We describe a new electrochemical method for reversible, pump-free control of liquid eutectic gallium and indium (EGaIn) in a capillary. Electrochemical deposition (or removal) of a surface oxide on the EGaIn significantly lowers (or increases) its interfacial tension as a means to induce the liquid metal in (or out) of the capillary. A fabricated prototype demonstrates this method in a reconfigurable antenna application in which EGaIn forms the radiating element. By inducing a change in the physical length of the EGaIn, the operating frequency of the antenna tunes over a large bandwidth. This purely electrochemical mechanism uses low, DC voltages to tune the antenna continuously and reversibly between 0.66 GHz and 3.4 GHz resulting in a 5:1 tuning range. Gain and radiation pattern measurements agree with electromagnetic simulations of the device, and its measured radiation efficiency varies from 41% to 70% over its tuning range.

Reconfigurable radiofrequency (RF) electronics are critical to enabling adaptive, multifunctional radios for future wireless sensing and communications. Conventionally, reconfigurable components have employed switched circuit elements (e.g., diodes and varactors) at a few locations to modify the current distribution on the device and change its RF properties. However, such localized loading can realize only a limited number and range of states. To construct more versatile tunable systems, liquid metals have recently been used in a variety of reconfigurable microwave components—filters,1,2 frequency selective surfaces (FSS),3 and antennas.4–6 In these applications, the liquid conductors are pneumatically actuated via pumps or contact pressure to change RF current paths.1–6 While the enhanced control over the conductor length and location provided by a liquid conductor greatly enhances the tuning range of the devices, the introduction of pumps and microfluidic elements adds to system complexity and requires a closed fluid path, limiting the device topology.

In this letter, we demonstrate a reconfigurable antenna that uses only electrical potential to actuate a liquid metal in a capillary, without the need for mechanical pumps. Our approach is enabled by the ability to deposit or remove a thin surface oxide on the metal using only electrical bias. The presence of the oxide layer significantly lowers the interfacial tension of the metal and allows it to flow into capillaries, whereas the removal of the oxide does just the opposite. This new, pump-free mechanism for the control of the metal offers the same advantages of continuously variable lengths, but with easily integrated electrical control.

Gallium alloys have attracted attention for reconfigurable electronics because of their liquid state at room temperature and their non-toxicity compared to mercury.2 The alloy used in this work is EGaIn, the eutectic (lowest melting point composition) of gallium (75%) and indium (25%) with conductivity2 of 3.4 × 106 S/m. EGaIn reacts with air to form a surface oxide that can stick to surfaces including the inner walls of capillaries.7 This adhesion limits the ability to reconfigure the metal, but can be avoided by injecting the metal into capillaries pre-filled with electrolyte. The electrolyte forms a slip layer between the oxide and the walls of the capillary.8,9

The presence of this slip layer enables continuous electrowetting (CEW), which is an effective method for moving plugs of metal through capillaries using only voltage,10–12 as shown in Figure 1(a). However, CEW has at least three limitations: (1) The metal plugs typically have constant length. (2) The metal plugs move in the channel, making it difficult to couple directly to a fixed external RF feed. (3) It is challenging to generate sufficient pressure differential to inject metal from a reservoir into a capillary.

FIG. 1.

Illustrative comparison between (a) continuous electrowetting using an EGaIn plug and (b) electrically controlled capillarity (ECC) of an EGaIn filament. The yellow arrows represent the direction of EGaIn motion under the indicated bias polarity.

FIG. 1.

Illustrative comparison between (a) continuous electrowetting using an EGaIn plug and (b) electrically controlled capillarity (ECC) of an EGaIn filament. The yellow arrows represent the direction of EGaIn motion under the indicated bias polarity.

Close modal

In contrast to CEW, our approach (Figure 1(b)) changes the length of liquid metal in a capillary using voltage to withdraw it from the capillary and into a reservoir (or conversely, to inject it from the reservoir and into the capillary). Oxidation of the leading surface substantially lowers the interfacial tension of the metal.13 When tuned such that the Laplace pressure of the metal in the reservoir exceeds that of the metal in the capillary, the metal flows into the capillary and the electrical length increases. Reversing the polarity of the DC potential electrochemically reduces the oxide and returns the metal to a state of large tension (a process called “recapillarity”14 because of the importance of reduction reactions to induce capillarity). In the absence of the oxide layer, Laplace pressure moves the EGaIn towards the external metal reservoir, shortening the metal filament and filling the vacated space with electrolyte.

