Abstract

Titanium dioxide (TiO2) thin films were sputtered by radio frequency (RF) magnetron sputtering method and have been employed as the sensing membrane of an extended gate field effect transistor (EGFET) for pH sensing detection application. The TiO2 thin films were deposited onto indium tin oxide (ITO) coated glass substrates at room temperature and 200°C, respectively. The effect of deposition temperature on thin film properties and pH detection application was analyzed. The TiO2 samples used as the sensing membrane for EGFET pH-sensor and the current-voltage (I-V), hysteresis, and drift characteristics were examined. The sensitivity of TiO2 EGFET sensing membrane was obtained from the transfer characteristic (I-V) curves for different substrate heating temperatures. TiO2 thin film sputtered at room temperature achieved higher sensitivity of 59.89 mV/pH compared to the one deposited at 200°C indicating lower sensitivity of 37.60 mV/pH. Moreover the hysteresis and the drift of TiO2 thin film deposited at room temperature showed lower values compared to the one at 200°C. We have also tested the effect of operating temperature on the performance of the EGFET pH-sensing and found that the temperature effect was very minimal.

1. Introduction

TiO2 is a type of metal oxide material that has become an important engineering material due to the versatility of electrical and optical properties that have been used in various applications in many fields including dye-sensitized solar cell [1, 2], photocatalysts [3], working electrodes in semiconductor [4], dielectric layers material [5, 6], gas sensor [7], drug sensor [8], and also sensing membrane for pH detection [912]. TiO2 is an interesting material due to its advantages such as good chemical and temperature stability, high permittivity and refractivity index, high dielectric constant, and biocompatibility.

TiO2 has been widely used as a pH sensor in order to detect H+ ions. There are two types of electrochemical sensor that typically have been studied for pH sensing which are ion sensitive field effect transistor (ISFET) [9] and extended gate field effect transistor (EGFET) [10, 13]. The distinction between these two devices is that ISFET consists of field effect transistor (FET) with sensing membrane that was fabricated together with FET and in contact directly with the electrolyte, while EGFET consists of FET and sensing membrane that are separated into two parts and only the sensing membrane is in contact directly with the electrolyte while the FET is being isolated. EGFET configuration gives more advantages compared to ISFET like low cost, less sensitivity to light and temperature, and also simple packaging [14].

In this work, we report the application of TiO2 as the sensing membrane of an EGFET pH sensor and its dependence on the deposition temperature are discussed. The deposition temperatures are kept at low range due to the possibility of employing the process for flexible substrates in the future. Then, the physical properties of deposited thin films were examined and also the EGFET sensors were characterized for their transfer characteristics, hysteresis, drift, and temperature compensation test. Comparison between EGFET and ISFET was also simulated experimentally to prove the operating temperature stability of EGFET.

2. Methodology

2.1. Fabrication of TiO2 Sensing Membrane

The TiO2 sensing membrane was deposited on ITO coated glass that acts as the substrate. At the same time, the conductive layer of ITO was used to be connected to the gate of commercial transistor (NDP6060L). Prior to the deposition process taking place, ITO coated glass was cleaned in an ultrasonic bath for 10 min at 50°C using methanol and deionized water (DIW) alternately to assure the ITO is free from dirt and contamination. Afterward, ITO was dried by nitrogen gas before being ready to be used. Before the deposition process, the target was undergoing presputtered process for about 300 sec to eliminate any contamination on the target surface and stabilize the sputtering conditions. TiO2 thin films were then deposited on top of cleaned ITO coated glass by RF magnetron sputtering system (SNTEK RSP 5004). The TiO2 target had high purity of 99.99% and a diameter of 4 inches. The details of RF magnetron sputtering conditions for TiO2 thin films are summarized in Table 1. The field emission scanning electron microscope, FESEM (JEOL-JSM 7600F), and Raman spectroscopy (Jobin Yvon Horiba) were used to analyze the surface morphology and structural properties of thin films. Meanwhile, the element composition of TiO2 was observed by energy dispersive X-ray spectroscopy (EDX) and the surface roughness of TiO2 was analyzed by using atomic force microscopy (AFM).

