Pd/ZnO nanorods based sensor for highly selective detection of extremely low concentration hydrogen

We report highly hydrogen selective Pd contacted ZnO nanorods based sensor detecting low concentration even at low operating temperature of 50 °C. The sensor performance was investigated for various gases such as H2, CH4, H2S and CO2 at different operating temperatures from 50 °C to 175 °C for various gas concentrations ranging from 7 ppm to 10,000 ppm (1%). The sensor is highly efficient as it detects hydrogen even at low concentration of ~7 ppm and at operating temperature of 50 °C. The sensor’s minimum limit of detection and relative response at 175 °C were found 7 ppm with ~38.7% for H2, 110 ppm with ~6.08% for CH4, 500 ppm with ~10.06% for H2S and 1% with ~11.87% for CO2. Here, Pd exhibits dual characteristics as metal contact and excellent catalyst to hydrogen molecules. The activation energy was calculated for all the gases and found lowest ~3.658 kJ/mol for H2. Low activation energy accelerates desorption reactions and enhances the sensor’s performance.

such as nanotubes, nanorods (NRs), nanowires and nanoneedles shows high sensitivity than thin films because of high aspect ratio and higher diffusion of gases [26][27][28] . Resistive gas sensors with noble metals such as Pd, Pt and Au exhibit superior sensing response for hydrogen gas due to their catalytic activity and solubility of hydrogen atom 29,30 . Among them, Pd has been most studied in hydrogen sensor owing to its excellent capability of adsorption of hydrogen molecules and their dissociation into hydrogen atoms. Moreover, the diffusion of hydrogen atoms is accelerated throughout the surface and leads to increase the interaction with active sites of the surface 31 . Basu et al. has fabricated ZnO thin films based sensor (Pd/ZnO/p-Si and Pd/ZnO/Zn) and found highly hydrogen selective sensor in the range of 2000-20000 ppm H 2 concentration 32 . Lupan et al. fabricated single Cd doped ZnO nanowire using focused ion beam and H 2 was detected down to 100 ppm at room temperature with good sensitivity while poor response to CH 4 , C 2 H 5 OH, O 2 , LPG and NH 3 33 . Mondal et al. synthesized the ZnO-SnO 2 composite type hydrogen sensor for different concentration of tested gases at 150 °C and the cross sensitivity of this sensor to CH 4 and CO was not high at same temperature 34 4 , H 2 S and CO 2 at various concentration and operating temperature ranging from 50 °C to 175 °C. Gas sensing mechanism based on activation energy is discussed to explain high selectivity towards hydrogen gas in comparison to other gases.

Results
ZnO NRs were uniformly grown on Si substrate and AFM images are shown in Fig. 1(a,b). Nanorods are highly crystalline and grown along c-axis which is reported in our previous study 36 . The schematic diagram and top view of the device are shown in Fig. 1(c,d). The sensor's performance depends on chemisorption reaction of reactive gases with adsorbed oxygen ions on ZnO NRs surface. The performance of Pd contacted ZnO NRs based sensor may be influenced by gas type, concentrations and operating temperatures. The sensor's relative response has been studied for various gases such as H 2 , CH 4 , H 2 S and CO 2 at different concentrations and operating temperatures. Figure 2 shows sensor's relative response curve with time for 7 ppm, 12 ppm, 55 ppm, 110 ppm, 500 ppm, 1000 ppm and 10,000 ppm (1%) concentrations of gases (a) H 2 , (b) CH 4 , (c) H 2 S and (d) CO 2 at operating temperature ranging from 50 °C to 175 °C. The response of the sensor increases with hydrogen concentration and temperature. It was found that the sensor is able to detect very low hydrogen concentration of ~7 ppm and at low temperature (50 °C). The relative response was measured 13.86% at 50 °C and 38.47% at 175 °C. The change in relative response increases with operating temperature due to creating more chemisorbed oxygen ions on the ZnO NRs surface which further enhances desorption of oxygen. The sensor response increases from 13.86% to 37.13% with concentration from 7 to 10,000 ppm and operating temperature 50 °C. A large variation in sensor's response is observed with low concentration below 1000 ppm since hydrogen molecules react with adsorbed oxygen ions and decrease the depletion region as well as reduce the barrier height at Pd/ZnO junction. Above 1000 ppm of hydrogen chemisorbed reaction is almost saturated on ZnO NRs which is shown by very less increase in sensor's response from 86.39% to 91.26% at 175 °C of operating temperature. To study the selectivity the relative response of the device was measured for various gases such as CH 4 , H 2 S and CO 2 with different concentration from 7 ppm to 1%. Figure 2b shows relative response curve with time of CH 4 gas and minimum detection limit was found 110 ppm at 100 °C. During loading/deloading of the gas, methane gas molecules react with adsorbed oxygen ions on ZnO NRs surface and show decrease/increase of ZnO NRs resistance, which is similar to hydrogen gas behaviour. Although CH 4 gas is highly stable below 100 °C and with increasing operating temperature above this limit, C-H bonds starts breaking and it gives more H atoms to react with adsorbed oxygen ions and decreases sensor's resistance. Furthermore, the relative response of H 2 S gas is shown in Fig. 2c and minimum detection limit was observed 500 ppm at 125 °C and maximum relative response is around 30% for 1%. This indicates less reactivity of H 2 S molecules with adsorbed oxygen in comparison to hydrogen and methane gases. Sensor's relative response was also measured with CO 2 gas and shown in Fig. 2d. The minimum concentration was detected 1% at 125 °C, because of its highest stable nature in comparison to other reactive gases. Hence, the sensor shows high relative response as well as high selectivity for hydrogen gas in comparison to other gases. Sensor's relative response variation with operating temperature and concentration is shown in Fig. 3 for (a) H 2 , (b) CH 4 , (c) H 2 S and (d) CO 2 . Figure 4(a-c) depicts hysteresis studies for H 2 , CH 4 and H 2 S at 150 °C for increasing and decreasing gas concentrations. A stable behaviour of the sensors has been observed for each target gas.