Combined, we call these two techniques “electrochemically controlled capillarity” (ECC). In contrast to CEW, this ECC enables enormous changes in surface tension (>500 mN/m without the oxide to near zero with it)15 that allows the metal to pump in and out of a reservoir.

We use ECC to reconfigure an EGaIn antenna continuously within a capillary. A monopole antenna radiates most effectively when its length is a quarter wavelength and the resonant impedance of the antenna matches the 50Ω system impedance. Thus, adjusting the physical length of the metal tunes the frequency continuously over a range that is limited primarily by the length of the capillary. If the design of the monopole allows a large ratio of lengths, the antenna can then tune over a very wide range of operating frequencies.

Figure 2 presents a schematic and photograph of the reconfigurable monopole antenna. A glass capillary filled with EGaIn spans an acrylic fixture horizontally with each end of the capillary held by a reservoir cut out of the fixture. EGaIn fills the lower reservoir and connects to an SMA (Subminiature Version A) connector by a short copper wire inserted into the reservoir. Electrolyte fills the upper reservoir to establish the DC current loop that controls the electrochemical actuation mechanism. The use of 1 M NaOH avoids build-up of excess oxide which could otherwise provide an impediment to flow.15 The DC voltage combines with the RF signal generated by the Vector Network Analyzer (VNA) through a bias tee. The combined port connects to the SMA connector and feeds the antenna via direct contact with the liquid metal reservoir. In addition, to choke off the RF signal from the bias lines, a conical RF inductor (inductance = 0.531 μH, Coilcraft, Cary, IL) connects the electrolyte to the external bias lines.

FIG. 2.

(a) Schematic of the tunable monopole antenna. The DC polarity shown is for the injection process. The polarity is inverted to withdraw the EGaIn from the channel. Orientation for the view: zenith angle (θ) is referenced from +z axis, azimuth angle-φ is referenced from +x axis to +y axis. (b) A photograph of the antenna, feed, and reservoir. The inset shows a zoomed image of the EGaIn-NaOH interface under no bias (oxide skin removed).

FIG. 2.

(a) Schematic of the tunable monopole antenna. The DC polarity shown is for the injection process. The polarity is inverted to withdraw the EGaIn from the channel. Orientation for the view: zenith angle (θ) is referenced from +z axis, azimuth angle-φ is referenced from +x axis to +y axis. (b) A photograph of the antenna, feed, and reservoir. The inset shows a zoomed image of the EGaIn-NaOH interface under no bias (oxide skin removed).

Close modal

A power supply generates a DC potential between the electrolyte and EGaIn reservoirs. Application of a positive DC potential to the metal injects EGaIn into the glass capillary and displaces the electrolyte in the channel. Reversing the voltage polarity relative to that in Figure 2 causes EGaIn to withdraw towards the reservoir. The speed of the movement of EGaIn varies with the applied bias and is faster in the withdrawal direction. Previous studies show that the withdrawal velocity can be as large as 20 cm/s and depends on the magnitude of the DC current.8 Here, the EGaIn withdrew from a glass capillary (0.7 mm inner diameter) at 3.6 mm/s using just −0.7 V. The same capillary requires a +7.7 V bias to inject the EGaIn at a much slower 0.6 mm/s rate. The larger voltage is necessary to drive oxidation of the surface and to overcome the potential drop through the electrolyte in the capillary. The presence of oxide, which only forms at positive bias, likely provides a partial mechanical impediment to flow and explains why injection is slower than withdrawal. The asymmetry of the actuation mechanism is further accentuated by the fact the EGaIn tends to withdraw slowly even in the absence of voltage due to the removal of the oxide skin by the electrolyte. A small positive voltage helps to hold the metal in a selected position.

