By Topic

• Abstract

SECTION I

## INTRODUCTION

RECENTLY chemical sensors that can be used for biochemical analyses are attracting interest for applications in biochemical, environmental, and medical diagnosis. One of the most useful parameters in biochemistry is the hydrogen-ion concentration, or pH. Ion-sensitive field-effect transistors (ISFETs) are one of the leading pH sensors using semiconductor technologies [1], [2], [3], [4], [5]. Moreover, charge-transfer-type pH sensors that are based on the charge-coupled device (CCD) image sensor technologies are proposed [6].

The sensitivity of a pH sensor, such as an ISFET is described by the Nernst equation, and, theoretically, at room temperature, the maximum sensitivity is 59 mV/pH. In these pH sensors, amplification noise decreases the signal-to-noise ratio (SNR) when amplifying output signals [7]. On the other hand, in the charge-transfer-type pH sensors, SNR is increased using the charge-accumulation technique when amplifying output signals [8], [9]. Moreover, the charge-transfer-type pH sensor was applied as a 32×32 arrayed pH imaging sensor and as a fused sensor for pH and optical imaging [10], [11], [12], [13].

SECTION II

## pH SENSING USING ISFETS

An ISFETs source and drain are constructed the same as those for a metal-oxide-semiconductor field-effect transistor (MOSFET). The reference electrode that acts like the gate electrode is separated from the channel by an ion selective membrane and the sample solution, as shown in Fig. 1. Fig. 2 shows plots of drain current $I_{\rm ds}$ versus reference electrode-to-source voltage $V_{\rm gs}$. Threshold voltage $V_{t}$ shifts according to the Nernst equation, which is given by TeX Source $$V_{{\rm gate}}={\rm const.}-2.303\times{{{\rm RT}}\over{F}}\times{\rm pH}\eqno{\hbox{(1)}}$$ where $V_{\rm gate}$ is the electrochemical potential between the sample solution to the ion selective membrane; $R$, the gas constant; $T$, the absolute temperature; and $F$, the Faraday's constant. Moreover, the rate of change in $V_{t}$ with respect to the pH of the sample solution is affected by the material of the ion selective membrane and given by TeX Source $${{\partial V_{t}}\over{\partial{\rm pH}}}=2.303\times{{{\rm RT}}\over{F}}\times{{\beta}\over{\beta+1}}\eqno{\hbox{(2)}}$$ where $\beta$ is the sensitivity coefficient. Typical materials used the ion selective membrane are ${\rm SiO}_{2}$ (silicon dioxide), ${\rm Si}_{3}{\rm N}_{4}$ (silicon nitride), ${\rm Al}_{2}{\rm O}_{3}$ (aluminum oxide), and ${\rm Ta}_{2}{\rm O}_{5}$ (tantalum oxide) [14], [15], [16], [17], [18], [19].

Fig. 1. Difference in structure between MOSFET and ISFET. An ISFETs source and drain are constructed the same as those of a MOSFET.
Fig. 2. Output curve of an ISFET. Plots show drain current $I_{\rm ds}$ versus reference electrode-to-source voltage $V_{\rm gs}$.
SECTION III

## pH SENSING USING CHARGE-TRANSFER-TYPE pH SENSORS

The structure of a charge-transfer-type pH sensor is based on a CCD image sensor, as shown in Fig. 3. An input control gate (ICG) and a transfer gate (TG) are gate electrodes that control the flow of signal electrons, and the sensing area consists of three layers: ${\rm Si}_{3}{\rm N}_{4}$, ${\rm SiO}_{2}$, and a silicon substrate.

Fig. 3. Cross-sectional structure of a charge-transfer-type pH sensor. The structure is based on CCD image sensor technologies.

Fig. 4 shows a potential diagram for the measurement of the pH signal. A potential well is formed at the surface of the silicon substrate beneath the sensing area. As the potential in the sensing area varies, the electrochemical potential in the ${\rm Si}_{3}{\rm N}_{4}$ film changes, and the depth of the potential well in the silicon substrate beneath the sensing area varies. As the potential increases, the depth of this potential well increases. The depth of the potential well is defined by the following equation: TeX Source $$\Delta V_{{\rm sig}}=\left({V_{{\rm ref}}+V_{{\rm gate}}}\right)-V_{{\rm ICG}}\eqno{\hbox{(3)}}$$ where $V_{\rm ref}$ is the reference electrode voltage; $V_{\rm gate}$, the electrochemical potential between the sample solution and the ion selective membrane; and $V_{\rm ICG}$, the ICG electrode voltage. Then, capacitance $C_{\rm sens}$ is formed by the potential well beneath the sensing area.