Discussion
Gas sensing mechanism is basically surface phenomenon. The fast adsorption-desorption cycle of chemisorbed oxygen on the surface of ZnO nanorods is highly desirable to make excellent gas sensor. Thus surface sensing properties for various gases could be understood by activation energy phenomenon. The rate of resistance changes with respect to temperature for particular target gas may be assumed by Arrhenius equation which can be expressed as 37 : where R is resistance of the sensor, K is the Boltzmann constant, T is absolute temperature and ΔE is the activation energy. For activation energy calculation, firstly the rate of change in resistances were evaluated from resistance verses time curves on exposure of (a) H 2 , (b) CH 4 , (c) H 2 S and (d) CO 2 gases with 1% concentration at different operating temperature and shown in Fig. 5(a-d). It can be seen that the resistance of the sensor is decreased drastically with increase in temperature for hydrogen gas as compared to the other reactive gases. Arrhenius Plot of rate of resistance change with temperature after exposer of 1% concentration of (a) H 2 , (b) CH 4 , (c) H 2 S and (d) CO 2 gases are shown in Fig. 6(a-d). The activation energy has been evaluated from slope of Arrhenius plot which is basically linear fitting of resistance rate with respect to temperature.  38 . The selectivity histogram of the sensor is shown in Fig. 7 for different concentration at 175 °C operating temperature. The sensor is able to detect minimum 7 ppm hydrogen concentration with ~38.47% relative response at 50 °C. However, for other gases minimum gas concentration sensing limits is quite high in comparison to hydrogen gas. As we know that hydrogen is smallest molecule with lowest activation energy which enhances surface reaction and corresponds to large change in depletion region of ZnO NRs. The minimum detected concentration for CH 4 gas was observed 110 ppm with 6.5% sensor response which is 10.11 times less than hydrogen gas response (~65.75). H 2 S gas has quite high activation energy (13.556 kJ/mol) in comparison to both Hydrogen and methane gas, shows minimum sensing concentration to 500 ppm with relatively low sensor response ~10.82%. These sensor response is around 7 times less than hydrogen response (~76.56%) for 500 ppm concentration. CO 2 is highly stable gas with highest activation energy which makes it least reactive gas in comparison to other gases. For CO 2 , minimum gas detection limit was 1% with sensor response ~12.29% which is 7.42 times less than hydrogen response (~91.26%). Cross selectivity has been calculated for Pd/ZnO nanorods based sensor to acquire more clarity for better selectivity. Figure 8 depicts cross selectivity response curve for various gases mixture (500 ppm H 2 , 500 ppm H 2 + 500 ppm CH 4 and 500 ppm H 2 + 500 ppm CH 4 + 500 ppm H 2 S) at 150 °C operating temperature. It is found that mixing H 2 gas with other gases with same concentration, there is slight increase in sensor's response in comparison to pure hydrogen gas. This result implies that Pd/ZnO nanorods based sensor shows high selectivity towards hydrogen gas. Gas sensing mechanism. Based on experimental investigation a gas sensing mechanism is proposed and shown in Fig. 9. From the figure, the relative change in depletion region of ZnO can be seen in presence of (a) Air, (b) H 2 , (c) CH 4 , (d) H 2 S and (e) CO 2 . The sensing mechanism is mainly influenced by activation energy of target gases on the surface of ZnO nanorods and a change in barrier height between Pd and ZnO NRs. When sensor chamber is loaded with target gas, these target gas molecule reacts with adsorbed oxygen ions. By removing chemisorbed oxygen ions from ZnO nanorods surface, trapped electron moves back to conduction region and decreases depletion region width of ZnO nanorods to a large extent [39][40][41] as shown in Fig. 9(b-e). For n-type semiconductor materials, CO 2 gas molecules also react with adsorbed oxygen ions and returns excess electrons to material. These reactions causes decreased sensor's resistance while loading of CO 2 gas 42 . Due to increasing activation energy of CH 4 , H 2 S and CO 2 gases in comparison to H 2 , maximum sensor response is observed for hydrogen at even ppm level with maximum change in depletion region. As activation energy of target gases increases, sensor response decreases with relatively small change in depletion region. These results clearly depict gas sensing mechanism schematic in Fig. 9(b-e). Along with activation energy, operating temperature and target gas concentration also plays an important role in sensor's response. With increasing temperature, chemisorption  of oxygen ions on ZnO nanorods surface increases as conduction band electron get sufficient energy to overcome barrier. Due to these enhanced chemisorbed oxygen ions, there is large change in depletion region as well as in sensors resistance. While loading with target gases, it enhances desorption of oxygen ions due to which relative change in sensor resistance increases. With increasing gas concentration, it provides large reaction with adsorbed oxygen ions which further enhances sensor's response.
In case of hydrogen, the dissociation of the molecules over the Pd contact is faster and diffuse the hydrogen atom to the interface due to smallest molecular size 23,43 . Moreover, the conduction of ZnO nanorods is correlated with cross section and given as follow 44 , where G, n, e, l, and µ are electrical conductivity, electron concentration, length of NR and electron mobility, respectively. Fast oxygen desorption will take place when ZnO NRs are exposed to hydrogen and increase the  conduction cross section channel area. When Pd is exposed to hydrogen gas, hydrogen molecule gets diffused into Pd which reduces Schottky barrier to a large extent. Because of Schottky barrier reduction along with depletion region reduction, the combine effect maximizes sensors response and makes ppm level detection possible. However as compared with hydrogen molecules, other target gases such as CH 4 and H 2 S gases are less reactive to Pd contact and slower diffusion due to larger molecular weight and size. Because of these factors, there is a less change in Schottky barrier reduction which causes lesser sensor response in comparisons to hydrogen gas. While relative response of CO 2 gas is lowest occurred by the (i) more activation energy is required to break double bond  than single bond (ii) highest molecular weight as compared to H 2 , CH 4 and H 2 S and (iii) highest activation energy of CO 2 gas provides the small relative response.

