CO-Sensing Properties of Diode-Type Gas Sensors Employing Anodized Titania and Noble-Metal Electrodes under Hydrogen Atmosphere

CO-sensing properties of diode-type sensors employing an anodized TiO2 film and noble-metal (M) electrodes (M/TiO2 sensor, M: Pd, Pt, and Pd-nPt, n: the amount of Pt (wt %) in the Pd-nPt electrode) were investigated at 50–250 ◦C in dry or wet H2. All the M/TiO2 sensors showed nonlinear I–V characteristics as a diode device in air and N2, but the I–V characteristics of the sensors were actually linear in H2 because of the negligible small height of Schottky barrier at their M/TiO2 interface. The Pd/TiO2 sensor showed no CO response in H2, but the Pt/TiO2 and Pd-nPt/TiO2 sensors responded to CO in H2. Among them, the Pd-64Pt/TiO2 sensor showed the largest CO response at 100 ◦C in H2. The reason why the mixing of Pd with Pt was effective in improving the CO response is probably because of a decrease in the amount of dissolved hydrogen species, an increase in the amount of dissociatively adsorbed hydrogen species, and an increase in the amount of adsorbed CO species in CO balanced with H2 by the mixing of Pt into Pd. The interference from moisture in the target gas on the CO response should be largely improved from a practical application perspective.


Introduction
Proton exchange membrane fuel cells (PEMFCs) have great potential as main power-supplying devices for transportation systems such as fuel cell electric vehicles (PEFCs) as well as distributed stationary and portable applications, in the near future, because of zero emission, compact size, low temperature operation, and other advantages.The fuel of the PEMFCs, hydrogen (H 2 ), is mainly produced by steam reforming of common natural gas containing methane in our modern society, and thus the given amount of carbon monoxide (CO) is mixed in the produced H 2 -based gas.However, even a small number of CO molecules strongly reduce the electrocatalytic activity of platinum (Pt)-based nanoparticles loaded on the carbon substrate of gas-diffusion electrodes for PEMFCs, which is well known as the "CO-poisoning effect" [1].Therefore, highly sensitive CO sensors operable under H 2 -based atmosphere are very convenient for monitoring the concentration of CO in the steam-reforming and various other processes such as a water gas shift reaction.
Various types of gas sensors, such as chemiresistor-type sensors using oxides [2][3][4][5][6], polymers [7], or metal salts [8][9][10], electrochemical sensors [11][12][13], and solid-electrolyte sensors [14,15], have been developed to detect CO sensitively and selectively, under reducing atmosphere.However, none of the sensors have CO-sensing properties sufficient to quantify the concentration of residue CO left in the reformed gas.On the other hand, we have already demonstrated that the diode-type gas sensors employing a titania (TiO 2 ) [16][17][18][19][20][21] or niobia (Nb 2 O 5 ) [22,23] film, which was prepared by the anodization of a constituent metal plate, and noble-metal electrodes (mainly, palladium (Pd) and/or Pt) showed quite excellent H 2 -sensing properties, especially under inert atmosphere (N 2 ), because H 2 was dissociatively adsorbed and dissolved into the noble-metal electrode and thus the drastic reduction of the work function as well as the height of the Schottky barrier of noble-metal/oxide interface.The excellent H 2 -sensing properties of the diode-type gas sensors under inert atmosphere motivated us to investigate their sensing properties to other gases under a more specific environment.In this study, therefore, CO-sensing properties of the diode-type sensors employing an anodized TiO 2 film and Pd and/or Pt electrodes have been investigated in comparison with their H 2 -sensing properties in air as well as in N 2 .

Fabrication of Diode-Type Gas Sensors
Figure 1 shows schematic drawing of a diode-type gas sensor employing an anodized TiO 2 film on a Ti plate and Pd and/or Pt electrodes, which was fabricated as follows.A half part of a Ti plate (5.0 × 10.0 × 0.5 mm 3 ) was anodized in 0.5 M H 2 SO 4 aqueous solution at 20 • C for 30 min at a current density of 50 mA•cm −2 , after the Ti plate was polished by using a buffing machine (Marumoto Struers K. K., Osaka, Japan, S5629) employing 3 kinds of diamond aqueous suspensions (diameter of diamond powders suspended: 9 µm, 3 µm, and 0.5 µm) sequentially.A pair of noble-metal (Pd or Pt) electrodes was fabricated on the surface of both the TiO 2 thin film and the Ti plate by radio-frequency (rf) magnetron sputtering (Shimadzu, HSR-552S).The mixing of Pd with Pt was also conducted by simultaneous deposition utilizing both Pd and Pt targets, and the obtained electrodes were denoted as Pd-nPt, where n is the amount of Pt (wt %) in the Pd-nPt electrodes.The fabrication conditions of all the electrodes were shown in Table 1, together with their composition and thickness, and the deposition time for all the films was 7 min.The composition of the Pd-nPt electrodes was measured by energy dispersive X-ray spectroscopy (EDS; JEOL Ltd., Tokyo, Japan, JED-2300) equipped with scanning electron microscopy (SEM; JEOL Ltd., JSM-7500F).The obtained sensors with noble-metal (M) electrodes were denoted as M/TiO 2 (M: Pd, Pt, or Pd-nPt).Each electrode was connected with an Au lead wire by using a Pt paste, and the electrical contact was ensured by firing at 600 • C for 1 h in air.
Chemosensors 2018, 6, x 2 of 12 anodization of a constituent metal plate, and noble-metal electrodes (mainly, palladium (Pd) and/or Pt) showed quite excellent H2-sensing properties, especially under inert atmosphere (N2), because H2 was dissociatively adsorbed and dissolved into the noble-metal electrode and thus the drastic reduction of the work function as well as the height of the Schottky barrier of noble-metal/oxide interface.The excellent H2-sensing properties of the diode-type gas sensors under inert atmosphere motivated us to investigate their sensing properties to other gases under a more specific environment.In this study, therefore, CO-sensing properties of the diode-type sensors employing an anodized TiO2 film and Pd and/or Pt electrodes have been investigated in comparison with their H2-sensing properties in air as well as in N2.

Fabrication of Diode-Type Gas Sensors
Figure 1 shows schematic drawing of a diode-type gas sensor employing an anodized TiO2 film on a Ti plate and Pd and/or Pt electrodes, which was fabricated as follows.A half part of a Ti plate (5.0 × 10.0 × 0.5 mm 3 ) was anodized in 0.5 M H2SO4 aqueous solution at 20 °C for 30 min at a current density of 50 mA•cm −2 , after the Ti plate was polished by using a buffing machine (Marumoto Struers K. K., Osaka, Japan, S5629) employing 3 kinds of diamond aqueous suspensions (diameter of diamond powders suspended: 9 μm, 3 μm, and 0.5 μm) sequentially.A pair of noble-metal (Pd or Pt) electrodes was fabricated on the surface of both the TiO2 thin film and the Ti plate by radio-frequency (rf) magnetron sputtering (Shimadzu, HSR-552S).The mixing of Pd with Pt was also conducted by simultaneous deposition utilizing both Pd and Pt targets, and the obtained electrodes were denoted as Pd-nPt, where n is the amount of Pt (wt %) in the Pd-nPt electrodes.The fabrication conditions of all the electrodes were shown in Table 1, together with their composition and thickness, and the deposition time for all the films was 7 min.The composition of the Pd-nPt electrodes was measured by energy dispersive X-ray spectroscopy (EDS; JEOL Ltd., Tokyo, Japan, JED-2300) equipped with scanning electron microscopy (SEM; JEOL Ltd., JSM-7500F).The obtained sensors with noble-metal (M) electrodes were denoted as M/TiO2 (M: Pd, Pt, or Pd-nPt).Each electrode was connected with an Au lead wire by using a Pt paste, and the electrical contact was ensured by firing at 600 °C for 1 h in air.

