Professional Documents
Culture Documents
A Ba0.6Sr0.4TiO3–ZnO–B2O3 composite ink was prepared and used for the manufacturing of fully inkjet-printed metal-
insulator-metal varactors. The dielectric thick films were co-fired with printed silver electrodes at 850°C and show a fine grained
microstructure. The films have a relative permittivity of er = 129 and a dielectric loss of tan d = 0.043 at f = 3 GHz. Printed
varactors with different dielectric film thickness were prepared. The characterization of the printed structures and 3D electromag-
netic simulations of the layout reveal the strong influence of electrode inductance and fringing effects on the properties of the
components. The printed varactors reach tunabilities between 14.4% and 16.4% by applying a tuning field of 5 V/lm. To dem-
onstrate the capability of the inkjet printing process for the preparation of tunable microwave devices, a fully inkjet-printed phase
shifter was fabricated. It is based on seven pairs of the printed varactors and reaches a phase shift of 180° and a figure of merit
of 19°/dB at 3.4 GHz.
(a) (b)
Fig. 7. Measured capacitance (a) and quality factor (b) of the fully inkjet-printed metal-insulator-metal varactors with different thickness
of the dielectric layer. The dashed line indicates the quality factor of the dielectric Ba0.6Sr0.4TiO3–ZnO–B2O3 layer.
coefficient of determination was R2 > 0.99 in all cases capacitance values Csim with the theoretical values Cideal
(2 9 9 regressions), which verifies the suitability of the (calculated according to Eq. 2) reveals the effect of
model. The inductance was calculated from Eq. (5) and fringing fields in the printed components. While the
gives values of L = (0.40 0.03) nH for 1 layer of effect of the fringing fields on capacitance is already
BST–ZnO–B2O3 and L = (0.36 0.01) nH for two 16 % for one layer of BST–ZnO–B2O3, it reaches
layers. For an ideal conductor with length l and thickness 28% for three layers.
t w, the inductance can be estimated according to Figure 7b shows the quality factor Q of the printed
Terman30 with the equation: structures. All samples undergo a decreasing quality fac-
tor with increasing frequency. This is caused by the
2l w
L 0:2l ln þ 0:5 þ 0:2235 nH/mm ð6Þ increasing dissipation factor of the dielectric as well as
w l the decreasing skin depth in the conductor films. All
For an electrode length of l = 0.8 mm, this estima- samples exhibit a nearly identical behavior and values
tion leads to an inductance of L 0.5 nH, which is in between Q = 35 and 37 at f = 0.5 GHz. The compari-
reasonable agreement with the measure values. son with the quality factor of the dielectric film
A 3D electromagnetic simulation (CST Microwave QBST = (tan d)1 reveals that the conductor loss is negli-
Studio, CST, Germany) was carried out, to determine gible at low frequencies. However, for f > 0.5 GHz, a
the effect of fringing fields on the capacitance of the deviation is observed. This is caused by the increasing
printed varactors. For the simulation, the geometric loss of the printed electrodes with higher frequency. For
model from Figure 1b was used and the overall capaci- f > 1.5 GHz, an additional minor deviation between the
tance of the varactors was calculated. Table II shows printed varactors is observed, which is also due to reso-
the parameters and results of the simulation. The nance effects in the samples with low thickness of the
simulated capacitances (Csim = 1.30–4.15 pF) are in dielectric layer.
reasonable agreement with the measured ones Figure 8 shows the tunability of the printed varac-
(Cmeas = 1.5–4.2 pF). The comparison of the simulated tors at f = 0.5 GHz. The tuning field strength E was
v = 40 mm/s were chosen to achieve a drop spacing of was determined to be d50 = 200 nm. Table I summa-
p = 80 lm. The varactor electrodes were prepared by rizes the measured properties of the dielectric ink and
printing four layers of silver each. This number of layers the conductive silver ink. The medium particle diameter
was chosen as a compromise between adequate thickness of both inks is below 1% of the nozzle diameter used for
of the electrodes (hAg = 2 lm) to reach a high conduc- inkjet printing, which was suitable to prevent clogging
tivity and sufficiently narrow electrodes to reach small during printing. The Ohnesorge number Oh was calcu-
capacitances. For the waveguides of the presented phase lated after Equation (3). For both inks, the values are
shifter, eight layers of silver were printed to reach an within the limits for a stable drop ejection, that is,
electrode thickness of about 4 lm. 0.1 < Oh < 1. The printability was confirmed experi-
After drying, the specimens were sintered for 1 h at mentally by successfully printing both inks in different
850°C in a tube furnace with heating rates and cooling experiments over a period of several days.
