Thermochemical hydrogen sensor based on chalcogenide nanowire arrays

The electrolyte used to synthesize the Bi_xTe_1−x nanowires was prepared by dissolving TeO₂ (99.995%, Alfa Aesar, Inc.) and Bi(NO₃)₃·5H₂O (98%, Acros Organic) in concentrated nitric acid (60%, Junsei Chemical). The electrolyte used to synthesize the Sb_xTe_1−x nanowires was prepared by separately dissolving TeO₂ in nitric acid and dissolving Sb₂O₃ (99.9%, Sigma Aldrich) in L-tartaric acid (99.5%, Sigma Aldrich) solution, and then mixing the two solutions to make an electrolyte consisting of 0.02 M SbO⁺, 0.01 M HTeO₂⁺, 0.5 M L-tartaric acid and 1 M nitric acid. All electrochemical depositions were carried out in a 250 ml electrochemical cell using a potentiostat (AMETEK, VersaSTAT3) with a standard three-electrode cell, which consisted of Ag/AgCl as the reference electrode, a platinum-coated titanium strip as the counter electrode and aluminum as the working electrode. In all sensors, a 200 nm thick gold seed layer was first deposited onto an AAO template by means of sputtering to enable subsequent plating of nanowires inside the template.

Herein, we fabricated two types of TCH sensors. One was a monomorphic-type TCH sensor composed of an n-type Bi₂Te₃ nanowire array; the process used to produce this sample is illustrated in figures 1(a)–(d). A gold layer was deposited onto an AAO template with a 200 nm pore size; this layer served both as a working electrode and as a seed layer for the Bi₂Te₃ nanowires (figure 1(b)). Electrochemical deposition of the Bi₂Te₃ nanowires into the template was carried out by applying a voltage (figure 1(c)), and then gold nanowire array were deposited using electrodeposition with gold electrolyte (Techni Gold 25 RTU, TECHNIC) over the Bi₂Te₃ nanowire arrays to form low-resistance Ohmic contacts at the interface between the Bi₂Te₃ nanowires and the electrode, which was deposited by means of sputtering (figure 1(d)). It is important to assure Ohmic contact because high resistance in a TE device can decrease its TE efficiency.

Zoom In Zoom Out Reset image size

The second sensor type was an n–p junction-type TCH sensor composed of n-type Bi₂Te₃ and p-type Sb₂Te₃ nanowire arrays. Partial masking of the AAO template was required to fabricate this device type; schematic illustrations of the fabrication of n–p junction TE layers by electrochemical deposition into the AAO template are given in figures 1(e)–(i). Two square gold films (total area: 50 mm²) were deposited onto the bottom of the AAO template by means of sputtering (figure 1(f)). The template area over one of these two gold electrodes was sealed with stop-off lacquer (Microstop, Pyramid Plastics, Inc.) and then the template was dried for 1 h in an ambient atmosphere. Then, Sb_xTe_1−x nanowires were deposited into the area over the other gold electrode by electrochemical deposition (figure 1(g)). After deposition, the Sb_xTe_1−x nanowire-deposited AAO template was dipped in acetone to remove the lacquer. And then the device was annealed at 120 °C for 1 h to synthesize the Sb₂Te₃ phase. Next, the side of the AAO template filled with the Sb_xTe_1−x nanowires was sealed with lacquer, and Bi_xTe_1−x nanowires were electrochemically deposited into the open area of the template over the other gold electrode (figure 1(h)). And then a top-side gold nanowire arrays were deposited over both nanowire regions by electrochemical deposition. Finally, gold electrode was deposited using sputter equipment (figure 1(i)).

A Pt/γ-Al₂O₃ catalyst was used to oxidize hydrogen gas on the surface of the devices, thereby providing a heat source that would create a temperature gradient across the TE layer of the devices composed of either n-type only or both n- and p-type chalcogenide nanowire arrays. In other words, the Pt/γ-Al₂O₃ catalyst responded to hydrogen gas exposure by heating the device, thereby inducing a detectable electric current through the TE nanowire arrays. A catalyst precursor solution of Pt/γ-alumina was prepared by a sol–gel method, using a method based on that reported by Yoldas et al [18]. Aluminum hydroxide was obtained by the hydrolysis of aluminum isopropoxide (AIP, ≥99.9%, Sigma Aldrich) in 90 °C deionized water for 1 h. During this reaction, a condenser was used to prevent loss of the volatile materials from the chemical reactor. Aluminum hydroxide precipitated to the bottom of the reactor after the reaction. To peptize the precipitated aluminum hydroxide, nitric acid (60%, Junsei) was added to adjust the AIP/H₂O/HNO₃ mole ratio of the solution to 1:100:0.7, and the solution was then stirred at 90 °C for 5 h. The resulting alumina sol was allowed to cool slowly at room temperature and was then dissolved in a solution of H₂PtCl (8 wt%, Sigma Aldrich) with additional stirring for 1 h, thereby obtaining a yellowish Pt/γ-Al₂O₃ precursor solution. This solution was dried at 80 °C for 24 h in an oven to remove water and other volatile components, and then the dried Pt/γ-Al₂O₃ precursor was deagglomerated by grinding using a mortar. Finally, it was calcined at 600 °C for 1 h to remove the organic additive. Mesoporous alumina was obtained to use as the heating catalyst on TE layer as-fabricated by electrochemical deposition, this synthesis method of mesoporous alumina catalyst has been reported in our previous investigation [19].

