IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 13, 2014 915
Optimization of Inkjet Printing of Patch Antennas on
Low-Cost Fibrous Substrates
Hossein Saghlatoon, Lauri Sydänheimo, Member, IEEE, Leena Ukkonen,Member,IEEE,and
Manos Tentzeris, Fellow, IEEE
Abstract—In this letter, the inkjet-printing procedure is used
to implement microwave circuits on low-cost ﬁbrous substrate,
cardboard. As a ﬁrst step for environmentally friendly electronics
applications, the high-frequency properties of the cardboard sub-
strate are extracted using the two-transmission-line method and
dielectric probe measurement. To provide an accurate model for
the inkjet-printed conductors, the conductivity and thicknessof
the printed silver traces are analyzed. The surface of the substrate
is pretreated using a dielectric ink to reduce the penetration of
the conductor ink into the ﬁbrous substrate and to diminish the
conductor loss at high frequencies. As a technology demonstrator,
a patch antenna is printed on a cardboard substrate, and the
simulation and measurement results are compared to study the
reliability of the obtained parameters. After initial experimental
veriﬁcation, simulation models were ﬁne-tuned in order to provide
a predictive method for design and fabrication of low-cost RF
circuits. The achieved model can be used to design and fabri-
cate low-cost RF structures on ﬁbrous environmentally friendly
Index Terms—Dielectric characterization, green electronics,
inkjet printing, low-cost substrate, paper RF, patch antenna, silver
ink, surface treatment, two transmission lines method.
AS AN organic substrate, paper has a great potential for
the realization of environmentally friendly and recyclable
electronics , . It is tempting to use paper as a substrate
for electronic circuits because it is ultra-low-cost, recyclable,
available everywhere, and compatible with printing . Addi-
tive printing technologies, such as inkjet printing, feature the
capability to increase the fabrication speed and enhance the pro-
duction versatility, both features that are critical for mass pro-
duction. Moreover, inkjet printing has been an enabling tech-
nology to print electronics on unconventional substrates such
as wood, ceramic, and paper –. The electrical properties of
the low-cost ﬁbrous types of paper, such as cardboard, should be
characterized, especially in microwave frequencies because the
RF properties of different paper materials are not identical .
In addition, it is challenging to effectively fabricate circuits on
Manuscript received December 13, 2013; revised February 20, 2014 and
April 04, 2014; accepted April 17, 2014. Date of publication May 08, 2014;
date of current version May 22, 2014.
H. Saghlatoon, L. Sydänheimo, and L. Ukkonen are with the TampereUniver-
sity of Technology, 33720 Tampere, Finland (e-mail: hossein.saghlatoon@tut.ﬁ;
M. M. Tentzeris is with the Georgia Institute of Technology, Atlanta, GA
30332-250 USA (e-mail: email@example.com).
Color versions of one or more of the ﬁgures in this letter are available online
Digital Object Identiﬁer 10.1109/LAWP.2014.2322572
ﬁbrous substrates using inkjet printing because the printed ink
penetrates into the substrate and reduces the conductivity and
quality of the printed pattern .
In this letter, one typical cardboard paper material is used
as the substrate and silver nanoparticle ink as the conductor.
The rest of this letter is organized as follows. In Section II, the
speciﬁcs of the cardboard electrical characterization technique
are discussed. Then, the measurement results for the conduc-
tivity, thickness, and surface roughness of inkjet-printed silver
traces before and after treatment of the substrate are proposed
according to Section III. Section IV discusses the design of a
patch antenna according the obtained parameters, and the effect
of inaccuracies in the fabrication and measurement procedures.
Finally, conclusions are presented in Section V.
II. RF CHARACTERIZATION OF THE CARDBOARD
To enable the effective implementation of efﬁcient microwave
structures on a cardboard substrate, its RF properties have to be
measured and analyzed. There are different ways to characterize
an unknown substrate such as split-cylinder resonator, ring
resonator, and other high- structures , . One of the most
accurate methods is the two-transmission-line method. In this
method, two identical microstrip lines with different lengths are
implemented on the substrate. The effective permittivity
of the substrate can be obtained from the phase difference of the
signals, and the loss tangent is acquired from the insertion loss. It
is not necessary to de-embed the effect of SMA connectors and
junctions because all the calculations are based on the transmis-
sion-line length difference effectively removing these parasitic
effects , . Using (1) and (2) the effective permittivity and
total loss can be calculated, respectively
where is the free-space speed of light, is the phase dif-
ference of the output signals for the long and the short lengths,
is the frequency, and is the length difference of the two
microstrip lines , .