Measurements of Gas-Sensing Properties
In order to confirm fundamental properties as their diode-type sensing devices, the current (I)voltage (V) characteristics of the representative 3 sensors (Pd/TiO2, Pt/TiO2, and Pd-64Pt/TiO2 sensors) were investigated in a base gas (dry air or N2) and in 8000 ppm H2 balanced with a base gas, and their sensing properties to 8000 ppm H2 balanced with dry air or N2 were measured at 250 °C when a dc voltage of +100 mV was applied to the sensors under forward bias condition (M(+)-TiO2-Ti(−)).In addition, a dc voltage of +1.0 mV was applied to all the sensors under the forward bias condition and the sensing properties to 1-80 ppm CO balanced with H2 under dry or wet (absolute humidity (AH): ca.12.8 g•m −3 ) atmospheres were measured at 50-250 °C (mainly, at 100 °C) after annealing under H2

Measurements of Gas-Sensing Properties
In order to confirm fundamental properties as their diode-type sensing devices, the current (I)-voltage (V) characteristics of the representative 3 sensors (Pd/TiO 2 , Pt/TiO 2 , and Pd-64Pt/TiO 2 sensors) were investigated in a base gas (dry air or N 2 ) and in 8000 ppm H 2 balanced with a base gas, and their sensing properties to 8000 ppm H 2 balanced with dry air or N 2 were measured at 250 • C when a dc voltage of +100 mV was applied to the sensors under forward bias condition (M(+)-TiO 2 -Ti(−)).In addition, a dc voltage of +1.0 mV was applied to all the sensors under the forward bias condition and the sensing properties to 1-80 ppm CO balanced with H 2 under dry or wet (absolute humidity (AH): ca.12.8 g•m −3 ) atmospheres were measured at 50-250 • C (mainly, at 100 • C) after annealing under H 2 atmosphere at 400 • C for 1 h.The I-V characteristics of the Pd-64Pt/TiO 2 sensor as a representative were also evaluated in an applied voltage range of −1.0 to +1.0 V in dry N 2 , dry H 2 or 80 ppm CO balanced with dry H 2 .All the sensors were externally heated in a test chamber by utilizing an electric furnace, and all the test gases were continuously supplied into the test chamber at a flow rate of 100 cm 3 •min −1 .
The magnitude of response was defined as the absolute value of "I s − I b ", where I s and I b represented sensor-current values in a sample gas (at 10 min after switching to the sample gas) and in a base gas, respectively.90% response time (T RS ) was defined as a period necessary to reach 90% sensor-current value of "I s − I b " from the time switching to the sample gas, while 90% recovery time (T RC ) was defined as that necessary to reach 90% sensor-current value of "I b -I s " from the time switching to the base gas.The 90% response and recovery times numerically contain a delay period from the gas-switching time to the response-and recovery-starting times, ca.64 s, in this study, since the dead volume of the gas-flow pathway and the chamber in the measurement apparatus is ca.106 cm 3 .

Microstructure
Figure 2 shows SEM photographs of surfaces of a Ti plate, a bare TiO 2 film fabricated on a Ti plate by anodization, and a TiO 2 film coated with a Pd-64Pt, as a representative of M/TiO 2 , by rf magnetron sputtering.The surface of a Ti plate polished with diamond suspensions became smoother than that of an untreated Ti plate, but many scratch lines with submicron intervals were observed in the same direction on the surface, since it was finally polished by using a buffing machine employing an aqueous suspension containing diamond powders with a diameter of ca.0.5 µm.The anodization of the Ti plate in 0.5 M H 2 SO 4 aqueous solution produced a relatively planar TiO 2 film with submicron pores, on the surface, without harmful effects of the submicron scratches.The prepared TiO 2 film (thickness: around 1 µm) consisted of dense columnar TiO 2 polycrystallites (main crystal phase: anatase and rutile for the TiO 2 film before and after heat treatment at 600 • C, respectively) and the microstructure remained unchanged even after the heat treatment, as shown in our previous studies [16].The Pd-64Pt film was uniformly deposited on the surface of the anodized TiO 2 film, and then the submicron pores of the anodized TiO 2 film were partly filled with the Pd-64Pt agglomerates (estimated size: several tens of nm in diameter) and the size of the pores considerably reduced after the Pd-64Pt deposition.The X-ray photoelectron spectroscopy (XPS) analysis of the Pd-64Pt electrode, which was fabricated with the same procedure, has demonstrated the effect of heat treatment on the composition, as below [19].
The bulk composition of the Pd-64Pt electrode right after the sputtering deposition (Pd: 36-33 wt %, Pt: 64-67 wt %) was very comparable to that which was estimated from the sputtering rate of Pd and Pt (Pd: 36 wt %, Pt: 64 wt %), while the weight percentage of Pt on the electrode surface right after the sputtering deposition (ca.55 wt %) was smaller than that which was estimated from the sputtering rate of Pd and Pt (namely n: 64 wt %).After firing at 600 • C for 1 h in air, Pd and Pt on the surface of the Pd-64Pt electrode was oxidized to PdO and PtO, respectively and the amount of the Pd component on the electrode surface increased by ca.90 wt %, because the given amount of Pd diffused toward the electrode surface due to the higher affinity of Pd for oxygen than Pt.The annealing at elevated temperatures under H 2 atmosphere reduced the PdO and PtO on the electrode to Pd and Pt, respectively. of Pt on the electrode surface right after the sputtering deposition (ca.55 wt %) was smaller than that which was estimated from the sputtering rate of Pd and Pt (namely n: 64 wt %).After firing at 600 °C for 1 h in air, Pd and Pt on the surface of the Pd-64Pt electrode was oxidized to PdO and PtO, respectively and the amount of the Pd component on the electrode surface increased by ca.90 wt %, because the given amount of Pd diffused toward the electrode surface due to the higher affinity of Pd for oxygen than Pt.The annealing at elevated temperatures under H2 atmosphere reduced the PdO and PtO on the electrode to Pd and Pt, respectively.

