An electrically tunable metasurface that absorbs continuous electromagnetic (EM) surface waves is proposed by taking advantage of varactor diodes embedded in the surface. On the one hand, the varactors perform as the main dissipating components due to their parasitic series resistance; on the other hand, they function as the tuning elements because the dissipation is highly dependent on their capacitance. Therefore, the absorption of the surface can be tuned by the direct current biasing voltage across the varactors, which is validated numerically and experimentally in this letter. This absorbing mechanism of the surface differs from prior surface-wave absorbers and can lead to greater flexibility for absorbing metasurfaces. In this work, a power-dependent absorbing performance is achieved by loading microwave power sensors. If incorporated with other types of sensors, the absorption could potentially be controlled by corresponding physical variables such as light, pressure, or temperature, thus giving rise to various absorbing applications in a complex EM environment.

Electromagnetic (EM) surface waves that propagate on interfaces between metal and open space are one of the main mechanisms of coupling harmful signals to electronic devices. In contrast to the direct pathway that is offered by antennas, which can be mitigated by using filters or limiters, the surface waves leak inside through inevitable cracks or gaps on the conducting enclosures of the devices, so they are much more difficult to deal with, causing interference or even damage to sensitive elements.1 

The suppression of microwave power has recently been a popular topic of research. Conventionally, microwaves can be dissipated by lossy foams or structures such as a Salisbury screen made of resistive materials.2,3 But they are usually heavy and bulky and can be as thick as one-quarter wavelength, so their applications are limited in many cases. Using the printed-circuit-board (PCB) technology, planar absorbers with small electrical thickness have been extensively developed.4–9 For surface waves, possible solutions may be an artificially engineered corrugated surface10 or the high-impedance surface (HIS),11 but these structures are only lossy if additional resistance is added. Based on the HIS, nonlinear metasurfaces that absorb high power surface waves but allow small signals have been proposed by loading the metal patterns with nonlinear passive electronic components, i.e., diodes, as well as capacitors and resistors.7–9 For example, Wakatsuchi et al. utilized diodes to rectify high power surface waves into a static field, whose energy is then stored in capacitors before being dissipated with resistors. A time period is needed for charging and discharging the capacitors, so these surfaces preferentially absorb high power pulsed signals.1,7,8 In other work, diodes were employed as switches to distinguish power levels by Kim9 and the metasurface was designed to preferentially absorb high power continuous signals. These devices are advantageous in protecting electronic systems from being impacted by high-power surface waves, while minimizing the attenuation on useful small signals used for communications.

The reactance of a HIS is determined by the sheet inductance, which is governed by its electrical thickness, and the sheet capacitance, which is due to the capacitance between neighboring metallic patches.11 Therefore, lumped varactors were used to tune the surface reactance.12–15 Meanwhile, the tunable absorbing behavior of the HIS for oblique incident plane waves was investigated,16 and a tunable metasurface cloak for reducing the scattering fields of a conducting cylinder was proposed.17 In these two papers, the varactors were merely considered to be tuning components, and their small parasitic resistance was totally neglected. In this letter, the key role of the resistance of the varactors in the absorbing performance of an impedance surface for surface waves will be discussed in detail. Since the varactors are the important components for both the capacitance-tuning and the absorbing behavior, this is a different absorption mechanism from Wakatsuchi's and Kim's works.7–9 In addition, if integrated with active sensing circuits, the absorbing application range can be largely expanded, e.g., tunable by surface wave power automatically, which will be presented in the following discussion. This is a step further from the basic direct current (DC) tuning mode and could be an inspiration for more advanced designs.

The field magnitude of surface waves decays with distance from the surface. As the surface reactance increases, the fields are bound closer to the surface.11 To illustrate this effect, several simulations are performed in the driven modal of HFSS 15.0, where an infinitely wide transverse-EM-like (TEM-like) waveguide is built with the boundary settings shown in Fig. 1(a). The top of the waveguide is set as a perfect electric conductor (PEC). Two side walls of the air box are set to be perfect magnetic conductors (PMCs). The bottom is set to be an impedance boundary with varying reactance. For this case, the dominant TEM mode is propagating, and its electric field (E-field) is perpendicular to the impedance surface.

FIG. 1.

Simulations in HFSS at 1.6 GHz. (a) TEM-like waveguide model. (b) Dissipation of the impedance boundary as a function of its reactance. E-fields with surface reactance of 300 jΩ and 900 jΩ are also shown.