The low thermal conductivity of the AAO template was a necessary property for its use as a matrix for synthesizing the one-dimensional TE nanostructures. If the thermal conductivity of the template were greater than that of the TE materials, the performance of the resulting TE device would be worsened due to heat loss by thermal diffusion into the template. The thermal conductivities of the TE materials bismuth telluride (bulk Bi₂Te₃) and antimony telluride (bulk Sb₂Te₃) are 1.5 W mK⁻¹ [20] and 1.6 W mK⁻¹ [21], respectively. Therefore, in this system the template must have lower thermal conductivity than these TE materials. A commercially available AAO template with 200 nm pores (Whatman Anodisc 13, Germany) was used to fabricate the TE layer of chalcogenide nanowire arrays; the thermal conductivity of the AAO template was measured by multiplying its specific heat capacity (С_p), density (ρ) and thermal diffusivity (λ) based upon the laser flash method:

$$\begin{eqnarray}&&k\;=\;Cp\cdot \rho \cdot \lambda .\end{eqnarray} \tag{ 1 }$$

The density of gold deposited on the AAO template was measured by Archimedes' principle, yielding the density of 2.683 ± 0.23 g cm⁻³. Before this measurement, the AAO template was immersed in boiling deionized water at about 100 °C to fill its pores. Specific heat capacity and thermal diffusivity were analyzed at 25 °C by differential scanning calorimetry (DSC, TA instrument, Q200) and LFA (NETZSCH, LFA 447); the measured values were 0.72 ± 0.14 J g⁻¹ °C⁻¹ and 0.54 ± 0.13 mm² s⁻¹, respectively. Thus, the thermal conductivity of the gold-deposited AAO template with 200 nm pores was determined to be 1.18 ± 0.22 W mK⁻¹ at 300 K. In previous studies, Borca-Tasciuc et al [22] reported the thermal conductivity of the AAO template to be 1.3 W mK⁻¹ at 300 K, and Ogden et al [23] reported values between 0.5 and 1.0 W mK⁻¹; our present results were similar to these results. These results suggest the possibility of a new type of TCH sensor using an AAO template.