Basic Diode Characteristics and H2-Sensing Properties in Dry Air and N2
In order to confirm the diode-type behavior of the M/TiO2 sensors fabricated in this study, I-V characteristics of typical three sensors, Pd/TiO2, Pt/TiO2, and Pd-64Pt/TiO2 sensors, were investigated at 250 °C in base gases (dry air and N2) and in 8000 ppm H2 balanced with the base gases, as shown in Figure 3.These sensors obviously showed nonlinear I-V characteristics under dry H2-free atmospheres, indicating that the Schottky contact was formed at all the M/TiO2 interfaces.In addition, the magnitude of current of all the M/TiO2 sensors in dry N2 under the forward bias was larger than that in dry air, and the magnitude of current of these sensors tended to increase with an increase in the amount of Pt in the M electrodes, especially in dry N2.Generally, the work function of Pt (5.26-5.69eV) is larger than that of Pd (4.87-5.25 eV) [24], and thus the Schottky barrier of the Pt/TiO2 interface (1.7-1.8 eV) tends to be larger than that of the Pd/TiO2 interface (1.2-1.3 eV) [25].On the other hand, the surface of the Pd electrode is easily oxidized to PdO under heat treatment at 600 °C in air, and the PdO is partially reduced in N2 at elevated temperatures [19,21].In contrast, the surface of the Pt electrode is not easily oxidized even under heat treatment at 600 °C in air [21].These results which were obtained by utilizing the XPS analysis have already been demonstrated in our previous papers [19,21].The electron affinity of PdO (ca.5.5 eV) is larger than the work function of Pd [26], and various defect and impurity levels are easily produced at the M/TiO2 interface [27,28].On the other hand, Pd and Pt in the bulk of the Pd-64Pt electrode was alloyed and the alloy Pd-Pt phase in the bulk is likely to be chemically and thermally stable as a metal at elevated temperatures, even in air [19,21].After the heat treatment at 600 °C in air, however, most of Pd in the vicinity of the surface of the Pd-64Pt film were oxidized to PdO, while Pt just on the surface was only oxidized to PtO [19].The degree of oxidation on the electrode surface probably had an impact on the electric properties of the M/TiO2 interface and thus these I-V characteristics dependent on the gaseous atmosphere.
The addition of H2 into both the base gases enhanced the magnitude of currents of these sensors.The behavior is due to dissociative adsorption of H2 molecules on the surface of M electrodes, subsequent dissolution of the hydrogen species into the M electrodes, and in turn reduction in these

Basic Diode Characteristics and H 2 -Sensing Properties in Dry Air and N 2
In order to confirm the diode-type behavior of the M/TiO 2 sensors fabricated in this study, I-V characteristics of typical three sensors, Pd/TiO 2 , Pt/TiO 2 , and Pd-64Pt/TiO 2 sensors, were investigated at 250 • C in base gases (dry air and N 2 ) and in 8000 ppm H 2 balanced with the base gases, as shown in Figure 3.These sensors obviously showed nonlinear I-V characteristics under dry H 2 -free atmospheres, indicating that the Schottky contact was formed at all the M/TiO 2 interfaces.In addition, the magnitude of current of all the M/TiO 2 sensors in dry N 2 under the forward bias was larger than that in dry air, and the magnitude of current of these sensors tended to increase with an increase in the amount of Pt in the M electrodes, especially in dry N 2 .Generally, the work function of Pt (5.26-5.69eV) is larger than that of Pd (4.87-5.25 eV) [24], and thus the Schottky barrier of the Pt/TiO 2 interface (1.7-1.8 eV) tends to be larger than that of the Pd/TiO 2 interface (1.2-1.3 eV) [25].On the other hand, the surface of the Pd electrode is easily oxidized to PdO under heat treatment at 600 • C in air, and the PdO is partially reduced in N 2 at elevated temperatures [19,21].In contrast, the surface of the Pt electrode is not easily oxidized even under heat treatment at 600 • C in air [21].These results which were obtained by utilizing the XPS analysis have already been demonstrated in our previous papers [19,21].The electron affinity of PdO (ca.5.5 eV) is larger than the work function of Pd [26], and various defect and impurity levels are easily produced at the M/TiO 2 interface [27,28].On the other hand, Pd and Pt in the bulk of the Pd-64Pt electrode was alloyed and the alloy Pd-Pt phase in the bulk is likely to be chemically and thermally stable as a metal at elevated temperatures, even in air [19,21].After the heat treatment at 600 • C in air, however, most of Pd in the vicinity of the surface of the Pd-64Pt film were oxidized to PdO, while Pt just on the surface was only oxidized to PtO [19].The degree of oxidation on the electrode surface probably had an impact on the electric properties of the M/TiO 2 interface and thus these I-V characteristics dependent on the gaseous atmosphere.forward bias.Namely, the mixing of oxygen into the H2-containing gaseous atmosphere largely reduced the magnitude of current of all the sensors.This indicates that the certain percentage of H2 was oxidized with oxygen species (e.g., oxygen adsorbates) on the surface of M electrodes and thus the effective H2 concentration on the surface reduced in dry air, whereas the H2 concentration on the M electrodes hardly decrease in dry N2 [21].Furthermore, the Pd/TiO2 and Pd-64Pt/TiO2 sensors showed ohmic-like I-V characteristics in H2 balanced with dry N2, due to the quite small Schottky barrier at the M/TiO2 interface.On the other hand, the Pt/TiO2 sensor managed to maintain the nonlinear I-V characteristics even in dry N2, which showed the smaller magnitude of currents in both dry N2 and air than the Pd/TiO2 and Pd-64Pt/TiO2 sensors.Figure 4 shows response transients of Pd/TiO2, Pt/TiO2, and Pd-64Pt/TiO2 sensors to 8000 ppm H2 at 250 °C at a forward bias of +100 mV, in dry air and N2.As expected from these I-V characteristics (Figure 3), the magnitude of currents of all the sensors was quite low in both the base gases, and the addition of H2 into the base gases drastically enhanced the magnitude of currents especially in dry N2.Thus, these sensors showed very large H2 responses, but these H2-sensing properties are quite dependent on oxygen concentration.Among these sensors, the Pd-64Pt/TiO2 sensors showed relatively large H2 responses in both dry air and N2.The addition of H 2 into both the base gases enhanced the magnitude of currents of these sensors.The behavior is due to dissociative adsorption of H 2 molecules on the surface of M electrodes, subsequent dissolution of the hydrogen species into the M electrodes, and in turn reduction in these work functions, leading to a decrease in the height of the Schottky barrier at the M/TiO 2 interface.In addition, the magnitude of currents in 8000 ppm H 2 balanced with dry N 2 was more than two orders of magnitude larger than those in 8000 ppm H 2 balanced with dry air, under the same applied forward bias.Namely, the mixing of oxygen into the H 2 -containing gaseous atmosphere largely reduced the magnitude of current of all the sensors.This indicates that the certain percentage of H 2 was oxidized with oxygen species (e.g., oxygen adsorbates) on the surface of M electrodes and thus the effective H 2 concentration on the surface reduced in dry air, whereas the H 2 concentration on the M electrodes hardly decrease in dry N 2 [21].Furthermore, the Pd/TiO 2 and Pd-64Pt/TiO 2 sensors showed ohmic-like I-V characteristics in H 2 balanced with dry N 2 , due to the quite small Schottky barrier at the M/TiO 2 interface.On the other hand, the Pt/TiO 2 sensor managed to maintain the nonlinear I-V characteristics even in dry N 2 , which showed the smaller magnitude of currents in both dry N 2 and air than the Pd/TiO 2 and Pd-64Pt/TiO 2 sensors.
Figure 4 shows response transients of Pd/TiO 2 , Pt/TiO 2 , and Pd-64Pt/TiO 2 sensors to 8000 ppm H 2 at 250 • C at a forward bias of +100 mV, in dry air and N 2 .As expected from these I-V characteristics (Figure 3), the magnitude of currents of all the sensors was quite low in both the base gases, and the addition of H 2 into the base gases drastically enhanced the magnitude of currents especially in dry N 2 .Thus, these sensors showed very large H 2 responses, but these H 2 -sensing properties are quite dependent on oxygen concentration.Among these sensors, the Pd-64Pt/TiO 2 sensors showed relatively large H 2 responses in both dry air and N 2 .