FIG. 1.

Simulations in HFSS at 1.6 GHz. (a) TEM-like waveguide model. (b) Dissipation of the impedance boundary as a function of its reactance. E-fields with surface reactance of 300 jΩ and 900 jΩ are also shown.

Close modal

Magnitudes of E-field distributions at the cross section of the waveguide with the impedance boundary having a reactance of 300 jΩ and 900 jΩ, respectively, are shown in Fig. 1(b), demonstrating the field-concentrating effect of the reactance of the impedance surface. Strong fields induce more intense RF currents on the impedance boundary, resulting in a higher dissipation rate by the resistance. This is proved by the increasing dissipation illustrated in Fig. 1(b) when a constant resistance of 40 Ω is set for the impedance boundary and the reactance is swept from 50 jΩ to 900 jΩ.

The impedance boundary is then implemented as a lattice of PCB mushroom-like cells. The geometry of each cell is shown in Fig. 2(a). It includes three metal layers supported by two substrates. On the top layer are an outer ring patch and an inner square patch. Five vias with diameter of d1 connect to the center of the square patch and the four arms of the ring patch. The vias penetrate through the substrates to the bottom layer and are isolated from the ground layer with clearances of d2. The surface impedance property of the lattice is analyzed with eigenmode solver in HFSS18 and is proved to be affected by the lumped capacitance across the slot. As can be observed in Fig. 2(b), when the capacitance changes from 1 to 3 pF, the reactance is enhanced from about 200 jΩ to as large as 1100 jΩ at 1.63 GHz. Because the lumped ports for the capacitance locate in x and y directions, the tuning effect is working for surface waves that propagate in the two directions on the surface. According to the above discussion, if the lattice is lossy, its dissipation varies with the capacitance, which will be analyzed by the co-simulations presented next.

FIG. 2.

(a) The geometry of the mushroom-like unit cell, not to scale. The dimensions are a = 22, g = 0.25, w = 6.5, s = 1, h1 = 2.54, h2 = 1, d1 = 0.5, d2 = 0.8 (all in mm). (b) Simulated surface reactance of the lattice versus capacitance across the slot.

FIG. 2.

(a) The geometry of the mushroom-like unit cell, not to scale. The dimensions are a = 22, g = 0.25, w = 6.5, s = 1, h1 = 2.54, h2 = 1, d1 = 0.5, d2 = 0.8 (all in mm). (b) Simulated surface reactance of the lattice versus capacitance across the slot.

Close modal

The co-simulations are performed using HFSS and Ansys Designer 8.0, which are for EM simulations and circuit simulations, respectively. In the driven modal of HFSS, a lattice consisting of a row of ten cells is distributed at the bottom of the TEM-like waveguide. For each cell, four rectangular lumped ports are set across the slot, and two lumped ports are set at the ring-shaped clearance regions between the vias and the ground layer, as shown in Fig. 2(a). Two ends of the waveguide are set as wave ports. After EM simulation in HFSS, the model is imported into Designer, where the wave ports are connected to source and load, respectively, the rectangular lumped ports are connected to electronic components, and the ring lumped components are connected to DC biasing circuits.

A resistor, Rslot, and a capacitor, Cslot, are at first connected in series to each lumped port across the slot. The ring lumped ports are left to be open. Simulations are performed with Rslot of 0 and 1.2 Ω, respectively. Absorption is calculated from scattering parameters by

A=1|S11|2|S21|2.
(1)

Results are shown in Figs. 3(a) and 3(b) as a function of Cslot. It can be seen that, when the resistance is zero, the absorption is small except for several peaks, which can be explained by the dissipation of lossy substrates at the resonances of the lattice. In contrast, the small resistance causes obvious changes to the absorption. At 1.57 GHz for instance, as the capacitance varies from 1.5 to 2.5 pF, the absorption ranges from about 0.15 to 0.99.

FIG. 3.

Absorbing behaviors of the lattice of the mushroom-like cells. (a) and (b) are simulated results with Rslot and Cslot connected in series across the slot: (a) Rslot = 0, (b) Rslot = 1.2 Ω. (c) and (d) are results with varactors connected across the slot: (c) simulated and (d) measured.

FIG. 3.

Absorbing behaviors of the lattice of the mushroom-like cells. (a) and (b) are simulated results with Rslot and Cslot connected in series across the slot: (a) Rslot = 0, (b) Rslot = 1.2 Ω. (c) and (d) are results with varactors connected across the slot: (c) simulated and (d) measured.