rates of 5 K/min. The BST–ZnO–B2O3 ink was printed on 1 lm
The printed structures were characterized by chro- Au-metallized Al2O3-substrates and sintered for 1 h at
matic white light interferometry (MicroProf, Fries 850°C. Figure 2c shows the microstructure of the sintered
Research & Technology, Bergisch Gladbach, Germany). thick film. Image analysis of the micrograph gives a mean
The electrical properties of the printed capacitances and grain size of d50 = 0.14 lm. This is a bit below the mea-
the phase shifter were determined using a vector network sured diameter of the particles in the initial ink
analyzer (37397C; Anritsu, Kanagawa, Japan). The mea- (d50 = 0.20 lm). This deviation is believed to be caused
surements were taken in a frequency range f = 0.1– by overestimating the particle size in the inks due to
10 GHz at different tuning voltages in a range of agglomerated particles as well as by underestimating the
U = 0–200 V. grain size due to overlapping grains. After sintering, 2.8-
lm thick Au-electrodes were lithographically patterned on
top of the thick films. This conventional structuring allows
Results and Discussion the preparation of capacitors with high geometrical accu-
racy and high-quality electrodes. The obtained metal-insu-
The inks were characterized with respect to their vis- lator-metal capacitor was used to determine the relative
cosity, surface tension, density, and particle size. Fig- permittivity and the dielectric loss of the material.
ure 2a shows a micrograph of the BST–ZnO–B2O3 ink. Figure 3a shows the relative permittivity and the
The measured particle size distribution of the ink is dielectric loss of the BST–ZnO–B2O3 thick films in
depicted in Fig. 2b. The mean particle size of the ink the frequency range of f = 0.1–10 GHz. The relative
Fig. 2. Scanning electron micrograph (a) and cumulative particle size distribution (b) of the Ba0.6Sr0.4TiO3–ZnO–B2O3 ink and micro-
graph of a thick film, sintered at 850°C (c).
(a) (b)
Fig. 3. Relative permittivity and dielectric loss of the Ba0.6Sr0.4TiO3–ZnO–B2O3 thick films over frequency (a) and tunability over tuning
field strength (b).
permittivity is nearly frequency independent in the stud- was calculated with the width (400 lm) and length
ied interval. The measured values at f = 3 GHz are (8 mm) of the lines. Figure 4 shows the microstructure
er = 129 3. This is considerably lower than the rela- and the sheet resistance R& of the printed electrodes.
tive permittivity of bulk Ba0.6Sr0.4TiO325 at room tem- For a thickness of hAg 0.8–1.5 lm, the silver films
perature and is due to the high porosity and the small undergo a complete separation due to the immense sin-
grain size of the printed films.24,26 Previous investiga- tering activity at 850°C. The films are not conductive in
tions showed that a lower porosity can be achieved by this case. For two printed layers (hAg 1.6–2 lm), the
using a mixture of coarse and fine grained powders27 or sheet resistance is in the range of R& = 50–65 mΩ/&.
a higher amount of sintering additive.19 The dielectric In this case, the films show a mostly closed network with
loss of the material increases with frequency. This behav- some separated areas. The increased path length and the
ior is known for tunable dielectric ceramics.4,22,26,27 At isolated areas lead to the comparable low conductance,
f = 3 GHz, the BST–ZnO–B2O3 thick films show a which corresponds to a specific conductivity of about
dielectric loss of tan d = 0.043 0.004. At 10 GHz, r 9 9 106 S/m. For three and more printed layers
the dielectric loss is increased to tan d = 0.073 0.007, (hAg > 2.4 lm), the sheet resistance further decreases
which is comparable to undoped screen-printed thick with increasing electrode thickness. This is in good
films, sintered at 1150–1200°C.26 Previous investigations agreement with the micrographs which show some pores,
have shown that the dielectric loss of BST thick films at but no separated areas in this case. The specific conduc-
microwave frequencies are mainly affected by impurities tance of the lines is r 35 9 106 S/m, which is
of the material, whereas the microstructure of the films about 56% of the conductivity of bulk silver (r
plays a less significant role.6,22,26 Figure 3b shows the 63 9 106 S/m at 20°C21) and therefore comparably
tunability of the films at different frequencies. In the high for printed films.
investigated range, the films show a frequency-indepen-
dent behavior, which is in good agreement with previous
investigations on BST thick films28 and thin films.5,29
By applying a tuning voltage of U = 200 V, a tuning
field of E = 87 V/lm is reached without breakdown.
This leads to an extraordinary high tunability of the
films, which is s = 75.3%. Furthermore, a tuning field
strength of E = 10 V/lm and a tunability of s = 22.8%
is reached by applying only 23 V tuning voltage.
Silver electrodes were printed and characterized with
respect to their morphology and conductivity. A layout
with 400 lm wide lines was printed and sintered at
850°C. The line thickness was varied by printing 1–5
layers, which results in a thickness between 0.8 and
4 lm. The microstructure was investigated by SEM.
Four-wire sensing was used to determine the electric Fig. 4. Sheet resistance over electrode thickness of printed silver
resistance of the printed lines. The sheet conductance electrodes, sintered at 850°C.
www.ceramics.org/ACT Inkjet Printed Metal-Insulator-Metal Capacitors E169
(a) (b)
Fig. 5. Topography (a) and microscopic image (b) of a fully inkjet-printed metal-insulator-metal varactor.