Typical I-V Characteristics and CO-Sensing Properties in H2
Figure 5 shows I-V characteristics of the Pd-64Pt/TiO2 sensor as a representative, in dry N2 at 100 °C and 250 °C, and in dry H2 and 80 ppm CO balanced with dry H2 at 100 °C.This sensor was annealed at 400 °C in H2 for 1 h, before the measurement, because of the enhancement in the thermal and chemical stability.The annealing at 400 °C in H2 increased the magnitude of current of the Pd-64Pt/TiO2 sensor in dry N2, and the current of the annealed Pd-64Pt/TiO2 sensor was a little larger than that of the non-annealed Pd-64Pt/TiO2 sensor at 250 °C (see Figure 3a(ii)), in dry N2.In addition, a decrease in the operating temperature reduced the magnitude of current, the nonlinearity of these I-V characteristics of the annealed Pd-64Pt/TiO2 sensor was comparable to that of the non-annealed Pd-64Pt/TiO2 sensor at 250 °C in dry N2.On the other hand, the I-V relationship in dry H2 was considerably linear, due to the negligibly small Schottky barrier, and the I-V relationship of the Pd/TiO2 and Pt/TiO2 sensors at 100 °C in dry H2 was also quite similar to that of the Pd-64Pt/TiO2 sensor (not shown here).The addition of 80 ppm CO into dry H2 reduced the magnitude of current of the Pd-64Pt/TiO2 sensor at 100 °C, which means that the sensor is capable of detecting CO in dry H2 at least.3a(ii)), in dry N 2 .In addition, a decrease in the operating temperature reduced the magnitude of current, the nonlinearity of these I-V characteristics of the annealed Pd-64Pt/TiO 2 sensor was comparable to that of the non-annealed Pd-64Pt/TiO 2 sensor at 250 • C in dry N 2 .On the other hand, the I-V relationship in dry H 2 was considerably linear, due to the negligibly small Schottky barrier, and the I-V relationship of the Pd/TiO 2 and Pt/TiO 2 sensors at 100 • C in dry H 2 was also quite similar to that of the Pd-64Pt/TiO 2 sensor (not shown here).The addition of 80 ppm CO into dry H 2 reduced the magnitude of current of the Pd-64Pt/TiO 2 sensor at 100 • C, which means that the sensor is capable of detecting CO in dry H 2 at least.

Typical I-V Characteristics and CO-Sensing Properties in H2
Figure 5 shows I-V characteristics of the Pd-64Pt/TiO2 sensor as a representative, in dry N2 at 100 °C and 250 °C, and in dry H2 and 80 ppm CO balanced with dry H2 at 100 °C.This sensor was annealed at 400 °C in H2 for 1 h, before the measurement, because of the enhancement in the thermal and chemical stability.The annealing at 400 °C in H2 increased the magnitude of current of the Pd-64Pt/TiO2 sensor in dry N2, and the current of the annealed Pd-64Pt/TiO2 sensor was a little larger than that of the non-annealed Pd-64Pt/TiO2 sensor at 250 °C (see Figure 3a(ii)), in dry N2.In addition, a decrease in the operating temperature reduced the magnitude of current, the nonlinearity of these I-V characteristics of the annealed Pd-64Pt/TiO2 sensor was comparable to that of the non-annealed Pd-64Pt/TiO2 sensor at 250 °C in dry N2.On the other hand, the I-V relationship in dry H2 was considerably linear, due to the negligibly small Schottky barrier, and the I-V relationship of the Pd/TiO2 and Pt/TiO2 sensors at 100 °C in dry H2 was also quite similar to that of the Pd-64Pt/TiO2 sensor (not shown here).The addition of 80 ppm CO into dry H2 reduced the magnitude of current of the Pd-64Pt/TiO2 sensor at 100 °C, which means that the sensor is capable of detecting CO in dry H2 at least.  Figure 6 shows response transients of three kinds of the annealed M/TiO 2 sensors (M: Pd, Pt, and Pd-64Pt) to 80 ppm CO at 100 • C in dry H 2 .The surface of the M electrodes is totally reduced with H 2 in both pre-treatment and operating conditions, and thus metallic Pd and/or Pt on the electrode surface were directly exposed to target gas [19].In addition, H 2 molecules as a base-gas component are dissociatively adsorbed on the surface and the part of the adsorbed hydrogen species is dissolved into the metals, which largely decrease the work function of the metals and thus change the M/TiO 2 interface entirely from the Schottky contact to quasi-ohmic contact.As they resulted in a large reduction in the sensor resistances, the quite small forward voltage, +1.0 mV, was applied to the sensors for the CO sensing at 100 • C in dry H 2 .The dependence of the I-V characteristics of the Pd-64Pt/TiO 2 sensor under the reducing atmospheres (Figure 5) promised that the addition of CO into dry H 2 increases the sensor resistance.Actually, the Pt/TiO 2 sensor expectedly showed a clear CO response in the negative direction, while the Pd/TiO 2 sensor showed no CO response.The solubility of hydrogen into Pt is extremely smaller than that into Pd [29], but H 2 molecules easily and largely adsorb on the Pt surface [30].In addition, CO molecules are well known to strongly adsorb on the Pt surface [2][3][4][5][6][7][8][9][10][11][12][13][14][15], especially at around 100 • C (temperatures at which PEMFCs generally operate) under H 2 -based reducing atmospheres.Therefore, the strongly adsorbed CO species probably interrupted the dissociatively adsorption of H 2 molecules, to increase the work function of Pt and the height of Schottky barrier at the M/TiO 2 interface, and thus to decrease the magnitude of current.In contrast, the solubility of hydrogen into Pd is quite large [29], and thus the amount of CO adsorbed on the Pd surface is generally much smaller than the total amount of hydrogen adsorbed on the Pd surface and dissolved into the Pd bulk.Therefore, the adsorption of CO molecules probably had negligible influence on the adsorption and absorption behavior of H 2 molecules.Furthermore, the mixing of Pt with a Pd electrode was quite effective in improving the magnitude of CO response.This is probably because a decrease in the amount of Pd (namely, an increase in the amount of Pt) in the M electrode decreased the amount of dissolved hydrogen species in H 2 [29] and increased the amount of dissociatively adsorbed hydrogen species [30], and then the adsorbed CO on the electrode surface interrupted the dissociatively adsorption of hydrogen, in 80 ppm CO balanced with H 2 .Further discussion will be done based on the additional data in the following section.On the other hand, the response and recovery speeds of the Pt/TiO 2 and Pd-64Pt/TiO 2 sensors were much slower than those for the H 2 response in both dry air and N 2 , and thus their drastic improvement is indispensable from a practical application perspective.
Figure 6 shows response transients of three kinds of the annealed M/TiO2 sensors (M: Pd, Pt, and Pd-64Pt) to 80 ppm CO at 100 °C in dry H2.The surface of the M electrodes is totally reduced with H2 in both pre-treatment and operating conditions, and thus metallic Pd and/or Pt on the electrode surface were directly exposed to target gas [19].In addition, H2 molecules as a base-gas component are dissociatively adsorbed on the surface and the part of the adsorbed hydrogen species is dissolved into the metals, which largely decrease the work function of the metals and thus change the M/TiO2 interface entirely from the Schottky contact to quasi-ohmic contact.As they resulted in a large reduction in the sensor resistances, the quite small forward voltage, +1.0 mV, was applied to the sensors for the CO sensing at 100 °C in dry H2.The dependence of the I-V characteristics of the Pd-64Pt/TiO2 sensor under the reducing atmospheres (Figure 5) promised that the addition of CO into dry H2 increases the sensor resistance.Actually, the Pt/TiO2 sensor expectedly showed a clear CO response in the negative direction, while the Pd/TiO2 sensor showed no CO response.The solubility of hydrogen into Pt is extremely smaller than that into Pd [29], but H2 molecules easily and largely adsorb on the Pt surface [30].In addition, CO molecules are well known to strongly adsorb on the Pt surface [2][3][4][5][6][7][8][9][10][11][12][13][14][15], especially at around 100 °C (temperatures at which PEMFCs generally operate) under H2-based reducing atmospheres.Therefore, the strongly adsorbed CO species probably interrupted the dissociatively adsorption of H2 molecules, to increase the work function of Pt and the height of Schottky barrier at the M/TiO2 interface, and thus to decrease the magnitude of current.In contrast, the solubility of hydrogen into Pd is quite large [29], and thus the amount of CO adsorbed on the Pd surface is generally much smaller than the total amount of hydrogen adsorbed on the Pd surface and dissolved into the Pd bulk.Therefore, the adsorption of CO molecules probably had negligible influence on the adsorption and absorption behavior of H2 molecules.Furthermore, the mixing of Pt with a Pd electrode was quite effective in improving the magnitude of CO response.This is probably because a decrease in the amount of Pd (namely, an increase in the amount of Pt) in the M electrode decreased the amount of dissolved hydrogen species in H2 [29] and increased the amount of dissociatively adsorbed hydrogen species [30], and then the adsorbed CO on the electrode surface interrupted the dissociatively adsorption of hydrogen, in 80 ppm CO balanced with H2.Further discussion will be done based on the additional data in the following section.On the other hand, the response and recovery speeds of the Pt/TiO2 and Pd-64Pt/TiO2 sensors were much slower than those for the H2 response in both dry air and N2, and thus their drastic improvement is indispensable from a practical application perspective.