To evaluate the performance and tuning range of the antenna, the VNA (Agilent E5071C) measures the reflection coefficient of the monopole in real time while injecting and withdrawing the EGaIn from the capillary with a DC voltage as in Figures 1 and 2. Although the tuning is continuous, we captured eleven discrete antenna states to compare the resonance frequencies during injection and withdrawal. In this experiment, the EGaIn antenna is first withdrawn from 75 mm (full length) to 65.5 mm by applying a 0 V DC bias and allowing the NaOH to reduce the oxide skin. The EGaIn is then held stable at 65.5 mm by applying a small positive DC voltage (with respect to the polarity shown in Figure 2) while the reflection coefficient data are stored. This process is repeated to obtain data at the six lengths shown in Figure 3(c). Finally, after withdrawing to 4 mm, the EGaIn is injected into the capillary with a positive DC voltage of up to +7.7 V. To verify the repeatability of the antenna tuning, reflection coefficient data are stored at the same lengths as during the withdrawal process. Figure 3(c) shows that the frequency response of the antenna is identical during withdrawal and injection as long as the length is the same, demonstrating that the tuning is consistent and reversible. Simulated reflection coefficient data are not shown, but closely match the measured results.

FIG. 3.

Reflection coefficient of the antenna, (a) lowest and (b) highest tuning range of the antenna, (c) lengths in the withdrawal and injection processes.

FIG. 3.

Reflection coefficient of the antenna, (a) lowest and (b) highest tuning range of the antenna, (c) lengths in the withdrawal and injection processes.

Close modal

The measured tuning range of the liquid metal antenna is from 0.66 GHz to 3.4 GHz with the length ranging from 75 mm to 4 mm resulting in a tuning ratio of 5.2:1. In comparison, varactor-based antennas typically have a smaller continuous tuning range due to the lumped nature of their loading reactance. For example, in Refs. 16–18, tuning ratios measure between 1.3 and 2.7 while requiring bias voltages as high as 30 V. The new approach described here requires a DC control voltage of less than 8 V but enables a much larger tuning ratio. The enhanced tuning range is primarily a function of the much larger change in the impedance achievable by shaping the antenna structure itself rather than changing only a lumped reactance at a point.

To characterize the 3D radiation pattern and efficiency of the antenna, we measured the tunable system in a Satimo SG 64 near-field measurement system at the Wireless Research Center of North Carolina (WRCNC). In the SG 64 system, the antenna under test (AUT) is placed in an anechoic chamber at the center of a ring of many dual-polarized antenna elements that sample the fields radiated by the AUT. The AUT is rotated 180° to measure the fields in all directions, and the total radiated power is calculated by integrating the radiated fields over a spherical surface.

Pattern measurements were conducted for two antenna states—a low frequency state at 55 mm length and a high frequency state at 17 mm length. Figure 4 plots the measured and simulated radiation patterns in the two states, 0.96 GHz and 2.5 GHz. The co-polar E-plane (y-z plane) patterns are as expected for a conventional monopole on a finite ground plane, and the measured co-polar H-plane (x-y plane) patterns are nearly omnidirectional in agreement with the simulation. The measured cross-polarization is slightly higher than in simulation, likely due to radiation and scattering from the bias lines. However, the cross-polarized gain is still more than 10 dB below the co-polar gain in the direction of maximum radiation.

FIG. 4.

Normalized radiation patterns for high frequency and low frequency states. High frequency state (2.5 GHz), antenna length = 17 mm (a) E-plane and (b) H-plane. Low frequency state (0.96 GHz), antenna length = 55 mm (c) E-plane and (d) H-plane.

FIG. 4.

Normalized radiation patterns for high frequency and low frequency states. High frequency state (2.5 GHz), antenna length = 17 mm (a) E-plane and (b) H-plane. Low frequency state (0.96 GHz), antenna length = 55 mm (c) E-plane and (d) H-plane.

Close modal

The measured antenna performance parameters are summarized in Table I. The measured total efficiency (including mismatch loss) is 41% in the low frequency state and 70% in the high frequency state, which is lower than the efficiency of a conventional monopole made with a solid conductor (>95%). Losses in this system could arise due to the low conductivity of the EGaIn itself or due to the lossy electrolyte. Because the efficiency of a non-tunable EGaIn dipole is quite high19 (90%), we expect that the primary source of the increased losses in the tunable system is the low conductivity electrolyte that enables actuation. To study the source of the losses, we used full wave electromagnetic simulations of the monopole in Ansys High Frequency Structure Simulator (HFSS) to compute the antenna's response to an excitation. Fields calculated from these simulations are used to generate the far field radiation patterns in Figure 4 and to compute the total radiated power by integrating the fields radiated in all directions.

