Energy-autonomous biosensing platform using supply-sensing CMOS integrated sensor and biofuel cell for next-generation healthcare Internet of Things

Figure 3 shows the circuit diagram of the first prototype. It consists of three parts: the fructose fuel cell, SCRO, and an inductive-coupling transmitter. By introducing digital-based SCRO, the required minimum supply voltage can be reduced. Besides, by introducing the current-driven inductive-coupling link, low-voltage operation can be possible.

Zoom In Zoom Out Reset image size

To minimize the supply voltage, a fully digital, current-driven architecture was employed. Implementing the proposed architecture using zero-V_th transistors enables a low-supply voltage of less than 0.4 V to be used. Zero-V_th transistors are commonly available on both the state-of-the-art and legacy CMOS technologies. Thus, our technique can be feasible even in other technologies. Because the supply-sensing scheme is unsuitable for pulse amplitude modulation owing to its nature, time-domain modulation must be employed. To minimize power consumption, pulse-interval modulation (PIM) was employed in this work as shown in Fig. 4.

Zoom In Zoom Out Reset image size

In the proposed supply-sensing biosensor platform, the bio fuel cell has two functions: energy harvesting and front-end sensing. Typical bio fuel cells can generate a voltage of less than 0.4 V.^4–6) Thus, to realize self-powered operation without large and expensive power management circuits such as up-converters, the circuits must operate with a supply voltage of less than 0.4 V.

In order for bio fuel cells to function as both a power source and sensing front-end, the anode and cathode must be designed carefully. Unlike typical biosensors based on one transducer, the proposed supply-sensing biosensor uses two transducers (anode and cathode). Thus, if the output power depends on unintended transducers, the proposed device cannot function as a sensor even if it functions well as power source.

For aiming practical application, glucose or lactate fuel cells are beneficial because it can be used for glucose or lactate monitoring. However, they are under development and not available for us. As a preliminary study, we have chosen fructose fuel cell. Though fructose is not available on the human body and it has no healthcare application, it is meaningful as a preliminary study. Since the energy generation mechanism of fructose fuel cell is similar to that of glucose and lactate fuel cells, this study will contribute the successive research and development.

In the case of our prototype fructose sensor, we use the following reactions: in the anode, the capability to product current depends on fructose concentration. In the cathode, the capability to product current depends on oxygen concentration. To sense fructose, the total output current must depend on not oxygen but fructose, which we achieve by adjusting the sizes of the anode and cathode.

The biofuel cell in this work is an enzymatic fuel cell. Figure 5(a) shows the conceptual image of the mechanism of power generation from the bio fuel cells. Two kinds of enzymes, d-fructose dehydrogenase (FDH; EC 1.1.99.11, 169.9 U mg⁻¹, ca. 140 kDa, from Gluconobactor, purchased from Toyobo Enzyme) and laccase (LAC; EC 1.10.3.2, 108 U mg⁻¹, ca. 60 kDa, from Trametes sp, purchased from Daiwa Kasei) are immobilized to anode and cathode, respectively. They can directly catalyze the oxidation of d-fructose and the reduction of dioxygen, respectively. We used the FDH as received without further purification. While, the LAC was purified by anion exchange chromatography with a DEAE-Toyoperal column.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It is really important for the bio fuel cells to be safe in human use from the viewpoint of practical application. These enzymes are safe and suitable for human use. By exploiting these enzymes, following chemical reactions are occurred.

$$\begin{align*} \text{Anode}&{:}\ \text{O$_{2}$} + \text{2H$^{+}$} \rightarrow \text{H$_{2}$O} \end{align*}$$

$$\begin{align*} \text{Cathode}&{:}\ \text{Fructose} \rightarrow \text{5-keto-$\rm{\small D}$-fructose} + \text{2H$^{+}$} \end{align*}$$

As a result of the above chemical reaction, electrons are transferred from the anode to cathode, and energy can be available at the load devices.

To achieve PIM, the supply voltage must be modulated using pulse intervals. To enable low-voltage operation, we implemented an SCRO. The SCRO consists of a normal ring oscillator with inverter cells. As the inverter cell, the standard cell using one NMOS and one PMOS from the foundry were utilized.

For the wireless transmitter, we implemented a current-driven, inductive-coupling transmitter. Considering the limited power budget, proximity communication without any security protection was adopted. Commonly-used electrical proximity communication can be categorized into two approaches: the capacitive-coupling links and inductive-coupling links. Figure 6 shows their conceptual operating principles. The voltage obtained at the receiver side of a capacitive-coupling link is determined by the ratio of the coupled capacitance to the total capacitance. Thus, a voltage that is higher than the transmit voltage cannot be received.