Impacts of Various Factors on CO-Sensing Properties in H2
Figure 7 shows operating temperature dependences of responses of the annealed Pt/TiO2 and Pd-64Pt/TiO2 sensors to 80 ppm CO in dry H2.The magnitude of CO response of the Pt/TiO2 sensor

Impacts of Various Factors on CO-Sensing Properties in H 2
Figure 7 shows operating temperature dependences of responses of the annealed Pt/TiO 2 and Pd-64Pt/TiO 2 sensors to 80 ppm CO in dry H 2 .The magnitude of CO response of the Pt/TiO 2 sensor monotonically decreased with an increase in operating temperature.The solubility of hydrogen into Pt was generally much smaller than that into Pd, and gradually increased with an increase in temperature [29].In addition, the CO coverage on Pt also decreased monotonically with an increase in temperature [31].These factors probably determined the relatively large CO response of the Pt/TiO 2 sensor at lower operating temperatures.On the other hand, the CO response of the Pd-64Pt/TiO 2 sensor was smaller than that of the Pt/TiO 2 sensor at 50 • C, probably due to the large amounts of dissociatively adsorbed and dissolved hydrogen species and relatively small amount of adsorbed CO on Pd-64Pt in comparison with those of Pt.However, the magnitude of the CO response increased with an increase in operating temperature, and the Pd-64Pt/TiO 2 sensor showed quite large CO responses at 100-150 • C. The amount of hydrogen dissolved into Pd abruptly decreases with an increase in the temperature, especially in the temperature range of 100-150 • C [29], whereas the CO coverage on Pt also decreased monotonically with an increase in the temperature [31].These two factors may increase the ratio of the amount of adsorbed CO species to the amounts of dissociatively adsorbed and dissolved hydrogen species at around 100-150 • C. In addition, the magnitude of CO responses at 200-250 • C were smaller than those at 50 • C, probably because the amount of CO adsorbed on the electrode surface decreased with an increase in the operating temperature.However, even the CO response of the Pd-64Pt/TiO 2 sensor at 200-250 • C was still a little larger than that of the Pt/TiO 2 sensor.Meanwhile, this gaseous atmosphere, namely dry H 2 containing CO, have a high possibility of progressing the hydrogenation of CO on the electrode surface, to produce some kinds of hydrocarbons and alcohols (Fischer-Tropsch process; e.g., nCO [32] and thus decrease the effective concentration of CO.The effect of these catalytic reactions on the electrode surface on the CO-sensing mechanism should be clarified in the future, in order to enhance the CO-sensing properties in dry H 2 . Chemosensors 2018, 6, x 8 of 12 monotonically decreased with an increase in operating temperature.The solubility of hydrogen into Pt was generally much smaller than that into Pd, and gradually increased with an increase in temperature [29].In addition, the CO coverage on Pt also decreased monotonically with an increase in temperature [31].These factors probably determined the relatively large CO response of the Pt/TiO2 sensor at lower operating temperatures.On the other hand, the CO response of the Pd-64Pt/TiO2 sensor was smaller than that of the Pt/TiO2 sensor at 50 °C, probably due to the large amounts of dissociatively adsorbed and dissolved hydrogen species and relatively small amount of adsorbed CO on Pd-64Pt in comparison with those of Pt.However, the magnitude of the CO response increased with an increase in operating temperature, and the Pd-64Pt/TiO2 sensor showed quite large CO responses at 100-150 °C.The amount of hydrogen dissolved into Pd abruptly decreases with an increase in the temperature, especially in the temperature range of 100-150 °C [29], whereas the CO coverage on Pt also decreased monotonically with an increase in the temperature [31].These two factors may increase the ratio of the amount of adsorbed CO species to the amounts of dissociatively adsorbed and dissolved hydrogen species at around 100-150 °C.In addition, the magnitude of CO responses at 200-250 °C were smaller than those at 50 °C, probably because the amount of CO adsorbed on the electrode surface decreased with an increase in the operating temperature.However, even the CO response of the Pd-64Pt/TiO2 sensor at 200-250 °C was still a little larger than that of the Pt/TiO2 sensor.Meanwhile, this gaseous atmosphere, namely dry H2 containing CO, have a high possibility of progressing the hydrogenation of CO on the electrode surface, to produce some kinds of hydrocarbons and alcohols (Fischer-Tropsch process; e.g., nCO + (2n + 1)H2 → CnH2n+2 + nH2O) [32] and thus decrease the effective concentration of CO.The effect of these catalytic reactions on the electrode surface on the CO-sensing mechanism should be clarified in the future, in order to enhance the CO-sensing properties in dry H2. Figure 8 shows variations in responses to 80 ppm CO and response and recovery times of all annealed Pd-nPt/TiO2 sensors at 100 °C in dry H2, as a function of the amount of Pt in the Pd-nPt electrodes.The mixing of Pt into Pd enhanced the magnitude of CO response, and vice versa.This fact indicates that the balance between the adsorption property of CO and the adsorption and dissolution properties of H2 on the electrode surface is the most important in enhancing the COsensing properties of the Pd-nPt sensors in H2.Consequently, the Pd-64Pt/TiO2 sensor showed the largest CO response among all the Pd-nPt/TiO2 sensors.On the other hand, the composition of the electrode had a minimal effect on the response and recovery speeds of the Pd-nPt/TiO2 sensors, too.Namely, the Pd-64Pt/TiO2 sensor showed the fastest response speed and the slowest recovery speed among them, probably because the response and recovery speeds were simply dependent on only the magnitude of CO responses.Figure 8 shows variations in responses to 80 ppm CO and response and recovery times of all annealed Pd-nPt/TiO 2 sensors at 100 • C in dry H 2 , as a function of the amount of Pt in the Pd-nPt electrodes.The mixing of Pt into Pd enhanced the magnitude of CO response, and vice versa.This fact indicates that the balance between the adsorption property of CO and the adsorption and dissolution properties of H 2 on the electrode surface is the most important in enhancing the CO-sensing properties of the Pd-nPt sensors in H 2 .Consequently, the Pd-64Pt/TiO 2 sensor showed the largest CO response among all the Pd-nPt/TiO 2 sensors.On the other hand, the composition of the electrode had a minimal effect on the response and recovery speeds of the Pd-nPt/TiO 2 sensors, too.Namely, the Pd-64Pt/TiO 2 sensor showed the fastest response speed and the slowest recovery speed among them, probably because the response and recovery speeds were simply dependent on only the magnitude of CO responses.