Stoichiometry is important and necessary in chemistry because it can dictate the properties of materials, such as their electrical and TE properties [24]. In other words, optimum stoichiometry of a material leads to high performance in terms of its TE properties. The content of Bi deposited in the nanowires was studied as a function of the potential applied during deposition; the Bi content increased from about 36 to 83 at% Bi with increasing negative potential from 0.125 to −0.075 V (figure 2(a)). In this case, the Bi used to form Bi_xTe_1−x was strongly influenced by the Bi³⁺ concentration when compared to the relatively lower HTeO₂⁺ concentration. The stoichiometric composition was realized by deposition under the potential of 0.075 V in an electrolyte containing 0.07 M Bi³⁺ and 0.01 M HTeO₂⁺ in 1 M HNO₃. Like Bi, the deposited Sb content increased from about 4 to 38 at% Sb with increasing negative potential from −0.10 to −0.19 V in an electrolyte containing 0.02 M SbO⁺, 0.01 M HTeO₂⁺, 0.5 M C₄H₆O₆, and 1 M HNO₃(figure 2(b)). In the fabrication of Sb_xTe_1−x nanowires, it was difficult to obtain the stoichiometric composition. To obtain stoichiometric Sb₂Te₃ nanowires, we annealed nanowires of a near-stoichiometric composition resulting from deposition at an applied voltage of −0.17 V. Annealing was performed at 120 °C for 1 h in atmospheric conditions. The morphology and composition of the Bi₂Te₃ and Sb₂Te₃ nanowires were investigated by field emission scanning electron microscopy and energy-dispersive x-ray spectroscopy (EDS). Figures S1(a)–(c) shows cross-sectional images and EDS composition results of Bi₂Te₃ nanowire arrays synthesized by electrochemical deposition with an applied voltage of 0.075 V. As shown, the Bi_xTe_1−x nanowires grew well and with high density, and their composition approached that of the Bi₂Te₃ phase. Cross-sectional views and EDS composition results of Sb_xTe_1−x nanowire arrays before and after annealing are shown in figures S1(d)–(i). The composition of as-synthesized Sb_xTe_1−x nanowire arrays synthesized with an applied voltage of −0.17 V were revealed to be 26.11% Sb and 73.89% Te. The annealing at 120 °C for 1 h, changed the composition of the Sb_xTe_1−x nanowire arrays to 37.24% Sb and 62.76% Te, which is near stoichiometric. X-ray diffraction (XRD) patterns were obtained for accurate analysis of the phase components and structure of each nanowire. The XRD pattern of the as-synthesized Bi₂Te₃ nanowires is shown in figure 3(a). All of the detected peaks were indexed to the rhombohedral structure. The XRD pattern of the phase corresponds with that of Bi₂Te₃ (JCPDS: 00-015-0863), and no other phases existed in the measured product. Annealing effects of Sb_xTe_1−x nanowires were investigated by XRD analysis (figure 3(b)). Before annealing, a Sb_xTe_1−x nanowire was revealed in which Sb_0.405Te_0.595 (00-045-1228) and Te (00-036-1452) showed a mixed amorphous structure. However, the crystal structure was changed from amorphous to crystalline by annealing, yielding a Sb₂Te₃ (00-015-0874) phase without other phases such as tellurium. Figures 4 and 5 show HR-TEM and fast Fourier transform (FFT)-converted SAED patterns of a Bi₂Te₃ nanowire and a Sb₂Te₃ nanowire, respectively. Actually, it is difficult to analyze the microstructure of a nanowire with the diameter of 200 nm. Therefore, we used nanowires under 200 nm for TEM analysis; these were grown in an AAO template with 100 nm pore diameter by means of electrochemical deposition, yielding as-deposited nanowires 100 nm in diameter (figure 4(a)). FFT-converted SAED patterns confirmed the synthesis of polycrystalline Bi₂Te₃ nanowires without elemental Bi or Te (figures 4(b)–(f)); FFT patterns of HR-TEM images of these nanowires were consistent with the d-spacing value of the (015), (110), (0, 2, 10) and (205) lattice planes of the rhombohedrally structured Bi₂Te₃ phase, consistent with XRD results. HR-TEM images and FFT-converted SAED patterns of a Sb₂Te₃ nanowire after annealing at 120 °C for 1 h are shown in figure 5. The nanowire was synthesized in an AAO template with the pore diameter of 100 nm, the same as that used to fabricate the Bi₂Te₃ nanowire, and yielding a nanowire 100 nm in diameter, with d-spacing values of the (110), (015) and (0, 0, 18) lattice planes consistent with the rhombohedrally structured Sb₂Te₃ phase. EDS analysis of the Bi₂Te₃ and Sb₂Te₃ nanowires was conducted to confirm their elemental contents. For the Bi₂Te₃ nanowire 100 nm in diameter, the EDS dot scan profiles showed the Bi/Te atomic ratio of 40.59 ± 0.6 to 59.41 ± 0.6 (figure S2(a)), indicating that the nanowire was composed of the stoichiometric Bi₂Te₃ phase. EDS dot scan profiles of the Sb₂Te₃ nanowire showed the Sb/Te atomic ratio of 34.93 ± 1 to 65.07 ± 1 (figure S2(b)), which was within the error detection range of 5% for stoichiometric Sb₂Te₃. To characterize the semiconducting behavior at the interface between the as-deposited nanowires and the electrolyte, Mott–Schottky measurement was performed [25] by means of an electrochemical workstation operated using the ac impedance method, which can demonstrate the carrier types (hole, p-type or electrons, n-type) and carrier concentrations of nanowire arrays. All these measurements were conducted in a conventional three-electrode system with the AAO template, using the deposited Bi₂Te₃ and Sb₂Te₃ nanowire arrays as the working electrode, Ag/AgCl with saturated KCl as a reference electrode and a Pt electrode as a counter electrode. The electrolyte used consisted of 0.1 M NaClO₄ and 0.1 M HClO₄ in deionized water. The capacitance of the chalcogenide nanowire-containing AAO template/electrolyte system was measured over the ac frequency range of 10–200 kHz with the potential ranging from −0.1 to 0.1 V versus Ag/AgCl at the scan rate of 1 mV s⁻¹. Mott–Schottky plots of the Bi₂Te₃ and Sb₂Te₃ nanowire arrays synthesized by electrochemical deposition show positive slopes for the Bi₂Te₃ nanowire arrays, indicating n-type semiconducting behavior, and negative slopes for the Sb₂Te₃ nanowire arrays, indicating p-type behavior (figure S3). The average carrier concentrations of the Bi₂Te₃ and Sb₂Te₃ nanowires were revealed to be 1.31 × 10²⁰ (±1.03 × 10²⁰) cm⁻³ and 3.70 × 10²⁰ (±1.96 × 10²⁰) cm⁻³, respectively; these values are similar to previously reported results [26, 27].