Close modal

For greater accuracy, the series resistors and capacitors are then replaced with SPICE models of Skyworks SMV 1405 varactors. Their DC biasing circuits are connected to the ring lumped ports. It is found from Fig. 3(c) that the curves are consistent with those in Fig. 3(b) while the voltage is swept from 5 to 0.5 V. When the series resistance of the varactors is set to be zero, similar results to those in Fig. 3(a) are obtained, which are not shown here for brevity, indicating the significant effect of the resistance in the tunable absorbing characteristic of the lattice.

The simulations are then validated by measurements. In the simulations, the model resembles an array with infinite width that is placed at the bottom of a TEM-like waveguide. This structure is time and memory-saving for the numerical calculations, but it is impractical for measurements. Therefore, a prototype with 10 × 4 cells is fabricated with the varactors and their DC biasing circuits. The upper substrate is Rogers RT/duroid 6006, and the lower one is FR4. The circuits are distributed at the bottom layer and connected with the top layers through the vias. The prototype is tested in a WR430 transverse-electric (TE) waveguide with its wires connected to external voltage sources. Two ends of the waveguide are connected to ports 1 and 2 of an Agilent E5071C vector network analyzer (VNA). The continuous microwave signal input into the waveguide is −5 dBm. According to measured scattering parameters, absorption is obtained from Equation (1). Since the dominant mode in the TE waveguide induces similar EM fields on the surface as the dominant mode in the TEM-like waveguide used in the simulations,9 both of the measured and simulated results should agree with each other in our intended frequency range. The results are plotted in Fig. 3(d), exhibiting the DC-voltage-dependent absorbing performance from about 1.55 to 1.65 GHz. At 1.58 GHz, for example, the absorption varies from 0.49 to 0.98 when the voltage decreases from 6 to 0.5 V. Differences between simulations and measurements happen as the frequency goes higher than those of our interest, i.e., the measured absorptions remain large, while the simulated ones drop. This can be attributed to the different structures used in the two cases. At these frequencies, the dominant mode in the TE waveguide cannot propagate through in the presence of the prototype; while the transmission in the TEM-like waveguide is believed to be higher order modes that propagate through the free space above the surface, which is out of the scope of this research.

If the DC circuits at the bottom layer are replaced by sensors that are able to output DC voltage signals, as illustrated in Fig. 4(a), the absorbing performance should be controlled by corresponding variables. Here, the microwave power sensors are employed as an example, which transform microwave signals ranging from 0.01 to 1 W into DC voltages from about 10 V to 1.8 V within a frequency range from 1.63 to 1.72 GHz.14 It is expected that, when the surface wave power increases, the voltage across the varactors drops accordingly, resulting in rising capacitance and thus an increase of absorption. The power-dependent performance is similar to Wakatsuchi's and Kim's designs, but obviously based on different principles.

FIG. 4.

(a) The configuration of the cell integrated with a sensor. (b) Pictures of a prototype embedded with microwave power sensors. (c) Measured absorption versus frequency as a function of surface wave power. (d) Measured absorption versus the power level at various frequencies.

FIG. 4.

(a) The configuration of the cell integrated with a sensor. (b) Pictures of a prototype embedded with microwave power sensors. (c) Measured absorption versus frequency as a function of surface wave power. (d) Measured absorption versus the power level at various frequencies.

Close modal

A prototype shown in Fig. 4(b) is tested inside the WR430 waveguide. The VNA is set to be working in the external test set mode.14 Signals generated from its source port are amplified by an Ophir 5022 RF power amplifier and then injected into the waveguide through a directional coupler. The accurate input, reflected, and transmitted signals are sent back to the VNA, respectively, through 30 dB attenuators. The input power varies from 0.1 to 1 W, and the resulting quantities versus frequency are shown in Fig. 4(c). As expected, the curves are very similar to the DC measurement curves shown in Fig. 3(d), showing that the absorption tends to increase with increasing power. The working frequency shifts slightly higher by 0.05 GHz, which can be accounted for by the impact of the sensors. The most convincing result occurs at 1.64 GHz, where the absorption increases from 0.65 to 0.94, as plotted in Fig. 4(d). It is also observed that, even with low power, the absorption is still about 0.6 due to the lossy circuit components with parasitic resistances and the substrates as well, which could be improved by utilizing more efficient components. Nevertheless, the tunability of the absorber by the surface wave power is clearly demonstrated.