By knowing the total loss and conductor loss, the dielec-
tric loss and then the loss tangent can be determined. The total
loss is the summation of dielectric loss, conductor loss, and ra-
diation loss. The radiation loss is very low and can be neglected
from the calculations because the lengths of the lines are rela-
tively small. In this case, to have a high efﬁcient traveling wave
1536-1225 © 2014 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.
916 IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 13, 2014
Fig. 1. Implemented microstrip lines on cardboard.
antenna, the length of the lines should be to . The con-
ductor loss is dependent on the physical properties of the con-
ductor, such as the conductivity and the width of the line. If the
properties of the conductor are known, the conductor loss is ap-
parent. The conductor loss of a certain transmission line can be
where is the angular frequency, is the conductivity of the
conductor, is the characteristic impedance of the line, and
is the width of the microstrip line. Moreover, the loss tangent
can be obtained using (5) which is valid for microstrip lines 
where is the dielectric loss, is the relative permittivity
of the substrate, is the effective relative permittivity of the
substrate, and is the phase constant in free space .
For the characterization of the highly ﬁbrous cardboard (Stora
Enso packaging thin paper), three microstrip lines with 15-, 10-,
and 5-cm lengths were realized on its surface utilizing a 50-
m-thick copper tape with bulk conductivity of S/m, ter-
minated in 50- SMA conductors, and having the same width of
2 mm for a characteristic impedance of 50 , as shown in Fig. 1.
Thelongestlineisalmost ; hence, the assumption for low
radiation loss is acceptable. For higher measurement reliability,
-parameters for the three different lengths were compared one
by one utilizing an Agilent PNA E8358A two-port vector net-
work analyzer (VNA) between 500 MHz and 3 GHz. To realize
the accuracy of this method, the relative permittivity and loss
tangent were measured using Agilent 85070 Dielectric Probe
Kit in the same frequency interval. The results for the average
permittivity and loss tangent of the substrate using these two
methods are depicted in Fig. 2. The results are in a sufﬁciently
good agreement. Depending on the density and water content of
cardboard, its permittivity can vary between 1.2 and 1.5 .
III. CHARACTERIZING THE PRINTED SILVER INK
The next important step for the inkjet printing of RF struc-
tures on cardboard is the characterization of the conductivity
and the realized thickness of the printed conductive traces.
NPS-JL silver ink with 55.5wt% metal content was used to
print using Dimatix DMP-2831 inkjet material printer. The
volume and diameter of each droplet was 10 pL and 140 m,
Fig. 2. Measured properties of cardboard. (a) Relative permittivity. (b) Loss
Fig. 3. Fabricated inkjet microstrip lines on cardboard.
respectively. To optimize the conductivity value, a sintering
process at 150 for 1 h with air circulation was applied, and
each RF pattern was printed . For appropriate conductivity,
each pattern was printed in four cycles. In each cycle, two
layers of silver ink were printed and sintered, so at the end there
were eight layers of silver ink with the total thickness of 3 m
Following this process, three microstrip lines with different
lengths were printed on cardboard as shown in Fig. 3. The di-
mensions of these lines are identical to the copper tape versions
for the purpose of easier comparison. In this case, the dielectric
loss is known, and by measuring the total loss, the conductor
loss can be obtained. The acquired value for the conductivity of
these structures is S/m.
Considering the cardboard substrate, there are numerous rea-
sons that led to a reduced value of conductivity compared to the
maximum achievable value S/m presented by the
manufacturer. These include the lower realized conductor thick-
ness compared to the skin depth and the high surface roughness.
