Z. Wu, M. Liang, W.-R. Ng, M. Gehm, and
[7] S. Gunnarsson, N. Wadefalk, I. Angelov, H. architecture for millimeter-wave radar,” H. Xin, “Terahertz horn antenna based on
Zirath, I. Kallfass, and A. Leuther, “A 220 IEEE Trans. Microw. Theory Techn., vol. 53, hollow-core electromagnetic crystal
GHz (G-band) microstrip MMIC single-end- no. 11, pp. 3588–3594, Nov. 2005. (EMXT) structure,” IEEE Trans. Antennas
ed resistive mixer,” IEEE Microw. Wireless [18] K. F. Brakora, J. Halloran, and K. Sarabandi, Propag., vol. 60, no. 12, pp. 5557–5563,
Compon. Lett., vol. 18, no. 3, pp. 215–217, “Design of 3-D monolithic MMW antennas Dec. 2012.
Mar. 2008. using ceramic stereolithography,” IEEE [26] K. Gbele, M. Liang, W.-R. Ng, M. E. Gehm,
[8] F. C. Commission et al., “Use of spectrum Trans. Antennas Propag., vol. 55, no. 3, and H. Xin, “Millimeter wave Luneburg lens
bands above 24 GHz for mobile services,” in pp. 790–797, Mar. 2007. antenna fabricated by polymer jetting rapid
Proc. FCC, Jul., 2016. [19] N. Delhote, D. Baillargeat, S. Verdeyme, C. prototyping,” in Proc. 39th Int. Conf. Infr.,
[9] R. Appleby and R. N. Anderton, “Millimeter- Delage, and C. Chaput, “Ceramic layer-by- Millim. Terahertz Waves, Tucson, TX, USA,
wave and submillimeter-wave imaging for layer stereolithography for the Sep. 2014, p. 1.