Chemosensors 2018, 6, x 9 of 12 Figure 9 shows concentration dependences of CO responses of the annealed Pt/TiO2 and Pd-64Pt/TiO2 sensors to 80 ppm CO at 100 °C in dry H2.The Pd-64Pt/TiO2 sensor showed a large response to a high concentration of CO (ca.403 μA for 80 ppm CO) with the excellent signal/noise (S/N) ratio (ca.143 for 80 ppm CO), but the response of the Pt/TiO2 sensor to 80 ppm CO and the S/N ratio was relatively small (ca.20 μA and ca. 10, respectively).In addition, the Pd-64Pt/TiO2 sensor easily detected even 1 ppm CO (the magnitude of response: ca.20 μA, S/N ratio: ca.10), while the Pt/TiO2 sensor showed no response to 1 ppm CO.As mentioned above, the CO response of the Pd-64Pt/TiO2 sensor was larger than that of the Pt/TiO2 sensor in all the concentration range.Unfortunately, both the sensors did not show a linear relationship between the CO response and the concentration.Figure 10 shows operating temperature dependences of responses of the annealed Pt/TiO2 and Pd-64Pt/TiO2 sensors to 80 ppm CO at 100 °C in wet H2 (AH: ca.12.8 g•m −3 ), and Figure 11 shows concentration dependences of CO responses of the annealed sensors at 100 °C in wet H2.Both the sensors showed the largest responses to 80 ppm CO at 100 °C in wet H2 (ca.18 μA (S/N ratio: ca.2.5) for the Pd-64Pt/TiO2 sensor and ca.6.0 μA (S/N ratio: ca.2.0) for the Pt/TiO2 sensor).In addition, the CO response of the Pd-64Pt/TiO2 sensor was much larger than that of the Pt/TiO2 sensor, and the magnitude of both the CO responses in wet H2 was much smaller than that in dry H2 in every operating temperature range.Nevertheless, the Pd-64Pt/TiO2 sensor managed to show a clear response even to 1 ppm CO (ca. 4 μA, S/N ratio: ca.1.5) in wet H2.The addition of moisture into the Figure 9 shows concentration dependences of CO responses of the annealed Pt/TiO 2 and Pd-64Pt/TiO 2 sensors to 80 ppm CO at 100 • C in dry H 2 .The Pd-64Pt/TiO 2 sensor showed a large response to a high concentration of CO (ca.403 µA for 80 ppm CO) with the excellent signal/noise (S/N) ratio (ca.143 for 80 ppm CO), but the response of the Pt/TiO 2 sensor to 80 ppm CO and the S/N ratio was relatively small (ca.20 µA and ca. 10, respectively).In addition, the Pd-64Pt/TiO 2 sensor easily detected even 1 ppm CO (the magnitude of response: ca.20 µA, S/N ratio: ca.10), while the Pt/TiO 2 sensor showed no response to 1 ppm CO.As mentioned above, the CO response of the Pd-64Pt/TiO 2 sensor was larger than that of the Pt/TiO 2 sensor in all the concentration range.Unfortunately, both the sensors did not show a linear relationship between the CO response and the concentration.Figure 9 shows concentration dependences of CO responses of the annealed Pt/TiO2 and Pd-64Pt/TiO2 sensors to 80 ppm CO at 100 °C in dry H2.The Pd-64Pt/TiO2 sensor showed a large response to a high concentration of CO (ca.403 μA for 80 ppm CO) with the excellent signal/noise (S/N) ratio (ca.143 for 80 ppm CO), but the response of the Pt/TiO2 sensor to 80 ppm CO and the S/N ratio was relatively small (ca.20 μA and ca. 10, respectively).In addition, the Pd-64Pt/TiO2 sensor easily detected even 1 ppm CO (the magnitude of response: ca.20 μA, S/N ratio: ca.10), while the Pt/TiO2 sensor showed no response to 1 ppm CO.As mentioned above, the CO response of the Pd-64Pt/TiO2 sensor was larger than that of the Pt/TiO2 sensor in all the concentration range.Unfortunately, both the sensors did not show a linear relationship between the CO response and the concentration.Figure 10 shows operating temperature dependences of responses of the annealed Pt/TiO2 and Pd-64Pt/TiO2 sensors to 80 ppm CO at 100 °C in wet H2 (AH: ca.12.8 g•m −3 ), and Figure 11 shows concentration dependences of CO responses of the annealed sensors at 100 °C in wet H2.Both the sensors showed the largest responses to 80 ppm CO at 100 °C in wet H2 (ca.18 μA (S/N ratio: ca.2.5) for the Pd-64Pt/TiO2 sensor and ca.6.0 μA (S/N ratio: ca.2.0) for the Pt/TiO2 sensor).In addition, the Figure 10 shows operating temperature dependences of responses of the annealed Pt/TiO 2 and Pd-64Pt/TiO 2 sensors to 80 ppm CO at 100 • C in wet H 2 (AH: ca.12.8 g•m −3 ), and Figure 11 shows concentration dependences of CO responses of the annealed sensors at 100 • C in wet H 2 .Both the sensors showed the largest responses to 80 ppm CO at 100 • C in wet H 2 (ca.18 µA (S/N ratio: ca.2.5) for the Pd-64Pt/TiO 2 sensor and ca.6.0 µA (S/N ratio: ca.2.0) for the Pt/TiO 2 sensor).In addition, the CO response of the Pd-64Pt/TiO 2 sensor was much larger than that of the Pt/TiO 2 sensor, and the magnitude of both the CO responses in wet H 2 was much smaller than that in dry H 2 in every operating temperature range.Nevertheless, the Pd-64Pt/TiO 2 sensor managed to show a clear response even to 1 ppm CO (ca. 4 µA, S/N ratio: ca.1.5) in wet H 2 .The addition of moisture into the H 2 base gas probably induced the adsorption of water molecules on the electrode surface.Thus, the inhibition of the adsorption of CO by the large amount of the adsorbed water molecules is one of important reasons to decrease the CO responses of both the sensors.However, the magnitude of current of these sensors in wet CO-free H 2 (base gas) was quite comparable to that in dry H 2 , probably because the adsorbed water molecules had little effect on the dissociatively adsorption and dissolution of hydrogen species.This behavior may indicate that the amounts of dissociatively adsorbed and dissolved hydrogen species were sufficiently saturated in the H 2 base gas.
Chemosensors 2018, 6, x 10 of 12 H2 base gas probably induced the adsorption of water molecules on the electrode surface.Thus, the inhibition of the adsorption of CO by the large amount of the adsorbed water molecules is one of important reasons to decrease the CO responses of both the sensors.However, the magnitude of current of these sensors in wet CO-free H2 (base gas) was quite comparable to that in dry H2, probably because the adsorbed water molecules had little effect on the dissociatively adsorption and dissolution of hydrogen species.This behavior may indicate that the amounts of dissociatively adsorbed and dissolved hydrogen species were sufficiently saturated in the H2 base gas.Another possible reason is that the water gas shift reaction (CO + H2O ⇄ CO2 + H2) [33], which proceeds on the electrode surface, decreases the effective concentration of CO on the electrode surface.The clarification of these adsorption and/or reaction mechanism of CO on the electrode surface is indispensable in enhancing these CO-sensing properties in future.