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Sensing properties, including gas detection and response time, were measured using a custom-made apparatus. The output voltage signal of each TCH sensor under test was measured using a nanovoltmeter (2182 A, Keithley) that was placed in contact with the thermochemical sensor. Temperature change due to hydrogen oxidation was monitored by using an IR thermometer (CM sensor, Raytek) located on the top side of the TE layer. As-synthesized 2 vol% mesoporous Pt/Al₂O₃ catalyst (0.05 g) was placed on top of the TE layer, the nanowire-embedded AAO template. The catalyst played a supporting role in the TCH sensor. The two internal electric wires were made of copper. The release of a H₂/air mixture gas for sensing was repeated with regular time interval for 180 s on, 600 s off at fixed a 100 sccm flow rate. Hydrogen sensing properties of the monomorphic-type and n–p junction-type TCH sensors were studied as a function of hydrogen concentration (0.5–5 vol%); for both sensor types, the output voltage signal increased with increasing temperature at the flow rate of 100 sccm (figure 6). For the monomorphic-type TCH sensors, the voltage signal reached 23.7 μV at the hydrogen concentration of 5 vol%, while the n–p junction-type TCH sensor yielded a 215 μV signal at 5 vol%, which is a 9-fold increase compared that of the monomorphic-type TCH sensor. The n–p junction-type TCH sensor was serially connected to n-type and p-type nanowire arrays, which gathered much more output voltage than that of the monomorphic-type TCH sensor under the same temperature gradient. The output voltage profiles were collected for various flow rates (50–300 sccm) of 1% hydrogen gas; like the effect of concentration, the change voltage showed constantly increase with respect to changes in the flow rate of hydrogen. Also, the n–p junction-type TCH sensor was 10.6 times as effective as the monomorphic-type TCH sensor at 1 vol% H₂/air (figure 7). For practical applications, the LDL of the hydrogen sensor should be defined. Here, it was determined by using an n–p junction-type TCH sensor to improve the sensitivity under low-temperature conditions. Low-concentration hydrogen detection of the n–p junction-type TCH sensor was studied by plotting its response to various hydrogen concentrations at the constant flow rate of 500 sccm; the sensor was able to detect 400 ppm of hydrogen gas without external interference such as moisture (figure 8).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Generally, the response of TE-type hydrogen sensors is relatively slow compared to that of semiconductor-type hydrogen sensors. Some studies were performed to achieve fast response in TE hydrogen sensors. A response time of 30 s at 3 vol% hydrogen in air was achieved by Luan et al [28], while Nishibori et al achieved an average response time of 2.5 s for 10 000 ppm hydrogen in air by using a Pt catalyst and heating to 100 °C [12]. However, the speed of detecting high concentrations of hydrogen gas at room temperature is still insufficient, and more systematic studies are required for practical applications. Specifically, rapid gas detection at high concentrations is important for practical applications. Output voltage time series of the n–p junction-type TCH sensor in response to hydrogen gas concentrations from 10 to 45 vol% H₂/air are shown in figure 9. All experiments were carried out using Pt/Al₂O₃ catalyst (0.004 g), and included the continuous release of a mixed gas of H₂/air for 120 s with a 500 sccm flow rate, which was stopped when voltage signal became saturated. The response times (t₉₀) of TCH sensor with hydrogen concentrations were calculated using this time-dependent voltage signal. It was found that the response time gradually decrease with increasing hydrogen concentration at fixed a 500 sccm flow rate. The upper TE layer was heated by the Pt/Al₂O₃ catalyst when it reacted with hydrogen, so the reaction with higher concentrations of hydrogen gas increased the output voltage signal of the TCH sensor within a shorter time. In other words, the temperature difference between the upper and lower part of the TCH sensor was reduced by increasing the temperature of upper part resulting in fast voltage saturation due to decline of delta-T (temperature difference) of the TE layer. This allowed fast response by the TCH sensor. Also, for concentrations studied above 30 vol% H₂/air, output voltage rapidly decreased and the sensor responded very rapidly. These results are caused by the use of dry air, which contains 79% nitrogen and 21% oxygen, and enables the n–p junction-type TCH sensor to detect the maximum 30 vol% hydrogen because this blend achieves the maximally efficient ratio of 2:1 hydrogen:oxygen. Consequently, rapid response times of TCH sensor were achieved because a small amount of catalyst easily reached the saturation temperature. Further, it is possible to adjust the response time by varying the catalyst amount without requiring the use of another system.

Zoom In Zoom Out Reset image size