Fig. 4. Procedure for pH measurement (conduction band structure). (a) Clock cycle is initiated by resetting the FD. (b) ID is pulsed briefly, and electrons flow into the potential well. (c) Signal electrons $Q_{\rm sig}$ are accumulated in $C_{\rm sens}$. (d) TG is turned on, and $Q_{\rm sig}$ is transferred to the FD. (e) FDs potential decreases with $\Delta V_{\rm out}$. (f) Output signal with the charge-accumulation technique.

The clock cycle is initiated by resetting the floating diffusion (FD), as shown in Fig. 4(a). Moreover, the constant voltage is applied to the ICG during a pH measurement. Next, the input diode (ID) is pulsed briefly. The IDs potential then decreases below the ICGs potential, and electrons flow into the potential well, as shown in Fig. 4(b). To reset the ID, signal electrons $Q_{\rm sig}$ are accumulated in $C_{\rm sens}$, as shown in Fig. 4(c). To balance the potential beneath the sensing area, extra electrons overflow from the sensing area to the ID. Next, the TG is turned on, and $Q_{\rm sig}$ is transferred to the FD, after which the TG is turned off again, as shown in Fig. 4(d). Finally, the FDs potential decreases with $\Delta V_{\rm out}$, as shown in Fig. 4(e). Remarkably, by repeating this process without resetting the FD, the output signal is amplified, as shown in Fig. 4(f), and the amplified output signal is given by TeX Source $$\Delta V_{{\rm out}}=N\times\Delta V_{{\rm sig}}\times{{C_{{\rm sens}}}\over{C_{{\rm FD}}}}\eqno{\hbox{(4)}}$$ where $C_{\rm FD}$ is the capacitance of the FD region, and $N$ is the repeat count. This process is named the charge-accumulation technique.

Fig. 5 shows plots of $\Delta V_{\rm out}$ versus $V_{\rm ref}$. As shown in Fig. 5(a), $V_{t}$ increases with the pH value, and $C_{\rm FD}$ limits the maximum value of $\Delta V_{\rm out}$. When using the charge-accumulation technique, the slope of an output signal becomes steep with an increase in the accumulated count, as shown in Fig. 5(b). In other words, pH sensitivity is improved directly with the accumulated count.

Fig. 5. Output curve of a charge-transfer-type pH sensor. Plots show output voltage $\Delta V_{\rm out}$ versus reference electrode voltage $V_{\rm ref}$. (a) pH sensing without the charge-accumulation technique. (b) pH sensing with the charge-accumulation technique.
SECTION IV

## OCCURRENCE FACTOR OF QUASI-SIGNAL

Fig. 6 shows the results of pH measurement using the charge-accumulation technique. The pH signal is obtained using a standard buffered solution with a pH value of 6.86. Blue and red plots show the results with one and ten accumulated counts, respectively. In these results, the leading edges of the output curves are distorted because of the quasi-signal. The change of the sensor output which occurs with electrical charges other than $Q_{\rm sig}$ is the quasi-signal. Moreover, the red plots have a larger quasi-signal than the blue plots. In other words, a pH measurement becomes more difficult by increasing the accumulated count. To take full advantage of the charge-accumulation technique's strengths in pH measurements, it is necessary to reduce the quasi-signal.

Fig. 6. Results of pH measurement using conventional charge-accumulation technique. Blue and red plots show the results with one and ten accumulated counts, respectively.

The quasi-signal is attributed to the following two main factors:

1. potential barriers between the ICG and the sensing area;
2. short electron spill time from the sensing area to the ID.

When potentials beneath the ICG and the sensing area balance, these problems occur. They are described below, respectively.