734  Proceedings of the IEEE | Vol. 105, No. 4, April 2017

 Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

[27] N. T. Nguyen, N. Delhote, M. Ettorre, D. [32] S. Qu, H. Yi, C. Chan, and K. Ng, “Low-cost [38] B. Zhang et al., “Metallic 3D printed
Baillargeat, L. Coq, and R. Sauleau, “Design discrete dielectric terahertz lens antenna antennas for millimeter- and submillimeter-
and characterization of 60-GHz integrated using 3D printing,” in Proc. IEEE Conf. wave applications,” IEEE Trans. THz Sci.
lens antennas fabricated through ceramic Antennas Meas. Appl., Antibes Juan-les-Pins, Technol., vol. 6, no. 4, pp. 592–600,
stereolithography,” IEEE Trans. Antennas France, Nov. 2014, pp. 1–3. Jul. 2016.
Propag., vol. 58, no. 8, pp. 2757–2762, [33] L. G, Menéndez, O. Kim, F. Persson, [39] “CST Microwave Studio,” Framingham,
Aug. 2010. M. Nielsen, and O. Breinbjerg, “3D printed MA, USA, Tech. Rep., 2016.
[28] P. T. Timbie, J. Grade, D. van der Veide, B. 20/30-GHz dual-band offset stepped-reflector
[40] H. Gulan et al., “Probe based antenna
Maffei, and G. Pisano, “Stereolithographed antenna,” in Proc. Eur. Conf. Antennas Propag.,
measurements up to 325 GHz for
mm-wave corrugated horn antennas,” in Lisbon, Portugal, May 2015, pp. 1–2.
upcoming millimeter-wave applications,”
Proc. 36th Int. Conf. Infr. Millim. Terahertz [34] M. D’Auria et al., “3-D printed metal-pipe in Proc. IEEE Int. Workshop Antenna
Waves, Houston, TX, USA, Oct. 2011, rectangular waveguides,” IEEE Trans. Technol., Karlsruhe, Germany, Mar. 2013,
pp. 1–3. Compon. Packag. Manuf. Technol., vol. 5, pp. 228–231.
[29] S. Pandey, B. Gupta, and A. Nahata, no. 9, pp. 1339–1349, Sep. 2015.
[41] B. Zhang and H. Zirath, “Metallic 3-D
“Terahertz plasmonic waveguides created via [35] K. Van Caekenberghe, P. Bleys, T. Craeghs, printed rectangular waveguides for
3D printing,” Opt. Exp., vol. 21, no. 21, M. Pelk, and S. Van Bael, “A W-band millimeter-wave applications,” IEEE Comp.
pp. 24422–24430, Oct. 2013. waveguide fabricated using selective laser Packag. Manuf. Technol., vol. 6, no. 5,
[30] P. Nayeri et al., “3D printed dielectric melting,” Microw. Opt. Technol. Lett., vol. 54, pp. 796–804, May 2016.
reflectarrays: Low-cost high-gain antennas at no. 11, pp. 2572–2575, Nov. 2012.
sub-millimeter waves,” IEEE Trans. Antennas [42] F. Warner, “Attenuation measurement,” in
[36] Micro Machined Process. BINC, accessed on
Propag., vol. 62, no. 4, pp. 2000–2008, Microwave Measurements, A. E. Bailey, Ed.,
Dec. 15, 2015. [Online]. Available: http://
Apr. 2014. 2nd ed. London, U.K.: IEE, 1989, pp. 132–134.
www.binc.biz/gb/index.php
[31] A. von Bieren, E. de Rijk, J.-P. Ansermet, and [37] B. Zhang, P. Linnér, C. Karnfelt, P. L. Tam, [43] HFSS, Version 16.0, Ansoft Corp. LLC,
A. Macor, “Monolithic metal-coated plastic U. Södervall, and H. Zirath, “Attempt of the Pittsburgh, PA, USA, 2016.
components for mm-Wave applications,” in metallic 3D printing technology for [44] B. Zhang and H. Zirath, “3D printed iris
Proc. 39th Int. Conf. Infr. Millim. Terahertz millimeter-wave antenna implementations,” bandpass filters for millimetre-wave
Waves, Tucson, TX, USA, Sep. 2014, in Proc. Asia–Pacific Microw. Conf., Nanjing, applications,” Electron. Lett., vol. 15, no. 22,
pp. 1–2. China, Dec. 2015, pp. 1–3. pp. 1791–1793, Oct. 2015.