2.2. EGFET Sensor: Measurement Setup
2.2.1. Transfer Characteristic (I-V) Measurement Setup

The deposited thin films were prepared for current-voltage (I-V) measurement. Before I-V measurements were carried out, the undeposited area of ITO was connected with metal wire and attached with silver paste. In order to avoid leakage current, the thin films were encapsulated with epoxy resin with 1 cm 1 cm of opening sensing area being defined at the deposited thin film. The completed sensing membrane was then connected to the gate of the commercial MOSFET (NDP 6060L). Then, the encapsulated thin film and reference electrode (Ag/AgCl) were immersed together in various pH buffer solutions of pH of 1.8, 4, 6.2, 7, 8.1, 10, and 12. As reported by [15], reference electrode is needed for proper operation of ISFET so that the electrolyte potential can be established with respect to the semiconductor substrate. By using Semiconductor Parametric Device Analyzer model B1500A, the transfer characteristics and drain current versus reference voltage () were obtained. In order to avoid light and environment effect, all measurements were performed in a dark box at room ambient. Figure 1 shows the current-voltage (I-V) measurement setup for EGFET using TiO2 sputtered sensing membrane.

2.2.2. Hysteresis and Drift Measurement Setting

The measurement system of hysteresis and drift consists of EGFET TiO2 sensing membrane, reference electrode, constant current constant voltage readout interfacing circuit (CVCC ROIC), data logger, and also a PC with signal recording software. For drift measurement, the deposited TiO2 sensing membrane and reference electrode were immersed in buffer solution of pH 4 for 12 hours and the output voltage of the TiO2 sensing membrane was recorded and stored by using a data logger. Meanwhile, a PC is used to display the graph of recording data. The measurement was repeated for various buffer solution of pH of 7, 10, and 12 for 12 hours, correspondingly. Afterward, using the same measurement setting of drift, the hysteresis curves were performed at pH loop of 7 4 7 10 7 and 7 10 7 4 7 for 500 sec.

2.2.3. EGFET and ISFET Comparison of Temperature Experimental Simulation

This experiment serves as the simulation to experiment the effect of operating temperature on the EGFET and ISFET sensor performance. As mentioned, for ISFET, since the sensing membrane is fabricated directly on the gate of the FET, the transistor will be immersed in the solution during measurement; thus the reading will be influenced by the temperature. On the other hand, for EGFET, only the sensing membrane which acts as the extended gate of the FET will be immersed during the measurement. In this work, we simulated ISFET, by immersing the sensing membrane and MOSFET with reference electrode in pH buffer solutions on top of the hot plate. The temperature of pH buffer solution varied from 24 to 70°C by increasing of 5°C per step. As for EGFET simulation, only the sensing membrane (extended gate) with MOSFET was immersed in pH buffer solutions on top of the hot plate for measurement and the temperature was varied in the same range. The measurement setup consisted of EGFET sensing membrane and reference electrode, commercial MOSFET, constant current constant voltage readout interfacing circuit (CVCC ROIC), digital multimeter, and a hot plate.

3. Result and Discussion

3.1. Physical Characteristics of the TiO2 Thin Films

Figures 2(a) and 2(b) show the surface morphologies of the TiO2 thin films deposited at room temperature and at 200°C, respectively. The images show that different deposition temperature results in different surface structure of thin films. Thin film deposited at room temperature shows a rough and nonuniform surface with mostly undistinguishable grain boundaries. On the other hand, the one at 200°C exhibits small grains and also shows the agglomeration of particles which has a more uniform surface structure. It might have happened due to the higher kinetic energy from the higher temperature of 200°C to restructure TiO2 particles. The material components of deposited TiO2 were confirmed by EDX characterization. The EDX shows that both deposited thin films contain titanium (Ti) and oxygen (O2) elements. Other elements that exist in the thin films like silicon (Si) and indium (In) were from the ITO glass substrate and the carbon (C) peak may originate from the environment contamination. The TiO2 thin film thickness measured by surface profiler for the thin film deposited at room temperature is 61.55 nm which is slightly thicker compared to that at 200°C which is 60 nm. It can be said that higher temperature induces more agglomeration and a denser film with regard to the film thickness direction resulting in a thinner and compact film. The surface roughness of deposited TiO2 thin films was analyzed by using atomic force microscopy (AFM), as shown in Figures 3(a) and 3(b). As the substrate temperature increased, the surface roughness of deposited thin films is decreasing. The roughness decreases from 7.123 nm to 2.841 nm as the substrate temperature increased from room temperature to 200°C, respectively. Via the site-binding model theory, the sensing performance would improve as the number of surface sites increased. The increasing size of grain could improve the properties of sensing performance when the deposited thin film was used as sensing membrane [16]. Figure 4 shows the Raman spectrum of the TiO2 thin films deposited at room temperature and at 200°C. The measurement was carried out at an excitation wavelength of 514.5 nm in the range of 200–1000 cm−1. Crystalline TiO2 known as anatase, rutile, and brookite will exhibit Raman shift peaks at 144, 447, and 153 cm−1 [1719]. However, from Figure 4, no sharp peak can be observed, which clearly indicates that both deposited TiO2 thin films have an amorphous structure.