First, factor 1 is described. In the fabrication process of the charge-transfer-type pH sensor, an ion selective membrane of ${\rm Si}_{3}{\rm N}_{4}$ covers the gate electrodes, as shown in Fig. 7. ${\rm Si}_{3}{\rm N}_{4}$ is filmed on the lateral side of the gate electrodes, and then the capacitance beneath the sensing area is formed nonuniformly. Therefore, a potential well is formed by the potential barrier, and the quasi-signal is accumulated during the pH measurement procedure, as shown in Fig. 7(a). On the other hand, ideally, there should be no potential barriers, as shown in Fig. 7(b). Moreover, since potentials beneath the TG and the sensing area never balance, the potential barrier does not appear between the TG and the sensing area.

Fig. 7. Quasi-signal caused by potential barriers between the ICG and the sensing area. (a) Nonideal condition. (b) Ideal condition.

Next, factor 2 is described. To balance the potential beneath the sensing area, it is necessary to spill electrons from the sensing area to the ID, as shown in Fig. 4(b) and (c). However, the spill time should become shorter with increasing sensing speed. A short spill time causes remnant electrical charges, such as the quasi-signal, as shown in Fig. 8. A SPECTRA device simulator was used to illustrate the electron density beneath the sensing area. The SPECTRA simulator is a 3-D, two-carrier software package that is used to simulate the electrical behavior of devices, either in a steady state or under transient conditions [20], [21]. Fig. 9 shows the simulated plots of the electron density beneath the sensing area versus the spill time. The difference in line color shows the difference in $\Delta V_{\rm sig}$ with an increment of 0.2 V between 0.0 and 2.0 V. From these results, the spill time should be at least 100 $\mu{\rm s}$, where $\Delta V_{\rm sig}$ is 0.0 V, and then the shortest time depends on $\Delta V_{\rm sig}$. To increase the sensing speed, the spill time should be shorter. Therefore, remnant electrical charges should be reduced by a means other than increasing the length of the spill time.

Fig. 8. Quasi-signal caused by short electron spill time from the sensing area to the ID.
Fig. 9. Simulated plots of the electron density beneath the sensing area versus the spill time. The difference in line color shows the difference in the depth of potential well $\Delta V_{\rm sig}$.
SECTION V

## REDUCTION OF QUASI-SIGNAL

Fig. 10 shows the concept behind the reduction in the quasi-signal. Only the linear region of the output curve is used for pH measurement by reduction in the nonlinear region. As mentioned above, $C_{\rm FD}$ limits the maximum value of $\Delta V_{\rm out}$. The novel charge-transfer-type pH sensor, which has an additional gate electrode on a sensing area for reducing the quasi-signal, is proposed. Fig. 11(a) and (b) are the cross-sectional view and the top view of the novel structure, respectively. This structure is named the quasi-signal-trap (Q-trap).

Fig. 10. Concept of the reduction of the quasi-signal. Only the linear region of the output curve is used for pH measurement by reducing the nonlinear region.
Fig. 11. Cross-sectional structure of a proposed charge-transfer-type pH sensor that has a Q-trap. (a) Cross-sectional view. (b) Top view.

Fig. 12 shows the mechanism of the reduction of quasi-signal using the Q-trap. As shown in Fig. 12(a) and $({\rm a}^{\prime})$, a potential dip is formed beneath the Q-trap by increasing the Q-trap voltage, and then the quasi-signal caused by the potential barriers is collected in the potential dip. Thus, remnant electrical charges that induce a quasi-signal are not transferred to the FD. In the same way, the quasi-signal caused by the short spill time is collected in the potential dip, as shown in Fig. 12(b) and (b'). Moreover, the spill time becomes shorter using the Q-trap, which will increase the sensing speed.

Fig. 12. Mechanism of the reduction of a quasi-signal using a Q-trap. (a) Quasi-signal caused by the potential barriers, and (a') the reduction mechanism. (b) Quasi-signal caused by the short electron spill time, and (b') the reduction mechanism.
SECTION VI

## VERIFICATION OF A Q-TRAP

The novel charge-transfer-type pH sensor with a Q-trap was successfully fabricated using 2-$\mu$-rule, single-poly-silicon, single-metal CMOS process technology. Fig. 13(a) and (b) show a computer aided design (CAD) layout and a photomicrograph of the fabricated sensor, respectively. The sensor size is 22.05×22.05 $\mu{\rm m}^{2}$, and the region that is enclosed by the red line is a Q-trap. The ion selective membrane is formed by the low pressure chemical vapor deposition method.

