Effect of Platinum Doping on the Morphology and Sensing Performance for CuO-Based Gas Sensor

Pristine and Pt-doped CuO nano-flowers were synthesized by a simple water bath heating method in this paper. Highly sensitive hydrogen sulfide (H2S) gas sensors based on Pt-doped CuO nano-flowers were fabricated. Scanning electron microscopy, X-ray diffraction, inductively coupled plasma atomic emission spectrometry, and energy dispersive X-ray spectroscopy were used to examine the characteristics and morphology of materials. The sensing performances of sensors with different concentrations of Pt dopants were evaluated at different operating temperatures. The results indicated that the CuO sensor doped with 1.25 wt % Pt exhibited the highest response (Rg/Ra, where Rg is the resistance in gas, and Ra is the resistance in air) of 135.1 to 10 ppm H2S at 40 ◦C, which was 13.1 times higher than the response of a pure CuO sensor. Pt doping also plays an important role for the enhancement of H2S selectivity against C2H5OH, NH3, H2, CH3COCH3, and NO2.


Introduction
Hydrogen sulfide (H 2 S) is a toxic, harmful, and malodorous gas, which is usually produced by petroleum refining, as well as agricultural and biogas production [1][2][3].The threshold limit value for H 2 S gas (recommended by the American Conference of Government Industrial Hygienists) is 10 ppm [4].Furthermore, H 2 S gas is liable to explode at a high temperature.Therefore, it is urgent to develop a new type of sensors for detecting H 2 S with a low operating temperature.Although researchers have developed some detecting methods, such as fluorescent probes [5], electrogenerated chemiluminescene [6], and chemical methods [7] to detect H 2 S, these methods still have shortcomings such as complex detection process, and high cost.Among the various detection methods, metal oxide semiconductor gas sensors have been widely used in the detecting of toxic, flammable, and explosive gases; monitoring air quality; and food processing, due to their simplicity, high sensitivity, low cost, and compatibility with modern electronic devices [8,9].In recent years, many researchers have utilized metal oxide semiconductor sensors to detect H 2 S gas, including CuO [10,11], In 2 O 3 [12], ZnO [13,14] SnO 2 [15], and WO 3 [16].Cupric oxide (CuO), as a typical p-type semiconductor with a band gap of 1.2 eV [17], has been widely investigated in lithium-ion batteries [18], super-capacitors [19], heterogeneous catalysts [20], and particularly in gas detection devices.The gas sensitivity of CuO-based gas sensors has been widely reported.Ramgir et al. [10] reported H 2 S gas sensors based on CuO thin films, and the results showed that the response of CuO thin films to sub-ppm (100-400 ppb) H 2 S at room temperature was highly reversible.However, once the concentration was higher than 400 ppb, the response became irreversible.To improve the performance of sensors, loading metal oxides with catalytic metals or noble metals nanoparticles on the surface has proved to be a very effective method [21].For example, Kang et al. [22] reported that 2.2 wt % Cr added to CuO nanostructures by a solvothermal reaction significantly increased the responses to 100 ppm NO 2 from 7.5 to 134.2.Kim et al. [23] prepared Pd functionalized one-dimensional CuO nanostructures, which revealed a high response of 167 at 50 ppm.However, the operating temperature was relatively high (300 • C).Therefore, there is still a great demand to develop a H 2 S gas sensor with excellent sensing performance, such as fast response/recovery time, good selectivity, high sensitivity, low detection limit, and relatively low operating temperature [24].
Pt as a catalytic additive has been proven to be effective in improving gas-sensing performance.Ivanov et al. [25] has shown that the Pt-doped SnO 2 material is less resistant, more sensitive, and shows faster response to ethanol than pure SnO 2 ; what's more, the response of the Pt-doped sensors was between 2 and 55 times higher than the commercial one.Wang et al. [26] prepared α-Fe 2 O 3 nanoparticles doped with different amounts of Pt.Compared with the pure α-Fe 2 O 3 based sensor, the Pt-doped α-Fe 2 O 3 based sensors showed the enhanced excellent sensing performance and good selectivity for H 2 S gas.However, reports on the gas-sensing performance of Pt-doped CuO are rather rare.We expected that Pt doping could be an efficient way to enhance the sensing properties of pure CuO.In the present work, the pristine and Pt-doped CuO based H 2 S gas sensors were designed.The Pt-doped CuO has been prepared using a facile one step method.For the purpose of comparison, the gas sensing performance of CuO sensors with different Pt concentrations was evaluated.The effect of Pt doping plays an important role in enhancing the sensing performance, and the resulting gas sensors are able to operate at a low temperature.Long recovery time is a common issue for the metal oxide semiconductor sensors at low operating temperature [27], and we found that applying a short electric current pulse is an appropriate way to reduce the recovery time.

Experimental
All chemicals used in this work were of analytical grade and were used without further purification, and were purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China).

Preparation of CuO
CuO nanoflowers were synthesized by a simple water bath heating method.The typical process is listed here: 24.16 g copper (II) nitrate hydrate was dissolved in 1000 mL deionized water to prepare copper nitrate solution (0.1 mol/L), and then 12.62 g hexamethylenetra-mine was then added into the above solution (20 mL), and stirred for 1 h to get a bright blue solution.After that, the flask was moved to a water bath container at 80 • C for 90 min, and the bright blue Cu-hydroxide precipitate converted into dark brown ones (copper oxide).Finally, the resulting mixture was allowed to cool to room temperature, and was subsequently subjected to centrifugation at 8000 rpm for 10 min.The samples were rinsed with deionized water three times, followed by three times with ethanol, and were dried in the oven at 80 • C.