3.2. EGFET Sensor Characterizations
3.2.1. Transfer Characteristics () and pH Sensitivity

Figures 5(a) and 5(b) show the transfer characteristics () of the sputtered TiO2 thin films in various buffer solutions of pH values between 1.8 and 12. Meanwhile, the corresponding versus pH plot is shown in the inset. The sensitivity and linearity of the sensor were derived from the slope and the linear regression of the graph, respectively. In this measurement, the drain voltage () was fixed at 500 mV while the reference voltage () was swept from −2 to 3 V. Both graphs show similar behaviour of transfer characteristics in which the threshold voltage shifts from left to the right. The shift of the threshold voltage to the right with increasing the pH value is due to the decreasing of the surface potential. The relation of surface potential voltage between sensing membrane and pH buffer solutions can be explained via the site-binding model theory [15]. Figure 6 shows the reactions that occur between the surface of sensing membrane and pH buffer solution. As can be seen from Figure 6, the surface of sensing membrane denoted as Ti can be in three different forms which are negative (TiO), positive (), and neutral (TiOH). The changes of the surface potential depend on the value of pH buffer solution. The dependence of surface potential voltage () between sensing membrane and pH buffer solution can be expressed as [20]where indicates the pH buffer solution value at the zero charge point, is the electron charge, is Boltzmann’s constant, is the absolute temperature, and is the sensitivity parameter. The relationship of and the surface sites per unit area could be expressed as [20]where and are acid equilibrium constant and basic equilibrium constant, respectively, and is the capacitance of the electrical double layer derived by the Gouy-Chapman-Stern model [21]. According to (1), higher linear response of surface potential voltage between sensing membrane and pH buffer solution could be obtained, thus resulting higher sensitivity parameter, β. Therefore, when was fixed, that applied to the commercial MOSFET (NDP 6060L) change with pH buffer solution values.

The sensitivity and linearity for both deposited TiO2 thin films of different substrate heating temperature are summarized in Table 2. The sensitivity and linearity of deposited TiO2 thin films were calculated from the output voltage taken from the corresponding point at = 100 µA. The room temperature deposited thin film shows sensitivity and linearity of 59.89 mV/pH and 0.93503, respectively. On the other hand, the sensitivity and linearity obtained by thin film at 200°C are 37.60 mV/pH and 0.95132. It is clear that the room temperature TiO2 thin film shows higher sensitivity and good linearity compared to that of 200°C. This result could be attributed to the higher surface roughness of TiO2 thin film deposited at room temperature compared to the one deposited at 200°C as observed in the previous surface morphology images (Figure 2) and also from AFM roughness analysis (Figure 3). As reported by many [2124], higher surface roughness gives better pH sensitivity. Although both thin films are amorphous as observed by Raman spectroscopy, it can be expected that the thin film deposited at room temperature has a lower degree of crystallinity compared to the sample deposited at 200°C which may facilitate more binding sites that relates to higher sensitivity [25]. By comparing to [26], Kao et al. used higher temperature of 800°C to obtain higher sensitivity. Meanwhile in this paper, we used lower temperature to produce higher sensitivity yet low cost.

3.2.2. Hysteresis and Drift Characteristics

A hysteresis characteristic of the sensor was measured for output offset voltage corresponding to the change in pH buffer solutions for specified time interval. The sensing membranes were dipped in the pH buffer solutions starting from pH loops of 7 4 7 10 7 and 7 10 7 4 7 for 500 sec in each pH value. Figures 7(a) and 7(b) show the typical hysteresis loops () and the hysteresis widths () are shown in Figures 7(c) and 7(d). As indicated in Figures 7(c) and 7(d), it is found that TiO2 thin films deposited at room temperature exhibit a lower output voltage compared to 200°C thin film. This is because of the higher surface roughness of TiO2 thin films deposited at room temperature, causing the reduction of memory effect [27]. Figure 7(c) demonstrates that the hysteresis width of TiO2 thin films deposited at room temperature is 5.3 mV and 9 mV in the pH 7 4 7 10 7 and 7 10 7 4 7 loops, respectively. Meanwhile, the hysteresis width for the one at 200°C is 7.9 mV and 13.8 mV in the same pH loops as shown in Figure 7(d). It is clearly seen that the acid-side hysteresis is smaller than the base-side hysteresis. According to Yao et al., the unsymmetrical hysteresis of pH sensor is due to the difference in the diffusion rates for H+ and OH ions into the buried sites of the sensing membrane [10]. The low hysteresis value indicates good characteristics of electrochemical sensor. This is because high value of hysteresis and drift will limit the accuracy obtained from the sensor [15]. The phenomena of drift and hysteresis could be attributed to the slow progressive hydration of the insulator film as well as to the transport of certain species throughout the insulator film, which affects the semiconductor-insulator interface [15].