Zoom In Zoom Out Reset image size

In contrast, the voltage received in the inductive-coupling link is determined by a product of the slew rate of the transmit current and mutual inductance. A high received voltage can be obtained, even with a low supply-voltage transmitter. Bio fuel cells can generate larger currents at lower tages because of their characteristics;⁴⁾ thus, a current-driven inductive-coupling link is preferable.

There is a trade-off between power supply voltage and the technology node. The lowest supply voltage was 0.7 V³⁵⁾ for the clock-based synchronous inductive-coupling link, and none of the conventional proximity communications could satisfy the requirement for operation with bio fuel cells. This work achieved the lowest supply voltage using the most cost-competitive technology node. The inductive-coupling transmitter was designed as described in the literature.²¹⁾

To verify the effectiveness of the proposed approach, a test chip was fabricated using 0.25-µm CMOS technology with a nominal supply voltage of 2.5 V. Figure 7 shows a microphotograph of the test chip. The measurement setup is shown in Fig. 8.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 9 and 10 show the measured current, power consumption, and output frequency as a function of supply voltage. Operation with 0.23 V power supply was verified. A supply of 0.23 V is sufficiently low for self-powered operation using a bio fuel cell and is the lowest supply voltage ever reported for proximity communications. Because the drain current of a zero-V_th transistor is proportional to the square of gate–source voltage V_GS from 0 to 0.4 V and is proportional to VGS from 0.4 V, the current consumption characteristics change at 0.4 V.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To verify the effectiveness of the proposed platform, self-powered operation using an organic bio fuel cell^1,2) was demonstrated. The bio fuel cell has a cloth-like feature. The enzymes for energy generation were immobilized on the bio fuel cell. Figure 11 shows a summary of the measured performance of the bio fuel cell. Figures 11(a) and 11(b) are performance of the anode and cathode, respectively. Figure 11(c) is the overall performance including both anode and cathode. By changing the load resistance, each measurement point was obtained. Figure 11(d) shows the measurement setup. The measurement was performed using a three-electrode system (BSA 730C electrochemical analyzer).

Zoom In Zoom Out Reset image size

Figure 11 shows successful self-powered operation using the bio fuel cell. By dipping the bio fuel cell into a fructose solution, the circuit transmitted a magnetic field, and its waveform appeared on the oscilloscope. This work is the first demonstration of a self-powered CMOS-based proximity transmission using bio fuel cells and SCRO.

Figure 12 shows a summary of the self-powered operation. Figure 13(a) shows the output voltage and current of the bio fuel cell as a function of fructose concentration. The voltage and current increased as the fructose concentration increased. Figure 13(b) shows the measured pulse rate of the output magnetic field from the proposed biosensor platform. The pulse rate increased as the fructose concentration increased. From these measured results, we have successfully confirmed the feasibility of the proposed self-powered supply-sensing biosensor platform.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 15 shows a block diagram of the proposed bio-sensing system, which consists of four building blocks: biofuel cells, a digital-based gate leakage timer, a supply-insensitive temperature sensor, and an inductive-coupling link. On the transmitter side, all circuitry is powered by a biofuel cell without regulation and voltage boosting. Consequently, the circuitry needs to be able to tolerate low-power and low-voltage operation. The digital-based gate leakage timer is used for supply sensing, and converts the supply voltage V_DD to the pulse period, which is depends on the bio-concentration. Because the supply-sensing scheme is intrinsically unsuitable for pulse amplitude modulation, pulse-interval modulation (PIM) is employed. The supply-insensitive time-domain temperature sensor, which is used to calibrate the supplied voltage from the biofuel cells, generates a temperature-dependent time-domain output. The inductive-coupling transmitter transmits the signal to the receiver side.

Zoom In Zoom Out Reset image size

For proper operation, the relationship between the timer and temperature sensor needs to be understood. Figure 16 shows the required output properties of the timer and temperature sensor. Based on these requirements, it is necessary that these devices oppositely sensitive to the supply voltage and temperature. The principle of operation the system when the bio-concentration and temperature change overtime is shown in Fig. 17. The timer converts the supply voltage into a period with inherently temperature-insensitive properties. Subsequently, the time-domain temperature sensor outputs the temperature information with supply-voltage-insensitive properties. The combination of the temperature-insensitive timer and supply-insensitive temperature sensor enables reliable healthcare monitoring using biofuel cells.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The test transmitter chip of the proposed system is shown in Fig. 18(a), and was fabricated using 65 nm CMOS technology with an active area of 0.0032 mm². The advanced CMOS technology was used because it enabled the low-voltage and low-power operation of the transmitter in a small footprint. In practical applications, the transmitter sensor board can be stacked on top of the biofuel cell, and the occupied area will therefore be constrained to that of the biofuel cell.