ABOUT THE AUTHORS

Bing Zhang (Senior Member, IEEE) was born in Yong-Xin Guo (Senior Member, IEEE) received
Taiyuan, China. He received the B.E. degree from the B.Eng. and M. Eng. degrees from Nanjing
the Civil Aviation University of China, Tianjin Shi, University of Science and Technology, Nan-
China, in 2004, the M.E. degree from Shanxi Uni- jing, China, in 1992 and 1995, respectively, and
versity, Shanxi Sheng, China, in 2008, and the the Ph.D. degree from City University of Hong
Ph.D. degree from Nanyang Technological Uni- Kong, Hong Kong, in 2001, all in electronic
versity, Singapore, in 2012, all in electrical and engineering.
electronic engineering. From September 2001 to January 2009, he
He was a Visiting Scholar with the Univer- was with the Institute for Infocomm Research,
sity of Nice Sophia Antipolis, Nice, France, in Singapore, as a Research Scientist. He joined
May 2012, and a Visiting Scientist with T�l�com Bretagne, Brest, France, the Department of Electrical and Computer Engineering, National Uni-
in November 2014 and April 2015. He was a Postdoctoral Researcher with versity of Singapore (NUS), as an Assistant Professor in February 2009
the Microwave Electronics Laboratory, Department of ­Microtechnology and was promoted to a tenured Associate Professor in January 2013. He is
and Nanoscience, Chalmers University of Technology, Gothenburg, Director of Center for Microwave and Radio Frequency at the Department
Sweden, from November 2011 to October 2015. He has been a Research Fel- of Electrical and Computer Engineering, NUS. Concurrently, he is a Sen-
low with the Department of Electrical and Computer Engineering, National ior Investigator at the National University of Singapore Suzhou Research
University of Singapore since November 2016. He is a part-time Consult- Institute (NUSRI), Suzhou, China and Director of Center of Advanced Micro-
ant with Sunrise Company, Ltd., Guangzhou, China, where he is involved in electronic Devices at NUSRI. He has authored or coauthored 182 interna-
developing mobile base station antennas. He is the Chief Technology Officer tional journal papers and 187 international conference papers. Thus far,
of Blue Ocean Information Technology Company, Ltd., Wuhan, China, where his publications have been cited more than 5100 times and his H-index is
he is involved in developing wearable electronic devices. He has authored 40 (source: Google Scholar). He holds seven granted/filed patents in the
one book chapter, one book, and two patents. His current research inter- United States and China. His current research interests include MMIC mod-
ests include design and codesign of RF passive and active devices, mmWave eling and design, RF energy harvesting and wireless power for biomedical
and THz antennas, packaging of mmWave and THz devices, and 3-D printing applications and IoTs, microstrip antennas for wireless communications,
technologies for millimeter-wave and THz applications. implantable/wearable antennas, on-chip antennas and antennas in pack-
Dr. Zhang was a recipient of the Foxconn Scholarship in 2008, the age. Eight Ph.D. students graduated from NUS under his supervision.
Singapore Ministry of Education Scholarship from 2009 to 2012, Dr. Guo is General Chair of the 2017 International Applied Compu-
the Dragon Venture Award in 2012, the Best Student Paper Award of the tational Electromagnetics Society (ACES) Symposium in Suzhou, China,
Asia-Pacific Conference on Antennas and Propagation in 2012, and the and was General Chair of the 2015 IEEE MTT-S International Microwave
Young Scientist Award of the International Union of Radio Science (URSI, ­Workshop Series on Advanced Materials and Processes for RF and THz
Commission B) in 2013. He is the TPC Member and TPC Chair of several Applications (IMWS-AMP 2015), Suzhou, China and the 2013 IEEE MTT-S
international conferences. He serves as a reviewer for journals, including International Microwave Workshop Series (IMWS2013) on ªRF and Wireless
the PROCEEDINGS OF THE IEEE; the IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION; Technologies for Biomedical and Healthcare Applicationsº in Singapore.
the IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY; the IEEE TRANSACTIONS He serves as a Technical Program Committee (TPC) Co-Chair of the IEEE
ON MICROWAVE THEORY AND TECHNIQUES; the IEEE TRANSACTIONS ON COMPONENTS, PACKAG- IMWS-AMP 2017 in Italy and was TPC Co-Chair of the IEEE International
ING AND MANUFACTURING TECHNOLOGY; IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS; Symposium on Radio Frequency Integration Technology (RFIT2009). He
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS; and IET Electronics Letters. has been a TPC member and session chair for numerous conferences

Vol. 105, No. 4, April 2017 | Proceedings of the IEEE   735

Zhang et al . : Investigation on 3-D -Printing Technologies for Millimeter-Wave and Teraher t z Applications