The thickness of the printed silver ink on cardboard was mea-
sured using an optical microscope. The ink penetrates into the
cardboard due to its highly ﬁbrous nature . By penetration
of the ink into the ﬁbrous substrate, the conductivity diminishes
because the printed silver nanoparticles cannot make a uniform
conductor. In addition, cardboard is not a good thermal con-
ductor; hence, in the sintering process, the printed silver ink is
not cured properly. Therefore, there is a demand to make the
surface smooth and ink proof. It can be easily seen in Fig. 4
that without a surface treatment approach, the ink penetration
into the substrate is around 55 m. Usually, different dielec-
tric inks are utilized to treat the surface , . We used
a conventional primer (composed of tetrahydrofurfuryl acry-
late, ethoxylated trimethylolpropane triacrylate, 2-hydroxy-2-
methyl-1-phenyl-propan-1-one, and bis-phenylphosphineoxide
SAGHLATOON et al.: OPTIMIZATION OF INKJET PRINTING OF PATCH ANTENNAS ON LOW-COST FIBROUS SUBSTRATES917
Fig. 4. Measured thickness of the ink (a) after treatment and (b) before
Fig. 5. Scanned surface of cardboard using optical proﬁlometer. (a) Before
treatment. (b) After treatment.
) to prepare a smooth surface for silver ink . For this pur-
pose, four layers of primer with 1016 dpi resolution were printed
separately on the rough surface of cardboard. After the printed
deposition of each layer, it was cured using ultraviolet light
(UV) for 15 min, and then 1 h at 150 in an oven. Afterwards,
silver ink was inkjet-printed on the treated surface utilizing the
conventional sintering approach.
The surface roughness of the printed silver ink on the treated
and untreated cardboard was scanned using an optical pro-
ﬁlometer with results shown in Fig. 5. As can be deduced,
the peak-to-valley distance is around 39 m before treatment,
whichisimprovedto11.5 m after treatment. Moreover,
the mean roughness and root mean squared roughness are
improved from 1.68 and 2.13 m before treatment to 1.3 and
1.62 m after treatment, respectively. After the preparation
of the smooth and ink-proof surface, the same microstrip
lines were printed to analyze the new value of conductivity,
which was found to be almost double S/m .The
other important parameter for the printed silver ink traces for
high-frequency designs is the conductor thickness uniformity.
The cross section of the treated cardboard with printed silver
ink is shown in Fig. 4. As can be seen, the thickness is almost
uniform in the whole printed area and equal to 3 m.
Fig. 6. Conﬁguration of the inkjet-printed patch antenna on cardboard.
Fig. 7. Simulation and measurement results for the proposed antenna. (a) Ra-
diation efﬁciency. (b) Radiation gain. (c) Return loss.
TAB L E I
OBTAINED VALUES FOR DIFFERENT PARAMETERS OF CARDBOARD
AND SILVER INK
IV. IMPLEMENTATION OF THE 2.45 GHZPATC H ANTENNA
As a proof of concept and without loss of generality, a mi-
crostrip patch antenna for 2.45 GHz with inset feeding was de-
signed for the acquired properties of the cardboard and of the
silver ink traces presented in Table I. The antenna is designed to
operate in the excitation mode, and its dimensions are
shown in Fig. 6.
This conﬁguration was simulated using the 3-D full-wave
electromagnetic simulator Ansys HFSS based on ﬁnite element
method, while the inkjet-printed prototype was measured using
the near-ﬁeld measurement equipment Satimo Starlab. The sim-
ulation and measurement results for the radiation gain, radiation
efﬁciency, and return loss are shown in Fig. 7. As can be seen,
there are slight discrepancies between the simulation and mea-
surement results. We believe these discrepancies arise due to
layer-to-layer alignment error and printer tolerances m .
However, it can be observed that this phenomenon increases the
918 IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 13, 2014
Fig. 8. (a) Current distribution in the patch antenna at 2.45 GHz. (b) Misalign-
ment of the printed layers.
Fig. 9. Simulation and measurement results with the optimized model. (a) Ra-
diation efﬁciency. (b) Radiation gain. (c) Return loss.
losses in the radiating edges of the patch antenna and of the
The results for the measured misalignment of different layers
using an optical microscope and for the simulated of the cur-
rent distribution magnitude are shown in Fig. 8. As can be seen
in Fig. 8(b), the misalignments of different layers are approxi-
mately in the range of 130 m and shown with the red shift. To
compensate for the effect of misaligned layers in the simulation
model, the conductivity of the radiating edges and of the feeding
line edges is reduced. The width and conductivity of the added
part in the simulations is 130 mand S/m. The afore-
mentioned two-transmission-line method was used to measure
the conductivity of the misaligned part.
