Functionalized EGaIn Electrodes with Tunable Reduced‐Graphene‐Oxide Assembled EGaIn Core–Shell Particles for Soft and Deformable Electrochemical Biosensors

Background

EGaIn is attractive for soft electronics because it is a deformable metallic conductor, but its surface forms a gallium oxide layer about 1-3 nm thick. In aqueous electrochemistry, this insulating oxide impedes electron transfer, the electrode produces large gallium redox currents that bury analyte signals, and removal of the oxide changes the liquid metal's shape. Particles of carbon nanomaterials or metals can stabilize EGaIn in solution, but for biosensing they must be patterned on a conductive collector without delaminating during bending and stretching. An earlier soft supercapacitor from Min-gu Kim and co-workers bonded oxidized carbon nanotubes to patterned EGaIn, but such electrodes lack the electronic conductivity and electrocatalytic activity needed for biosensors.

Design and approach

Bulk EGaIn is broken into submicron particles by probe sonication in an acidic graphene oxide (GO) dispersion. The acid removes the gallium oxide, GO assembles on the particle surface, and gallium reduces it to reduced graphene oxide (RGO) because GO has a higher electrochemical potential (+0.48 V) than gallium (−0.572 V). The resulting REGs can be further decorated with platinum or gold by galvanic replacement, giving metal-decorated REGs (M-REGs). The particles are drop-cast onto an EGaIn current collector (ECC) stamped on a polydimethylsiloxane (PDMS) substrate to form REG-integrated deformable EGaIn electrodes (RIDE) and metal-decorated versions (M-RIDE). The authors attribute adhesion without a binder to hydrogen bonding and electrostatic interaction between residual oxygen groups on RGO and the oxide on the collector, to fusion of ruptured particles with the collector during drying, and, for metal shells, to reactive wetting between EGaIn and the metal. The underlying ECC carries current, while the RGO and metal shells provide a stable, catalytic sensing surface.

How it was evaluated

Particles were characterized by scanning and transmission electron microscopy, electron diffraction, energy dispersive spectroscopy, Raman and X-ray photoelectron spectroscopy. Electrodes were stretched by about 30% and blown with nitrogen to test adhesion, and sheet resistance was measured under strain with and without the collector. Electrochemical behavior was compared with bare EGaIn electrodes by cyclic voltammetry in phosphate buffer solution (PBS) and with 5 mM ferri/ferrocyanide, including after repeated 30% stretch-release cycles. Ascorbic acid (AA), dopamine (DA) and uric acid (UA) were measured simultaneously by differential pulse voltammetry in 50 mM PBS at pH 7.0. Glucose oxidase (GOx) was immobilized on the electrodes for chronoamperometric glucose detection, with selectivity tested against AA, acetaminophen, UA and lactic acid. All analyte tests were performed in buffer with three measurements per condition.

Key results

The REGs averaged 342 nm in size and showed a core-shell structure, a graphitic d-spacing of 0.38 nm and a rise in the Raman D/G ratio from 0.87 to 1.15, consistent with GO reduction; platinum and gold nanoparticles were confirmed on the shells by diffraction and elemental mapping. The REG layer did not delaminate from the collector after about 30% stretching and nitrogen blowing, and electrodes with the collector showed negligible resistance change at 30% strain, whereas REG films without it disconnected at 5% strain.

Bare EGaIn electrodes in PBS settled to oxidation currents below 10 μA and showed no ferri/ferrocyanide signal, while RIDE stabilized at a capacitive current of 140 μA after cyclic voltammetry pretreatment. Peak separations were 0.198 V for RIDE, 0.186 V for Pt-RIDE and 0.176 V for Au-RIDE, and the electrochemically active areas were 21.79, 27.14 and 27.24 mm2, respectively. Peak currents and potentials changed negligibly after repeated 30% stretch-release cycles.

With 100 μM AA, 10 μM DA and 15 μM UA, RIDE showed separate oxidation peaks at −0.038, 0.131 and 0.264 V, a separation of 0.169 V between AA and DA and 0.133 V between DA and UA, whereas commonly used carbon electrodes showed signal interference. Linear ranges were 10-100 μM for AA, 1-10 μM for DA and 1.5-15 μM for UA, with limits of detection of 0.065, 0.105 and 0.068 μM and R2 = 0.996. Responses to 40 μM AA, 4 μM DA and 6 μM UA showed no significant loss over up to 200 cycles of 30% stretch-release.

GOx on RIDE without metal gave no significant current change with 5 mM glucose, indicating that the RGO surface did not effectively catalyze hydrogen peroxide oxidation in the tested potential range. With platinum, GOx-Pt-RIDE operated at 0.4 V and responded linearly over 100-1000 μM glucose with a reported sensitivity of 2.94 μA μM−1, a limit of detection of 65.82 μM and R2 = 0.996; GOx-Au-RIDE required 0.6 V and gave 2.63 μA μM−1, 56.02 μM and R2 = 0.997. Both retained their response to 0.5 mM glucose after 200 cycles of 30% stretch-release, and GOx-Pt-RIDE showed higher selectivity, which the authors attribute to its lower operating potential.

Open questions

The authors note that RIDE still produced a large oxidative current in the first voltammetric cycle, which they attribute to collector areas left uncovered and to EGaIn released from ruptured particles, so cyclic voltammetry pretreatment was used before measurements. REG films alone were not conductive enough to act as free-standing electrodes, and RIDE without platinum or gold did not effectively catalyze hydrogen peroxide oxidation for enzymatic glucose sensing in the tested potential range.

How it connects to other research

Fully solution-processed liquid-metal features reported in npj Flexible Electronics serve as highly conductive, stretchable conductors and interconnects; this study instead converts the liquid-metal surface into an electrochemical sensing interface by bonding RGO- and metal-coated EGaIn particles to the current collector.