The durability and reliability of electrochemical sensor were tested by using drift measurement. The drift measurement was completed in 300 Sec for each pH buffer solution. The drift characteristics of the TiO2 thin films measured for pH values of 4–12 plotted as output voltage over time are shown in Figures 8(a) and 8(b). The drift rates of room temperature sample in Figure 8(a) are 0.09766 mV/h, 2.6007 mV/h, 0.59152 mV/h, and 0.041692 mV/h for pH of 4, 7, 10, and 12, respectively. And those for the sample deposited at 200°C were 0.81103 mV/h, 4.4062 mV/h, 0.46916 mV/h, and 0.32618 mV/h. The sensing membrane deposited at room temperature has an obviously smaller drift compared to the one deposited at 200°C, thus indicating the good durability and reliability of pH sensor for 12 hours.

3.2.3. ISFET and EGFET Comparison of Temperature Experimental Simulation

Temperature experimental simulation is the analysis of output voltage response of sensing membranes toward the changes of temperature in various buffer solutions of pH values between 4 and 12. As can be seen in Figures 9(a) and 9(b), the EGFET temperature experimental simulation is shown; meanwhile the ISFET temperature experimental simulation was shown in Figures 9(c) and 9(d). It is apparent from these figures that output voltage response for both sensing membranes toward the changes of solution temperature for EGFET is more stable compared to ISFET. The output tolerance for both thin films deposited at room temperature and 200°C for EGFET and ISFET temperature experimental simulation were summarized in Table 3. For EGFET, both thin films gave lower output tolerance voltage per 1°C. In contrast with ISFET, both thin films showed higher output tolerance voltage per 1°C. From these results, it was proved that having an extended gate for the sensing membrane of sensor system can eliminate the temperature effect on the MOSFET because it was shown that the changes of pH buffer solution temperature did not give significant output voltage value toward the sensing membranes. Indirectly, this temperature compensation test illustrates the advantage of EGFET over ISFET. Hence, EGFET shows better temperature tolerance compared to ISFET which is good for sensor performance.

4. Conclusion

In this work, TiO2 thin films were prepared using RF magnetron sputtering method at different deposition temperatures onto ITO coated glass substrate for sensing membrane of the EGFET pH sensor. The surface morphology of TiO2 thin film deposited at room temperature showed a rough and nonuniform surface with mostly undistinguishable grain boundaries compared to that of 200°C which exhibits small grains and also shows the agglomeration of particles which has a more uniform surface structure. Typical transfer characteristics () for EGFET pH sensor were obtained for each deposited thin film at different temperatures. TiO2 thin film deposited at room temperature exhibited higher sensitivity and linearity of 59.89 mV/pH and 0.93503, respectively. In contrast, TiO2 thin film deposited at 200°C indicated slightly lower sensitivity and linearity of 37.60 mV/pH and 0.95132, correspondingly. This happened due to the higher surface roughness of TiO2 thin film deposited at room temperature compared to the one deposited at 200°C as observed in the surface morphology and AFM images. From Raman spectroscopy, both thin films are amorphous; thus it can be expected that the thin film deposited at room temperature has a lower degree of crystallinity compared to the sample deposited at 200°C which may facilitate more binding sites that relates to higher sensitivity. The hysteresis and drift of TiO2 thin film deposited at room temperature showed lower values compared to the one at 200°C. Low hysteresis and drift values indicate good characteristics of electrochemical sensor because high value of hysteresis and drift will limit the accuracy of the sensor. The temperature compensation test for both thin films presents likely the same output tolerance voltage; however, it gives a different output tolerance voltage for EGFET and ISFET. EGFET gave lower output tolerance voltage per 1°C compared to ISFET. It can be concluded that TiO2 thin films deposited at room temperature have good sensitivity, linearity, reliability, and durability which is suitable for pH sensing membrane.

Competing Interests

The authors declare that there are no competing interests regarding the publication of this paper.

Acknowledgments

The authors would like to thank all members of NANO-Electronic Center (NET), Universiti Teknologi MARA, UiTM, and NEMS and Photonics Cluster of MIMOS Berhad for all the research facilities. The work is partially supported by Ministry of Education Malaysia under the Niche Research Grant Scheme (Project Code 600-RMI/NRGS 5/3(6/2013)).