Fig. 9 depicts the simulation results for the radiation efﬁ-
ciency, radiation gain, and return loss for the new corrected
model. With this model, the matching between the simulated
and measured efﬁciency is improved by 1.5%, and the gain by
1 dB. Another potential source of measurement error is the inac-
curacy of the near-ﬁeld measurement device, which is dB
for the peak gain at 1880 MHz. Compared to the simulation
model, the roughness of the printed structure is another source
of error that will be modeled in the further studies.
V. C ONCLUSION
We have presented a novel method for the optimization of
inkjet printing of RF structures on ﬁbrous substrates. As a proof
of concept, the properties of a typically highly ﬁbrous card-
board substrate as well as of the inkjet-printed silver traces are
accurately characterized up to 3 GHz. A smoothness/substrate
hermeticity enhancing approach for the cardboard is presented,
and numerous uncertainties of inkjet printing approach are mod-
eled. The acquired values for the permittivity and loss tangent
of the cardboard, as well as the thickness and the conductivity
values for the printed silver traces are 1.78, 0.025, 3 m, and
S/m, respectively. Based on the developed method, a
patch antenna is simulated, fabricated with inkjet printing on
cardboard, and then measured. Finally, the simulation and mea-
surement results were compared, and acceptable agreement is
achieved. It can be concluded that the optimized model for card-
board and silver ink can be used in the fabrication and optimiza-
tion of RF structures on ultra-low-cost highly ﬁbrous substrates,
such as cardboard.
 S. Palacios et al., “Inkjet-printed planar antenna for a wireless sensor
on paper operating at wi-ﬁfrequency,” in Proc. IEEE APSURSI,Jul.
2012, pp. 1–2.
 M. Irimia-Vladu, “Green electronics: biodegradable and biocompatible
materials and devices for sustainable future,” Chem.Soc.Rev., vol. 43,
no. 2, pp. 588–610, 2014.
structures on paper substrates using inkjet-printing technology,” IEEE
Trans. Microw. Theory Tech., vol. 55, no. 12, pp. 2894–2901, Dec.
 J. Virkki, J. Virtanen, L. Sydanheimo, L. Ukkonen, and M. M.
Tentzeris, “Embedding inkjet-printed antennas into plywood struc-
tures for identiﬁcation and sensing,” in Proc. IEEE RFID-TA, Nov.
2012, pp. 34–39.
 A.A.Babaret al., “Inkjet-printable UHF RFID tag antenna on a ﬂex-
ible ceramic-polymer composite substrate,” in IEEE MTT-S Microw.
Symp. Dig., Jun. 2012, pp. 1–3.
treatment of ﬂexible material for printed electronics with carbon nan-
otube based ink,” Appl. Surface Sci., vol. 257, no. 8, pp. 3645–3651,
 L. Yang and M. M. Tentzeris, “Design and characterization of
novel paper-based inkjet-printed RFID and microwave structures for
telecommunication and sensing applications,” in Proc. IEEE MTT-S
Int. Microw. Symp., Jun. 2007, pp. 1633–1636.
termination of microwave dielectric constant by two microstrip line
method combined with EM simulation,” Microw. Opt. Technol. Lett.,
vol. 48, no. 11, pp. 2199–2201, 2006.
surement of substrate dielectric constant,” IEEE Trans. Microw. Theory
Tech., vol. MTT-35, no. 7, pp. 636–642, Jul. 1987.
 D. M. Pozar, Microwave Engineering, 1st ed. New York, NY, USA:
Wiley, 1990, ch. 4, pp. 126–204.
 M. Sivakumar and D. D. Deavours, “A dual-resonant microstrip an-
tenna for paperboard in the cold chain,” in Proc. IEEE Sarnoff Symp.,
2008, pp. 1–5.
 Harima, “NPS-JL silver nanoparticle ink,” NanoPaste Series, Metal
Paste for Thin Film Formation, Datasheet, 2012.
 R. Bollström et al., “A multilayer coated ﬁber-based substrate suit-
able for printed functionality,” Organic Electron., vol. 10, no. 5, pp.
 S. M. Hurst, B. Farshchian, J. Choi, J. Kim, and S. Park, “A univer-
sally applicable method for fabricating superhydrophobic polymer sur-
faces,” Colloids Surfaces A, Physicochem. Eng. Aspects, vol. 407, pp.