TABLE I.

Measured antenna performance.

StatesLength (mm)Holding voltage (V)Resonance frequency (GHz)Total efficiencyaMaximum realized gain (dBi)b
Low freq. 55 1.6 0.96 41% 1.08 
High freq. 17 2.2 2.5 70% 3.4 
StatesLength (mm)Holding voltage (V)Resonance frequency (GHz)Total efficiencyaMaximum realized gain (dBi)b
Low freq. 55 1.6 0.96 41% 1.08 
High freq. 17 2.2 2.5 70% 3.4 
a

Total efficiency is the ratio of the total radiated power to the total power available to the antenna.

b

Realized gain is the power (per unit solid angle) radiated by the antenna in given direction, relative to the power radiated (per unit solid angle) by an isotropic antenna that radiates all of the total available power. The maximum realized gain is the largest of this quantity over all angles.

Even though the total volume of electrolyte is small, our simulations indicate that electrolyte losses are greater than Ohmic losses in the EGaIn. Figure 5 compares the simulated total efficiency of the EGaIn monopole with lengths of 17 mm and 55 mm and electrolyte conductivity (σe) varying from 0 to 25 S/m. When the electrolyte is modeled as an ideal insulator (σe=0 S/m), the simulated efficiency approaches 100%, indicating that the EGaIn introduces negligible loss, particularly at high frequencies when the short EGaIn length forms a smaller percentage of the total radiating element. As the NaOH conductivity is increased, the total efficiency falls rapidly. Although the expected conductivity20 of the 1 M NaOH used in the experiment is 15 S/m, our simulations suggest that a reasonable value to compute the efficiency is in the range of 5–15 S/m.

FIG. 5.

Measured and simulated total efficiency for different antenna states and values of electrolyte conductivity (σe). (a) Antenna length = 55 mm and (b) antenna length = 17 mm. The red circle indicates the measured total efficiency at resonance.

FIG. 5.

Measured and simulated total efficiency for different antenna states and values of electrolyte conductivity (σe). (a) Antenna length = 55 mm and (b) antenna length = 17 mm. The red circle indicates the measured total efficiency at resonance.

Close modal

Based on the rapid decrease of the efficiency as the electrolyte conductivity increases, choosing a very low conductivity electrolyte would seem well-advised. However, the electrolyte conductivity also impacts the tuning speed, as larger σe allows greater DC current to drive the electrochemical reaction that induces the motion of the EGaIn.14 Thus, the choice of electrolyte represents a fundamental tradeoff between the total efficiency and reconfiguration speed for a fixed DC voltage and capillary geometry.

This letter demonstrates a continuously tunable monopole antenna by shortening and lengthening a liquid metal filament using electrochemical reactions via a DC bias. This novel approach for reconfiguration reshapes the EGaIn by reducing or depositing the oxide skin on the surface of the metal to tune the surface tension and induce motion without mechanical pumps. Because a continuous path for fluid flow is not necessary, this electrical actuation approach enables new topologies that are not possible with a pumped system, such as direct feeding of the radiating element. The slow switching speed represents an area for improvement via further research.

For EGaIn monopoles with lengths between 75 mm and 4 mm, the measured resonance frequency tunes from 0.66 GHz to 3.4 GHz for a tuning ratio to 5.2:1, which is beyond the ratio obtained by switch or varactor-based antennas. Furthermore, the measured total efficiency ranges from 41% to 70%, which, while lower than a conventional monopole, presents a tradeoff between efficiency and versatility that is evident in most tunable systems.

M.D.D. acknowledges the support from NSF CAREER (CMMI-0954321).