Zoom In Zoom Out Reset image size

The asynchronous receiver circuit was fabricated with 0.25 µm CMOS technology and had an active area of 0.035 mm², as shown in Fig. 18(b). This legacy CMOS technology was selected for cost reasons. The receiver sensor board was focused on verification of the inductive-coupling communications method, and did not include circuit techniques, such as calibration or statistical processing, in this work.

To validate the design of the timer and temperature sensor, the outputs of both the supply and temperature sensor must follow the properties shown in Fig. 3. The characteristics of the test chips of the timer and temperature sensor were measured inside a constant temperature oven. The experimental setup is shown in Fig. 12. The measured samples of the timer and temperature sensor operated correctly.

Figure 19 shows the measured output frequency across five samples and temperature sensitivity of the timer. The frequency increased with the supply voltage. The measurements revealed that the average frequency was 1.83 kHz at a supply voltage of 200 mV with a standard deviation of 130 Hz (or 6.83%). An almost linear relationship between the frequency and supply voltage was confirmed at room temperature. Keysight 34465A was used to characterize the resolution of the timer (supply-sensing sensor). The resolution was 0.001 mV with the average sensitivity of 5410 ppm/mV over the supply voltage range of 200–400 mV, as shown in Fig. 19(a). On the other hand, from the result of Chip 1, the temperature sensitivities were 3200 and 1330 ppm/°C at supply voltages of 200 and 400 mV, respectively, as shown in Fig. 19(b). The frequency variation resulted in a voltage inaccuracy of 31 and 24 mV at supply voltages of 200 and 400 mV, respectively. The temperature impact on the frequency variation is mainly due to the temperature-related threshold voltages of the two inverters in the Schmitt trigger of the timer. Therefore, the frequency variation of the timer against temperature was reduced at the higher supply voltage with the nearly superthreshold region operated digital circuitry in the timer. Figure 19(c) shows the measured linearity of the timer. The voltage inaccuracy of the timer was within −36/+43 mV. Figure 20 shows the measured power consumption of the timer as a function of the supply voltage. The power consumption of the timer reached a minimum of 7.3 nW at a supply voltage of 200 mV with a standard deviation of 3.1 nW (or 31%). Figure 21 shows the measured setup and communication results. Figure 22 shows the measured power consumption.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In this work, we firstly demonstrate the potential of using a combination of low-power CMOS OOK transmitter and glucose fuel cell for the development of glass-free CGMS contact lens. The recent progress in the field of biofuel cells has enabled the fabrication of glucose fuel cells on silicon wafer.³⁷⁾ It is capable of achieving an output of 2.3 µW/cm² and an open-circuit voltage of 0.55 V, which is maintained for at least 48 h. In the fuel cell, glucose oxidation and oxygen reduction occurs in the anode and cathode, respectively. Under an oxygen-rich environment with oxygen-permeable contact lens, the output power depends on glucose concentration. Therefore, a self-powered independent CGMS, which does not require any external signal nor power can be developed by combining it with a supply-modulated OOK transmitter. By using the fuel cells as both the power source and sensing transducer from tear's glucose concentration, the costly off-chip antenna for wireless power delivery and sensing electrodes can be avoided, which dramatically reduces production cost. Besides, since human blood glucose concentration is guaranteed to be higher than a certain level, the power supply from glucose is expected to be more stable than wireless power delivery or other energy harvesting sources.

A test chip was fabricated in 55-nm DDC CMOS technology featuring low-leakage and low-voltage operation. Except for the cross-couple NMOSs in the LC power oscillator and the switches in the switched-capacitor voltage multiplier, all the transistors are DDC transistors. For the purpose of reducing the fabrication cost, an on-chip antenna was implemented. The footprint of the CMOS prototype is 0.36 mm² (0.6 mm × 0.6 mm), which is the same as that of the CMOS read-out chip in literature.⁹⁾

Figure 23 shows the conceptual image of the proposed CGMS. The entire system can be divided into two parts: Glucose fuel cell and CMOS chip. The size of the fuel cell is same as that of the CMOS chip. The miniaturization was achieved by introducing RIE.³⁸⁾ Since the output power depends only on the glucose concentration under an oxygen-rich environment, the fuel cell can be used as both the sensing transducer as well as the power source.³⁹⁾ For example, higher glucose concentration results in higher output power and higher transmission rate, and vice versa. This architecture eliminates the requirement of a sensor part, which requires a large area as in the prior art.⁹⁾ The CMOS chip is capable of supply-modulated OOK wireless transmission. The chip consists of on-chip-inductor-based boost-converter, self-oscillating-voltage-doubler-based control circuit, and LC power oscillator. Two chips are integrated in the PCB board, and the total size is less than 1 mm³.