(a) (b)
Fig. 7. Measured capacitance (a) and quality factor (b) of the fully inkjet-printed metal-insulator-metal varactors with different thickness
of the dielectric layer. The dashed line indicates the quality factor of the dielectric Ba0.6Sr0.4TiO3–ZnO–B2O3 layer.
coefficient of determination was R2 > 0.99 in all cases capacitance values Csim with the theoretical values Cideal
(2 9 9 regressions), which verifies the suitability of the (calculated according to Eq. 2) reveals the effect of
model. The inductance was calculated from Eq. (5) and fringing fields in the printed components. While the
gives values of L = (0.40 0.03) nH for 1 layer of effect of the fringing fields on capacitance is already
BST–ZnO–B2O3 and L = (0.36 0.01) nH for two 16 % for one layer of BST–ZnO–B2O3, it reaches
layers. For an ideal conductor with length l and thickness 28% for three layers.
t w, the inductance can be estimated according to Figure 7b shows the quality factor Q of the printed
Terman30 with the equation: structures. All samples undergo a decreasing quality fac-
tor with increasing frequency. This is caused by the
2l w
L 0:2l ln þ 0:5 þ 0:2235 nH/mm ð6Þ increasing dissipation factor of the dielectric as well as
w l the decreasing skin depth in the conductor films. All
For an electrode length of l = 0.8 mm, this estima- samples exhibit a nearly identical behavior and values
tion leads to an inductance of L 0.5 nH, which is in between Q = 35 and 37 at f = 0.5 GHz. The compari-
reasonable agreement with the measure values. son with the quality factor of the dielectric film
A 3D electromagnetic simulation (CST Microwave QBST = (tan d)1 reveals that the conductor loss is negli-
Studio, CST, Germany) was carried out, to determine gible at low frequencies. However, for f > 0.5 GHz, a
the effect of fringing fields on the capacitance of the deviation is observed. This is caused by the increasing
printed varactors. For the simulation, the geometric loss of the printed electrodes with higher frequency. For
model from Figure 1b was used and the overall capaci- f > 1.5 GHz, an additional minor deviation between the
tance of the varactors was calculated. Table II shows printed varactors is observed, which is also due to reso-
the parameters and results of the simulation. The nance effects in the samples with low thickness of the
simulated capacitances (Csim = 1.30–4.15 pF) are in dielectric layer.
reasonable agreement with the measured ones Figure 8 shows the tunability of the printed varac-
(Cmeas = 1.5–4.2 pF). The comparison of the simulated tors at f = 0.5 GHz. The tuning field strength E was
Table III. Properties of the Fully Inkjet-Printed Metal-Insulator-Metal Varactors at f = 0.5 GHz
hBST/lm Cmeas/pF Q s/%*
1 layer of BST–ZnO–B2O3 2.9 0.1 4.1 0.6 35.5 0.9 15.9 0.9
2 layers of BST–ZnO–B2O3 5.6 0.1 2.5 0.2 36.8 0.5 16.4 0.9
3 layers of BST–ZnO–B2O3 9.0 0.3 1.5 0.1 38.6 0.8 14.4 0.7
*
E = 5 V/lm.
E172 International Journal of Applied Ceramic Technology—Friederich, et al. Vol. 12, No. S1, 2015
Fig. 9. Topography and microscopic image of the fully inkjet-printed phase shifter.
(a) (b)
Fig. 10. Phase shift (a) and figure of merit (b) of the fully inkjet-printed phase shifter for different bias voltages. The highlighted frequency
range marks the operating interval where a maximal phase shift between 90° and 180° is reached.
in the reported publications (Du = 4.5–63°15–18). This magnetic simulation of the varactors shows a reasonable
is believed to be due to the lower film thickness of the accordance of the calculated capacitances with the mea-
inkjet-printed layers, which allows a larger tuning of the sured ones (7–20% deviation). The simulation further
varactors at similar biasing voltages. reveals the strong influence of the fringing field effect on
the capacitance of the printed structures. Depending on
the thickness of the dielectric layer, the effect leads to
Conclusions capacitance values that are between 16% and 28%
higher than expected by using the formula of an ideal
A Ba0.6Sr0.4TiO3–ZnO–B2O3 composite ink was parallel-plate capacitor. The printed varactors reach tuna-
prepared and used for the manufacturing of fully inkjet- bilities up to 30% by applying a tuning field of
printed metal-insulator-metal varactors. Sintering at E = 9 V/lm. To demonstrate the capability of the ink-
850°C leads to a highly porous, fine grained dielectric jet printing process for tunable microwave devices, a
film with a permittivity of er = 129 and a dielectric loss fully inkjet-printed phase shifter was developed. It is
of tan d = 0.043 at f = 3 GHz. The printed varactors based on seven pairs of the printed varactors and reaches
show a capacitance which is influenced by resonance a phase shift of Du = 90–180° with a figure of merit
effects for frequencies above 1 GHz due to the large FoM = 19–21°/dB in the frequency range of f = 2.7–
inductance of the printed electrodes. The 3D electro- 3.4 GHz.
www.ceramics.org/ACT Inkjet Printed Metal-Insulator-Metal Capacitors E173