Conclusions
CO-sensing properties of M/TiO2 sensors (M: Pd, Pt, and Pd-nPt) in H2 were investigated in this study.The I-V characteristics of all the M/TiO2 sensors were nonlinear in air and N2, as a typical diode device, and they all showed large H2 responses under the same atmospheres.On the other hand, only the Pd/TiO2 sensor showed no CO response in H2, but the Pt/TiO2 and Pd-nPt/TiO2 sensors responded also to CO in H2, after they were annealed under H2 atmosphere at 400 °C.Among them, the Pd-64Pt/TiO2 sensor showed the largest CO response at 100 °C in H2.The Schottky barrier of all the sensors was negligibly small in H2, but the mixing of Pt into Pd decreased the amount of dissolved hydrogen species, and increased the amount of dissociatively adsorbed hydrogen on the electrode H2 base gas probably induced the adsorption of water molecules on the electrode surface.Thus, the inhibition of the adsorption of CO by the large amount of the adsorbed water molecules is one of important reasons to decrease the CO responses of both the sensors.However, the magnitude of current of these sensors in wet CO-free H2 (base gas) was quite comparable to that in dry H2, probably because the adsorbed water molecules had little effect on the dissociatively adsorption and dissolution of hydrogen species.This behavior may indicate that the amounts of dissociatively adsorbed and dissolved hydrogen species were sufficiently saturated in the H2 base gas.Another possible reason is that the water gas shift reaction (CO + H2O ⇄ CO2 + H2) [33], which proceeds on the electrode surface, decreases the effective concentration of CO on the electrode surface.The clarification of these adsorption and/or reaction mechanism of CO on the electrode surface is indispensable in enhancing these CO-sensing properties in future.

Conclusions
CO-sensing properties of M/TiO2 sensors (M: Pd, Pt, and Pd-nPt) in H2 were investigated in this study.The I-V characteristics of all the M/TiO2 sensors were nonlinear in air and N2, as a typical diode device, and they all showed large H2 responses under the same atmospheres.On the other hand, only the Pd/TiO2 sensor showed no CO response in H2, but the Pt/TiO2 and Pd-nPt/TiO2 sensors responded also to CO in H2, after they were annealed under H2 atmosphere at 400 °C.Among them, the Pd-64Pt/TiO2 sensor showed the largest CO response at 100 °C in H2.The Schottky barrier of all the sensors was negligibly small in H2, but the mixing of Pt into Pd decreased the amount of dissolved hydrogen species, and increased the amount of dissociatively adsorbed hydrogen on the electrode Another possible reason is that the water gas shift reaction (CO + H 2 O CO 2 + H 2 ) [33], which proceeds on the electrode surface, decreases the effective concentration of CO on the electrode surface.The clarification of these adsorption and/or reaction mechanism of CO on the electrode surface is indispensable in enhancing these CO-sensing properties in future.

Conclusions
CO-sensing properties of M/TiO 2 sensors (M: Pd, Pt, and Pd-nPt) in H 2 were investigated in this study.The I-V characteristics of all the M/TiO 2 sensors were nonlinear in air and N 2 , as a typical diode device, and they all showed large H 2 responses under the same atmospheres.On the other hand, only the Pd/TiO 2 sensor showed no CO response in H 2 , but the Pt/TiO 2 and Pd-nPt/TiO 2 sensors responded also to CO in H 2 , after they were annealed under H 2 atmosphere at 400 • C.Among them, the Pd-64Pt/TiO 2 sensor showed the largest CO response at 100 • C in H 2 .The Schottky barrier of all the sensors was negligibly small in H 2 , but the mixing of Pt into Pd decreased the amount of dissolved hydrogen species, and increased the amount of dissociatively adsorbed hydrogen on the electrode surface.In addition, the mixing of Pt into Pd also increased the amount of CO adsorbed on the electrode surface and the adsorbed CO species interrupted the adsorption of hydrogen species.The optimal balance between them, which was attained by the compositional control of the Pd-nPt electrode, probably enhanced the CO-sensing properties of the Pd-nPt/TiO 2 sensor.

Figure 1 .
Figure 1.Schematic drawing of a diode-type gas sensor.

Figure 1 .
Figure 1.Schematic drawing of a diode-type gas sensor.

Figure 2 .
Figure 2. SEM photographs of surfaces of (a) a Ti plate; (b) a bare TiO2 film; and (c) a Pd-64Pt-coated TiO2 film.The TiO2 film was fabricated on a Ti plate by anodization.

Figure 2 .
Figure 2. SEM photographs of surfaces of (a) a Ti plate; (b) a bare TiO 2 film; and (c) a Pd-64Pt-coated TiO 2 film.The TiO 2 film was fabricated on a Ti plate by anodization.