1.
A. T.
Ohta
,
G.
Shuyan
,
L. B.
Jun
,
H.
Wenqi
, and
W. A.
Shiroma
, in
Proceedings of 2012 IEEE International Conference on Wireless Information Technology and Systems (ICWITS)
(
2012
).
2.
M. R.
Khan
,
G. J.
Hayes
,
S.
Zhang
,
M. D.
Dickey
, and
G.
Lazzi
,
IEEE Microwave Wireless Compon. Lett.
22
(
11
),
577
(
2012
).
3.
M.
Li
and
N.
Behdad
,
IEEE Trans. Antennas and Propag.
60
(
6
),
2748
(
2012
).
4.
A. M.
Morishita
,
C. K. Y.
Kitamura
,
A. T.
Ohta
, and
W. A.
Shiroma
,
IEEE Antennas Wirel. Propag. Lett.
12
,
1388
(
2013
).
5.
A.
Dey
,
R.
Guldiken
, and
G.
Mumcu
, in
Proceedings of the IEEE Antennas and Propagation Society International Symposium (APS/URSI)
(
2013
), p.
392
.
6.
D.
Kim
,
R. G.
Pierce
,
R.
Henderson
,
S. J.
Doo
,
K.
Yoo
, and
J.-B.
Lee
,
Appl. Phys. Lett.
105
(
23
),
234104
(
2014
).
7.
M. D.
Dickey
,
R. C.
Chiechi
,
R. J.
Larsen
,
E. A.
Weiss
,
D. A.
Weitz
, and
G. M.
Whitesides
,
Adv. Funct. Mater.
18
(
7
),
1097
(
2008
).
8.
Mo. R.
Khan
,
C.
Trlica
,
J.-H.
So
,
M.
Valeri
, and
M. D.
Dickey
,
ACS Appl. Mater. Interfaces
6
(
24
),
22467
(
2014
).
9.
C.
Koo
,
B. E.
LeBlanc
,
M.
Kelley
,
H. E.
Fitzgerald
,
G. H.
Huff
, and
A.
Han
, “Manipulating Liquid Metal Droplets in Microfluidic Channels With Minimized Skin Residues Toward Tunable RF Applications,”
J. Microelectromech. Syst.
(in press).
10.
G.
Beni
,
S.
Hackwood
, and
J. L.
Jackel
,
Appl. Phys. Lett.
40
(
10
),
912
(
1982
).
11.
R. C.
Gough
,
A. M.
Morishita
,
J. H.
Dang
,
Hu.
Wenqi
,
W. A.
Shiroma
, and
A. T.
Ohta
,
IEEE Access
2
,
874
(
2014
).
12.
S.-Y.
Tang
,
V.
Sivan
,
K.
Khoshmanesh
,
A. P.
O'Mullane
,
X.
Tang
,
B.
Gol
,
N.
Eshtiaghi
,
F.
Lieder
,
P.
Petersen
,
A.
Mitchell
, and
K.
Kalantar-zadeh
,
Nanoscale
5
(
13
),
5949
(
2013
).
13.
M. R.
Khan
,
G. J.
Hayes
,
J.-H.
So
,
G.
Lazzi
, and
M. D.
Dickey
,
Appl. Phys. Lett.
99
(
1
),
013501
(
2011
).
14.
M. R.
Khan
,
C.
Trlica
, and
M. D.
Dickey
,
Adv. Funct. Mater.
25
(
5
),
671
(
2015
).
15.
M. R.
Khan
,
C. B.
Eaker
,
E. F.
Bowden
, and
M. D.
Dickey
,
Proc. Natl. Acad. Sci. U. S. A.
111
(
39
),
14047
(
2014
).
16.
N.
Behdad
and
K.
Sarabandi
,
IEEE Trans. Antennas Propag.
54
(
2
),
409
(
2006
).
17.
C. R.
White
and
G. M.
Rebeiz
,
IEEE Trans. Antennas Propag.
57
(
1
),
19
(
2009
).
18.
Z.
Shaozhen
,
D. G.
Holtby
,
K. L.
Ford
,
A.
Tennant
, and
R. J.
Langley
,
IEEE Trans. Antennas Propag.
61
(
4
),
2301
(
2013
).
19.
J.-H.
So
,
J.
Thelen
,
A.
Qusba
,
G. J.
Hayes
,
G.
Lazzi
, and
M. D.
Dickey
,
Adv. Funct. Mater.
19
(
22
),
3632
(
2009
).
20.
H.
Bianchi
,
H. R.
Corti
, and
R.
Fernández-Prini
,
J. Solution Chem.
23
(
11
),
1203
(
1994
).