In conclusion, the absorption-tuning mechanism of an impedance surface loaded with varactor diodes is investigated. The absorption is mainly due to the parasitic series resistance of the varactors and is strongly dependent on the reactance of the surface, which is determined by the capacitance of the varactors. Through changing the DC biasing voltage across the varactors, the reactance and thus the absorption can be adjusted. The tuning mechanism provides the absorbing property with potentials of environmental sensing if appropriate sensors are included. As an example, microwave power sensors are integrated into the surface, and the absorption is measured to be enhanced by the increasing surface wave power. Besides surface waves, this work can also potentially be extended to active absorbers for external plane waves.

This work was supported in part by the Science and Technology Innovation Foundation of Institute of Electronic Engineering, China Academy of Engineering Physics (Grant No. S20161301), in part by NSF (Grant No. 1306055), and in part by the China NSAF Fund (Grant No. U1230112).

1.
D. F.
Sievenpiper
,
IEEE Antennas Wireless Propag. Lett.
10
,
1516
1519
(
2011
).
2.
E.
Knott
,
Radar Cross Section Measurements
(
SciTech Publishing
,
Raleigh, NC
,
2006
).
3.
B. A.
Munk
,
Frequency Selective Surfaces: Theory and Design
(
Wiley-Interscience Publication
,
New York
,
2000
).
4.
Y.
Zou
,
L.
Jiang
,
S.
Wen
,
W.
Shu
,
Y.
Qing
,
Z.
Tang
,
H.
Luo
, and
D.
Fan
,
Appl. Phys. Lett.
93
,
261115
(
2008
).
5.
Y.
Chang
,
W.
Che
,
Y.
Han
, and
Y. L.
Chow
,
IEEE Antennas Wireless Propag. Lett.
13
,
1084
1087
(
2014
).
6.
J.
Lee
and
S.
Lim
,
Electron. Lett.
47
,
8
9
(
2011
).
7.
H.
Wakatsuchi
,
S.
Kim
,
J. J.
Rushton
, and
D. F.
Sievenpiper
,
Phys. Rev. Lett.
111
,
245501
(
2013
).
8.
H.
Wakatsuchi
,
S.
Kim
,
J. J.
Rushton
, and
D. F.
Sievenpiper
,
Appl. Phys. Lett.
102
,
214103
(
2013
).
9.
S.
Kim
,
H.
Wakatsuchi
,
J. J.
Rushton
, and
D. F.
Sievenpiper
,
Appl. Phys. Lett.
108
,
041903
(
2016
).
10.
11.
D.
Sievenpiper
,
L.
Zhang
,
R. F. J.
Broas
,
N. G.
Alexopolous
, and
E.
Yablonovitch
,
IEEE Trans. Microwave Theory Tech.
47
,
2059
2074
(
1999
).
12.
D.
Sievenpiper
,
J. H.
Schaffner
,
H. J.
Song
,
R. Y.
Loo
, and
G.
Tangonan
,
IEEE Trans. Antennas Propag.
51
,
2713
2722
(
2003
).
13.
D.
Sievenpiper
,
IEEE Trans. Antennas Propag.
53
,
236
247
(
2005
).
14.
Z.
Luo
,
X.
Chen
,
J.
Long
,
R.
Quarfoth
, and
D.
Sievenpiper
,
IEEE Trans. Antennas Propag.
63
,
1736
1745
(
2015
).
15.
Z.
Luo
,
X.
Chen
,
J.
Long
,
R.
Quarfoth
, and
D.
Sievenpiper
,
Appl. Phys. Lett.
106
,
211102
(
2015
).
16.
B.-Q.
Lin
,
S.-H.
Zhao
,
Q.-R.
Zheng
,
M.
Zhu
,
F.
Li
, and
H.-Y.
Zhang
,
Prog. Electromagn. Res., C
43
,
247
254
(
2013
).
17.
S.
Liu
,
H.-X.
Xu
,
H. C.
Zhang
, and
T. J.
Cui
,
Opt. Express
22
,
13403
13417
(
2014
).
18.
R.
Quarfoth
and
D.
Sievenpiper
, in
2011 IEEE International Symposium on Antennas and Propagation (APSURSI)
,
Spokane, WA
(
2011
), pp.
1498
1501
.