Figure 5 .
Figure 5. I-V characteristics of a Pd-64Pt/TiO2 sensor, which was annealed at 400 °C in H2 for 1 h, as a representative, (a) in dry N2 at 100 °C and 250 °C, and (b) in dry H2 and 80 ppm CO balanced with dry H2 at 100 °C.

3. 3 . 2 Figure 5
Figure5shows I-V characteristics of the Pd-64Pt/TiO 2 sensor as a representative, in dry N 2 at 100 • C and 250 • C, and in dry H 2 and 80 ppm CO balanced with dry H 2 at 100 • C.This sensor was annealed at 400 • C in H 2 for 1 h, before the measurement, because of the enhancement in the thermal and chemical stability.The annealing at 400 • C in H 2 increased the magnitude of current of the Pd-64Pt/TiO 2 sensor in dry N 2 , and the current of the annealed Pd-64Pt/TiO 2 sensor was a little larger than that of the non-annealed Pd-64Pt/TiO 2 sensor at 250 • C (see Figure3a(ii)), in dry N 2 .In addition, a decrease in the operating temperature reduced the magnitude of current, the nonlinearity of these I-V characteristics of the annealed Pd-64Pt/TiO 2 sensor was comparable to that of the non-annealed Pd-64Pt/TiO 2 sensor at 250 • C in dry N 2 .On the other hand, the I-V relationship in dry H 2 was considerably linear, due to the negligibly small Schottky barrier, and the I-V relationship of the Pd/TiO 2 and Pt/TiO 2 sensors at 100 • C in dry H 2 was also quite similar to that of the Pd-64Pt/TiO 2 sensor (not shown here).The addition of 80 ppm CO into dry H 2 reduced the magnitude of current of the Pd-64Pt/TiO 2 sensor at 100 • C, which means that the sensor is capable of detecting CO in dry H 2 at least.

Figure 5 .
Figure 5. I-V characteristics of a Pd-64Pt/TiO2 sensor, which was annealed at 400 °C in H2 for 1 h, as a representative, (a) in dry N2 at 100 °C and 250 °C, and (b) in dry H2 and 80 ppm CO balanced with dry H2 at 100 °C.

Figure 5 .
Figure 5. I-V characteristics of a Pd-64Pt/TiO 2 sensor, which was annealed at 400 • C in H 2 for 1 h, as a representative, (a) in dry N 2 at 100 • C and 250 • C, and (b) in dry H 2 and 80 ppm CO balanced with dry H 2 at 100 • C.

Figure 7 .
Figure 7. Operating temperature dependences of responses of Pt/TiO2 and Pd-64Pt/TiO2 sensors, which were annealed at 400 °C in H2 for 1 h in advance, to 80 ppm CO in dry H2 (applied forward bias: +1.0 mV).

Figure 7 .
Figure 7. Operating temperature dependences of responses of Pt/TiO 2 and Pd-64Pt/TiO 2 sensors, which were annealed at 400 • C in H 2 for 1 h in advance, to 80 ppm CO in dry H 2 (applied forward bias: +1.0 mV).

Figure 8 .
Figure 8. Variations in (a) responses to 80 ppm CO and (b) response and recovery times (TRP and TRC, respectively) of Pd-nPt/TiO2 sensors (including Pd/TiO2 (n = 0) and Pt/TiO2 (n = 100)), which were annealed at 400 °C in H2 for 1 h in advance, at 100 °C in dry H2 (applied forward bias: +1.0 mV), as a function of the amount of Pt (wt %) in the Pd-nPt electrodes (n).TRP and TRC of the Pd/TiO2 (Pd-0Pt/TiO2) sensor are not shown in this figure, because the sensor showed no response to CO.

Figure 8 .
Figure 8. Variations in (a) responses to 80 ppm CO and (b) response and recovery times (T RP and T RC , respectively) of Pd-nPt/TiO 2 sensors (including Pd/TiO 2 (n = 0) and Pt/TiO 2 (n = 100)), which were annealed at 400 • C in H 2 for 1 h in advance, at 100 • C in dry H 2 (applied forward bias: +1.0 mV), as a function of the amount of Pt (wt %) in the Pd-nPt electrodes (n).T RP and T RC of the Pd/TiO 2 (Pd-0Pt/TiO 2 ) sensor are not shown in this figure, because the sensor showed no response to CO.

Figure 8 .
Figure 8. Variations in (a) responses to 80 ppm CO and (b) response and recovery times (TRP and TRC, respectively) of Pd-nPt/TiO2 sensors (including Pd/TiO2 (n = 0) and Pt/TiO2 (n = 100)), which were annealed at 400 °C in H2 for 1 h in advance, at 100 °C in dry H2 (applied forward bias: +1.0 mV), as a function of the amount of Pt (wt %) in the Pd-nPt electrodes (n).TRP and TRC of the Pd/TiO2 (Pd-0Pt/TiO2) sensor are not shown in this figure, because the sensor showed no response to CO.

Figure 9 .
Figure 9. Concentration dependences of CO responses of Pt/TiO 2 and Pd-64Pt/TiO 2 sensors, which were annealed at 400 • C in H 2 for 1 h in advance, at 100 • C in dry H 2 (applied forward bias: +1.0 mV).

Figure 10 .
Figure 10.Operating temperature dependences of responses of Pt/TiO2 and Pd-64Pt/TiO2 sensors, which were annealed at 400 °C in H2 for 1 h in advance, to 80 ppm CO at 100 °C in wet H2 (applied forward bias: +1.0 mV).

Figure 11 .
Figure 11.Concentration dependences of CO responses of Pt/TiO2 and Pd-64Pt/TiO2 sensors, which were heat-treated at 400 °C in H2 for 1 h in advance, at 100 °C in wet H2 (applied forward bias: +1.0 mV).

Figure 10 .
Figure 10.Operating temperature dependences of responses of Pt/TiO 2 and Pd-64Pt/TiO 2 sensors, which were annealed at 400 • C in H 2 for 1 h in advance, to 80 ppm CO at 100 • C in wet H 2 (applied forward bias: +1.0 mV).

Figure 10 .
Figure 10.Operating temperature dependences of responses of Pt/TiO2 and Pd-64Pt/TiO2 sensors, which were annealed at 400 °C in H2 for 1 h in advance, to 80 ppm CO at 100 °C in wet H2 (applied forward bias: +1.0 mV).

Figure 11 .
Figure 11.Concentration dependences of CO responses of Pt/TiO2 and Pd-64Pt/TiO2 sensors, which were heat-treated at 400 °C in H2 for 1 h in advance, at 100 °C in wet H2 (applied forward bias: +1.0 mV).

Figure 11 .
Figure 11.Concentration dependences of CO responses of Pt/TiO 2 and Pd-64Pt/TiO 2 sensors, which were heat-treated at 400 • C in H 2 for 1 h in advance, at 100 • C in wet H 2 (applied forward bias: +1.0 mV).

Table 1 .
Composition and fabrication conditions of Pd, Pt, and Pd-nPt electrodes fabricated on a TiO 2 film by rf magnetron sputtering.