Fig. 13. (a) CAD layout and (b) photomicrograph of the fabricated sensor. The sensor size is 22.05×22.05 $\mu{\rm m}^{2}$, and the region that is enclosed by the red line is the Q-trap.

Fig. 14 shows the effectiveness of our approach for the reduction of the quasi-signal when using the devised sensor. The pH signal is obtained using a standard buffered solution with a pH value of 6.86. Red and blue plots show the results, with and without using a Q-trap, respectively. Without using a Q-trap, a quasi-signal occurs on the leading edge of the output curve. In contrast, the quasi-signal is reduced using the Q-trap. These results show the effectiveness of the reduction when a Q-trap is implemented for conventional charge-transfer-type pH sensors.

Fig. 14. Effectiveness of our approach for the reduction of the quasi-signal when using the devised sensor. Red and blue plots show the results with and without a Q-trap, respectively.

Fig. 15 shows the effectiveness of the charge-accumulation technique using a Q-trap. Meanwhile, the constant voltage is applied to a Q-trap during a pH measurement. Blue and red plots show the results with one and ten accumulated counts, respectively. The pH signal is obtained using a standard buffered solution with a pH value of 6.86. In these results, when $V_{\rm ref}$ is 1.3 V, $\Delta V_{\rm out}$ is 90 mV when using one charge-accumulation cycle and 870 mV when using ten charge-accumulation cycles. The output voltage is approximately ten times larger than the output when using one charge-accumulation cycle, according to (4). The error is caused by the nonlinearity of a source follower, which is connected across the output line.

Fig. 15. Effectiveness of the charge-accumulation technique using a Q-trap. Blue and red plots show the results with one and ten accumulated counts, respectively.

Finally, the pH sensitivity is measured by using the devised sensor. The pH signal is obtained using three standard buffered solutions with pH values of 4.01, 6.86, and 9.18. Fig. 16 shows the measurement results using ten charge-accumulation cycles. Blue and red plots show the results with one and ten accumulated counts, respectively. When using the charge-accumulation technique, the pH sensitivity is approximately 252 mV/pH. From these results, the pH sensitivity is improved by a factor of approximately 6.

Fig. 16. Results of pH measurement with using ten charge-accumulation cycles. The pH sensitivity is approximately 252 mV/pH.
SECTION VII

## CONCLUSION

The results clearly demonstrate the successful use of the novel charge-transfer-type pH sensor for highly-sensitive pH detection. To devise this sensor, we utilized a Q-trap, which reduced the quasi-signal caused by the potential dip and the short spill time.

It was confirmed that the devised sensor had a pH sensitivity of 252 mV/pH for ten charge-accumulation cycles (the sensitivity depends on the capacitance of FD). The most familiar pH sensor using a semiconductor is the ISFET. The sensitivity of an ISFET is approximately 59 mV/pH at room temperature. The sensitivity of the devised sensor was four times higher than that of an ISFET. Thus, our approach paves the way for detecting an unknown biochemical reaction with a small pH variation.

## Footnotes

This work was supported in part by the Japan Society for the Promotion of Science Grant-in-Aid for Scientific Research (S) under Grant 24226010 and in part by the Grant-in-Aid for JSPS Fellows. The review of this paper was arranged by Editor A. M. Ionescu.

H. Nakazawa, R. Otake, M. Futagawa, F. Dasai, and M. Ishida are with the Integrated Circuit and Sensor System Group, Toyohashi University of Technology, Toyohashi 441-8580, Japan (e-mail: nakazawa-h@int.ee.tut.ac.jp; otake-r@int.ee.tut.ac.jp; futagawa@batonzone.tut.ac.jp; dasai-f@int.ee.tut.ac.jp; ishida@ee.tut.ac.jp).

K. Sawada is with the Integrated Circuit and Sensor System Group, Toyohashi University of Technology, Toyohashi 441-8580, Japan, and also with the Core Research for Evolutional Science and Technology, Japan Science and Technology Agency, Tokyo 102-0075, Japan(e-mail: sawada@ee.tut.ac.jp).

Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org.

## References

No Data Available

## Cited By

No Data Available

None

## Multimedia

No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
ISSN:
None
INSPEC Accession Number:
None
Digital Object Identifier:
None
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available