Conclusion
We have demonstrated Pd contacted ZnO NRs based sensor which shows high sensor sensitivity and selectivity for hydrogen. The sensor was analysed for selectivity test towards H 2 , CH 4 , H 2 S and CO 2 and found best suited for H 2 even at extreme low concentration (~7 ppm) and low operating temperature 50 °C. Because of lowest activation energy ~3.658 kJ/mol of hydrogen in comparison to other reactive gases, it leads to maximum change in conduction cross section area of ZnO NRs which could be possible cause for excellent sensor's performance. A poor sensitivity was detected for CH 4 at 110 ppm (~6.5% response) which is 10.11 times less than H 2 (~65.75% response), for H 2 S gas at 500 ppm (~10.82% response) which is 7 times less than H 2 (~76.56% response) and for CO 2 at 1% (~12.29% response) and marked 7.42 times less than H 2 (~91.26%). The gas sensing mechanism for Pd contacted ZnO NRs based sensor was also proposed which mainly attributed to activation energy of reactive gases. The proposed sensor is not only highly hydrogen selective but also can operate at low temperature which reduces the power consumption and prove to be energy efficient in its class.

Experimental Techniques
Vertically aligned ZnO NRs were fabricated on n-Si substrate using RF sputtering. Substrate temperature was kept constant at 600 °C during the deposition and other parameters can be found elsewhere 36 . Uniform distribution of ZnO NRs was confirmed by using AFM (Park XE-70) and FESEM techniques. For sensor fabrication, circular palladium (Pd) electrodes of 500 µm diameter were deposited on ZnO NRs with the help of DC Sputtering technique where shadow mask was used. Thickness of Pd electrodes was kept around 150 nm and 500 µm inter-electrode spacing. Gas sensing characterization was carried out in a stainless steel sensing chamber. The chamber was evacuated down to 3 × 10 −3 mbar using a rotary pump. The testing gases were mixed with argon and introduced in the sensing chamber by injecting a pre-calibrated volume of target gas through mixing chamber and a gas tight syringe (Hamilton). Operating temperature was tuned from 50 °C to 175 °C. The sensor's response was measured by Semiconductor Parameter analyser (Keithey 4200-SCS) by applying a constant voltage of 1 V.