Zoom In Zoom Out Reset image size

Figure 24 shows the schematics of the proposed circuit of a boost converter and an RF LC-based power oscillator including its simulated result. To improve compatibility with volume-limited contact lens application, an off-chip-inductor-based boost converter⁴⁰⁾ cannot be adopted. However, the conventional switched-capacitor-based converter⁴¹⁾ cannot satisfy our requirement from the viewpoint of leakage current. For reducing leakage current with a small volume, we implemented on-chip-inductor-based boost converter. In order to compensate low-efficiency of the boost converter due to low-performance of an on-chip inductor, the LC power oscillator is activated with much lower frequency than the boost converter as shown in the simulated waveform. Time resolution of several seconds is sufficient for CGMS application. Low-V_th NMOSs are implemented in the cross-couple part of LC power oscillator. The control signals are fed into by the control circuitry described as followings.

Zoom In Zoom Out Reset image size

Figure 25 shows the schematic and simulated waveform of the control circuit. The control circuit consists of FF-based clock divider, pulse generators, switched-capacitor-based voltage multipliers. A self-oscillating voltage doubler⁴²⁾ provides the boost conversion and clock signal. Since insufficient slew rate in activation signal (A) increases dynamic loss in the boost converter, high slew rate operation must be achieved. Therefore, the supply voltage of the control circuit was provided by the voltage doubler. This is also helpful for reducing the required number of stages of large-leakage switched-capacitor-based voltage multiplier. The clock divider generates sub-Hz activation signal (B), of which division ratio is determined to guarantee the oscillation of LC power oscillator.

Zoom In Zoom Out Reset image size

Figure 26 shows the operating principle of the developed CMOS-compatible glucose fuel cell.^7,47) Ideal glucose fuel cells obtain energy from the following chemical reactions:

$$\begin{align*} \text{Anode}&{:}\ \text{C$_{6}$H$_{12}$O$_{6}$} + \text{6H$_{2}$O} \rightarrow \text{6CO$_{2}$} + \text{24H$^{+}$} + \text{24e$^{-}$} \end{align*}$$

$$\begin{align*} \text{Cathode}&{:}\ \text{6O$_{2}$} + \text{24H$^{+}$} + \text{24e$^{-}$} \rightarrow \text{12H$_{2}$O} \end{align*}$$

$$\begin{align*} \text{Whole}&{:}\ \text{C$_{6}$H$_{12}$O$_{6}$} + \text{6O$_{2}$} \rightarrow \text{6CO$_{2}$} + \text{6H$_{2}$O} \end{align*}$$

However, glucose is not completely oxidized when its oxidation is mediated by Raney platinum⁷⁾ but mainly oxidized to gluconic acid by the following chemical reactions:

$$\begin{align*} \text{Anode}&{:}\ \text{C$_{6}$H$_{12}$O$_{6}$} + \text{H$_{2}$O} \rightarrow \text{C$_{6}$H$_{12}$O$_{7}$} + \text{2H$^{+}$} + \text{2e$^{-}$} \end{align*}$$

$$\begin{align*} \text{Cathode}&{:}\ \frac{1}{2}\text{O$_{2}$} + \text{2H$^{+}$} + \text{2e$^{-}$} \rightarrow \text{H$_{2}$O} \end{align*}$$

$$\begin{align*} \text{Whole}&{:}\ \text{C$_{6}$H$_{12}$O$_{6}$} + \frac{1}{2}\text{O$_{2}$} \rightarrow \text{C$_{6}$H$_{12}$O$_{7}$} \end{align*}$$

As shown above, glucose oxidation and oxygen reduction are respectively caused at the anode and cathode. The theoretical maximum OCV is 1.30 V.^45,48)

Zoom In Zoom Out Reset image size

The fabrication process of wafer-scale glucose fuel cell is shown in Fig. 27. The development procedure is based on previous study.⁷⁾ The fabrication procedure of the cells is outlined as follows:

Zoom In Zoom Out Reset image size

1) A 6-in. silicon wafer with 1-µm thermal oxide was prepared. We used a wafer with 625 µm thickness.

2) The electrode was prepared by depositing 2-nm-thick Ti and 100-nm-thick Pt layers. Ti was used as an adhesion layer.