and workshops. He serves as an Associate Editor for the IEEE ANTENNAS AND Yue Ping Zhang (Fellow, IEEE) received the
WIRELESS PROPAGATION LETTERS, IET Microwaves, Antennas & Propagation, B.E. and M.E. degrees from Taiyuan Polytechnic
and Electronics Letters. He was a recipient of the Young Investigator ­Institute and Shanxi Mining Institute, Taiyuan
Award 2009, National University of Singapore. He received the 2013 Raj University of Technology, Shanxi, China, in 1982
Mittra Travel Grant Senior Researcher Award. He received the Best Poster and 1987, respectively, and the Ph.D. degree from
Award in 2014 International Conference on Wearable & Implantable Body the Chinese University of Hong Kong, Hong Kong,
Sensor Networks (BSN 2014), Zurich, Switzerland. He is a corecipient of in 1995, all in electronic engineering.
the Design Contest Award of the 20th International Symposium on Low He was with Shanxi Electronic Industry
Power Electronics and design (ISLPED), Rome, Italy, July 2015. His Ph.D. Bureau, Shanxi, China, from 1982 to 1984; the Uni-
students received Best Student Paper Awards from IEEE MTT-S IMWS-Bio versity of Liverpool, Liverpool, U.K., from 1990 to
2015 in Taiwan, IEEE iWEM 2013 in Hong Kong, 2011 National Microwave 1992; and the City University of Hong Kong, Hong Kong, from 1996 to 1997.
and Millimeter-Wave Conference at Qingdao, China, and IEEE ICMMT 2010 From 1987 to 1990, he taught at Shanxi Mining Institute and from 1997 to
in Chengdu, China. 1998, the University of Hong Kong. He was promoted to a Full Professor
with the Taiyuan University of Technology in 1996. He is currently a Pro-
fessor with the School of Electrical and Electronic Engineering, Nanyang
Herbert Zirath (Fellow, IEEE) was born in
Technological University, Singapore. He has delivered scores of invited
G�teborg, Sweden, on March 20, 1955. He
papers/keynote addresses at international scientific conferences. He has
received the M.Sc. and Ph.D. degrees in electrical
organized/chaired dozens of technical sessions of international symposia.
engineering from Chalmers University, G�teborg,
His current research interests include radio science and technology.
Sweden, in 1980 and 1986, respectively.
Dr. Zhang received the Sino-British Technical Collaboration Award
From 1986 to 1996, he was a Researcher at
in 1990 for his contribution to the advancement of subsurface radio sci-
the Radio and Space Science at Chalmers Uni-
ence and technology. He received the Best Paper Award from the Sec-
versity, engaged in developing a GaAs- and InP-
ond International Symposium on Communication Systems, Networks
based HEMT technology, including devices, mod-
and Digital Signal Processing, July 18Ð20, 2000, Bournemouth, U.K.,
els, and circuits. In spring/summer 1998, he was
and the Best Paper Prize from the Third IEEE International Workshop on
a Research Fellow at California Institute of technology (Caltech), Pasadena,
Antenna Technology, March 21Ð23, 2007, Cambridge, U.K. He received
CA, USA, engaged in the design of MMIC frequency multipliers and Class E
the William Mong Visiting Fellowship from the University of Hong Kong
Power amplifiers. Since 1996, he has been a Professor in High Speed Elec-
in 2005. He received the IEEE AP-S Schelkunoff Prize 2012. He was a
tronics at the Department of Microtechnology and Nanoscience, MC2, at
Guest Editor of the I­ nternational Journal of RF and Microwave Computer-
Chalmers University. He became Head of the Microwave Electronics Labo-
Aided Engineering and an Associate Editor of the International Journal of
ratory in 2001. At present, he leads a group of approximately 40 research-
Microwave Science and Technology. He serves as an Editor of ETRI Jour-
ers in the area of high-frequency semiconductor devices and circuits. His
nal, an Associate Editor of the IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION
main research interests include MMIC designs for wireless communica-
Propagation, and an Associate Editor of the International Journal of Elec-
tion and sensor applications based on III-V, III-N, Graphene, and silicon
tromagnetic Waves and Applications He serves on the Editorial Boards
devices. He is author/coauthor of more than 530 refereed journal/confer-
of a large number of journals, including the IEEE TRANSACTIONS ON MICROWAVE
ence papers, and holds five patents. He is a Research Fellow at Ericsson
THEORY AND TECHNIQUES and IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS .
AB, leading the development of a D-band (110Ð170 GHz) chipset for high
data rate wireless communication. He is a c-founder of Gotmic AB, a com-
pany developing highly integrated frontend MMIC chipsets for 60-GHz and
E-band wireless communication.

736  Proceedings of the IEEE | Vol. 105, No. 4, April 2017