A dual-band microwave sensor for glucose measurements utilizing an enclosed split ring metamaterial-based array

Diabetes is currently a major public health concern, partly exacerbated by the recent outbreak of coronavirus. Most of the published EM-wave based glucose sensors of this date allow a glucose concentration to be determined through a resonance frequency shift, inevitably with a questionable accuracy. To overcome the accuracy problem, a dual-band glucose sensor with dimensions 50 mm × 20 mm is proposed in this work to enable a glucose concentration to be measured at one resonance frequency band and cross-checked at another. An array of split-ring resonators (SRRs) was fabricated at a rectangular sensing area on the top surface of a 0.3 mm thick PET substrate, forming a metasurface with dual resonance bands at 4.5 GHz and 9.2 GHz. The backside of the PET substrate was fabricated with a defected ground plane designed to suppress the Q-factor associated with 4.5 GHz while leaving the Q-factor associated with 9.2 GHz unchanged. During a glucose concentration measurement, a drop of glucose solution was applied to the rectangular metasurface sensing area. The glucose concentration was determined in the form of a resonance frequency shift of the reflection coefficient at 4.5 GHz and a magnitude change of the reflection coefficient at 9.2 GHz. Consistent with our theoretical prediction, the fabricated sensor has indeed exhibited a dual resonant band characteristics, with one resonance occurring at 4.5 GHz and the other at 9.2 GHz. By measuring the reflection coefficient near 4.5 GHz, a positive and linear correlation in the log scale was observed between the glucose concentration and the resonant frequency shift with a sensitivity of 0.6 MHz/(mgdL⁻¹). At 9.2 GHz, there was no significant resonant frequency shift with varying glucose concentrations, but the magnitude of the reflection coefficient changed with the glucose concentration nonlinearly in an amount-dependent manner, with a sensitivity of 16.6 dB per unit glucose concentration within the clinical diabetic range. Overall, the log scale of the glucose concentration has exhibited a positive and linear correlation within the clinical diabetic range with both the resonant frequency shift at 4.5 GHz and the magnitude change at 9.2 GHz, thereby allowing the glucose concentration to be measured at 4.5 GHz and further cross-checked at 9.2 GHz at the same time.