3) A porous-Pt layer in the anode area was formed. The overall approach is etching Al from Al/Pt alloy. We deposited 100-nm-thick Al over Pt (deposited in 2) by evaporation and annealed to alloy the Al and Pt. After annealing, we etched the Al from Al/Pt alloy.^7,49)

4) We formed a 0.83% Nafion solution via $1:5$ dilution in isopropyl alcohol (IPA) from Nafion liquid dispersion (Nafion^® perfluorinated resin solution 5 wt % in mixture of lower aliphatic alcohols and water, contains 45% water, Sigma-Aldrich) and spin-coated it at 1000 rpm. After spin coating of the Nafion solution, we heated the wafer at 60 °C for 30 min and 120 °C for 30 min. Patterned by photolithography using 1.5-µm OFPR-800 and etched resist and Nafion layer O₂ ashing (reactive ion etching, RIE) until the resist disappeared. The Nafion layer existed after etching. The Nafion layer separates the anode and cathode by preventing the passage of anions through the membrane.

5) We formed CNT-based layer on the Nafion layer. The CNT [carbon nanotube, single-walled ≥95% carbon basis (≥99% as carbon nanotubes), 0.84 nm average diameter, Sigma-Aldrich] was mixed in 0.83% Nafion solution described in 4). The CNT-based solution was dropped on the wafer heated about 100 °C (±20 °C). Patterned by photolithography using 1.5-µm OFPR-800 and etched resist and Nafion layer O₂ ashing (RIE) until the resist disappeared. The CNT-based layer existed after etching. The difference between the method reported in Ref. 29 and the above process is the introduction of RIE instead of wet etching in the present process for the fabrication of the separator and cathode in step 4) and 5). The fabricated cells are in 0.36 mm² (0.6 mm × 0.6 mm) and the wafer was cut into 0.7-mm square including 0.6-mm device size and 0.1-mm margin. The microphotograph and measurement setup is shown in Fig. 28.

Zoom In Zoom Out Reset image size

The dependence of the measured output voltage and power on current is shown in Fig. 29. Figure 30 shows the dependence of the highest OCV in the measured samples. The measurement was performed with PBS-based 30-mM glucose concentration, which simulates the blood sugar level of diabetic patients in serious situation. Each point of the graph was measured by various load resistances. An OCV of 228 mV was verified. The power was 0.53 nW at peak with a load resistance of 2 MΩ, which corresponds to 1.32 µW/cm² considering an anode area of 0.04 mm². Considering the internal resistance of the tester (10 MΩ), the peak power was 5.2 nW.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 31 shows the correlation between the output voltage and glucose concentration. The output voltage get higher as glucose concentration increase. The PBS in the glucose solution resulted in an offset voltage when the glucose concentration was 0 mM. The sample cell used in this measurement was different from that shown previously.

Zoom In Zoom Out Reset image size

Figure 32 shows the performance variation of the developed glucose fuel cells. We measured 132 samples to evaluate the performance variation. Six samples achieved over 100 mV of OCV.

Zoom In Zoom Out Reset image size

Figure 33 illustrates testchip microphotograph, the measurement setup, and measured waveform of the supply current, and measured period of wireless transmission dependence on input supply voltage of the developed CMOS prototype. For simulating real-life environment with glucose fuel cell, only two signal are fed into the chip using a manual prober (Apollowave α100) and SMU (Keysight B2912A). The period of spike due to oscillation of LC power oscillator was increased by increment of input supply voltage from 320 to 400 mV. Periodical boost conversion whose period monotonically depends on the supply voltage was verified. This result implies the capability for supply-modulated OOK transmission.

Zoom In Zoom Out Reset image size

Figure 34 depicts the measured spectrum obtained by the developed CMOS prototype and its measurement setup. A 5-cm wireless transmission with 4.87 GHz center frequency was confirmed by the magnetic probe (LANGER MFA01), the amplifier (Beehive 150A), and the spectrum analyzer (Anritsu MS2830A). A wireless transmission of −34 dBm power under 10-dB attenuation was verified. The communication distance can be extended by introducing an external antenna.

Zoom In Zoom Out Reset image size

Figure 35 shows the performance comparison among the state-of-the-art low-power CMOS glucose sensor [1], a boost converter [5], and a pulsed transmitter with two input terminals [6]. Our proposed sensor outperforms the others in terms of current consumption while maintaining a competitive size. A current consumption of 6.1 nA at 0.32 V can be obtained using the 0.6-mm square glucose fuel cell [2,3]. Thus, this work demonstrates the feasibility of glass-free CGMS contact lens.

Zoom In Zoom Out Reset image size