Pt-Doped CuO Flowers
Pt-doped CuO flowers were prepared by the following procedure: Chloroplatinic acid hexahydrate (1 g) was dissolved in deionized water (100 mL) to prepare mother liquor, then, the above solution (12.95 mL) was diluted to 250 mL to prepare (hydro)chloroplatinic acid (1 mmol/L).After that, a copper nitrate solution (20 mL) and (hydro)chloroplatinic acid were mixed together, and the remaining steps were the same as mentioned above.For a convenient comparison, we indicate doping with different concentrations of Pt by 0.75Pt-CuO, 1.25Pt-CuO, and 1.50Pt-CuO, respectively, which means the Pt doping concentration in the copper nitrate solution were 0.75 wt %, 1.25 wt %, and 1.50 wt %, respectively.

Characterization
The crystallinity and phase composition of the CuO specimens were analyzed by X-ray diffraction with a monochromatized Cu target radiation resource (λ = 1.5418Å), scanning from 10-80 • (XRD, D8-Advance, Bruker, Germany), as well as X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Thermo Scientific, London, UK).The morphology of the powders was investigated by field emission scanning electron microscopy (FE-SEM, S-4800, Hitachi Co. Ltd., Tokyo, Japan) at an accelerating voltage of 10 kV.Energy Dispersive X-Ray (EDX) spectroscopy was used to confirm the composition of the CuO nanostructures.The surface areas were measured by the Brunauer-Emmett-Teller method (BET, Tristar II 3020, Micromeritics Instrument Co Ltd., Norcross, GA, USA).The concentrations of Pt-doped specimens were determined by inductively coupled plasma atomic emission spectrometer (ICP-AES, A-6300, Thermo Scientific, Waltham, MA, USA).

Fabrication and Characterization of Gas Sensors
Gas sensors based on bare CuO and Pt-doped CuO products were fabricated.The gas sensors were fabricated as follows: The as-synthesized Pt-doped CuO flowers were uniformly mixed with deionized water to prepare a paste, and were then coated onto a ceramic tube with a pair of gold electrodes at both ends to form a thick film.The ceramic tubes were heated and dried under infrared radiant heating.Figure 1a shows a schematic structure of the ceramic electrode.The sensing element was calcined at 500 • C for 2 h to remove the bound water and to form a porous structure on the surface for gas sensing.After cooling down, a heating wire was placed inside the ceramic tube to make sure the sensors can work at different operating temperatures.Finally, the sensors were aged at 5.0 V for 7 days, in order to stabilize it.The photograph of the gas sensor is shown in Figure 1b.

Characterization
The crystallinity and phase composition of the CuO specimens were analyzed by X-ray diffraction with a monochromatized Cu target radiation resource (λ = 1.5418Å), scanning from 10°-80° (XRD, D8-Advance, Bruker, Germany), as well as X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Thermo Scientific, London, UK).The morphology of the powders was investigated by field emission scanning electron microscopy (FE-SEM, S-4800, Hitachi Co. Ltd., Tokyo, Japan) at an accelerating voltage of 10 kV.Energy Dispersive X-Ray (EDX) spectroscopy was used to confirm the composition of the CuO nanostructures.The surface areas were measured by the Brunauer-Emmett-Teller method (BET, Tristar II 3020, Micromeritics Instrument Co Ltd., Norcross, GA, USA).The concentrations of Pt-doped specimens were determined by inductively coupled plasma atomic emission spectrometer (ICP-AES, A-6300, Thermo Scientific, Waltham, MA, USA).

Fabrication and Characterization of Gas Sensors
Gas sensors based on bare CuO and Pt-doped CuO products were fabricated.The gas sensors were fabricated as follows: The as-synthesized Pt-doped CuO flowers were uniformly mixed with deionized water to prepare a paste, and were then coated onto a ceramic tube with a pair of gold electrodes at both ends to form a thick film.The ceramic tubes were heated and dried under infrared radiant heating.Figure 1a shows a schematic structure of the ceramic electrode.The sensing element was calcined at 500 °C for 2 h to remove the bound water and to form a porous structure on the surface for gas sensing.After cooling down, a heating wire was placed inside the ceramic tube to make sure the sensors can work at different operating temperatures.Finally, the sensors were aged at 5.0 V for 7 days, in order to stabilize it.The photograph of the gas sensor is shown in Figure 1b.
The WS-30A Gas Sensing Analysis System (Weisheng Tech Co, Ltd., Zhengzhou, China) was used to measure the performance of the gas sensor, and the static testing method was applied to examine the gas sensing performance.The gas sensors were placed into a closed test chamber (15 L in volume), and the target gases were injected into the test chamber through a rubber plug.The electrical properties changed once the sensor was exposed to the target gas.When the response reached a constant value, the sensors were exposed to air again, and the response returned to its original value.The response was defined as the ratio of Ra to Rg or Rg to Ra, where Ra and Rg are the resistances in the air and target gas, respectively.The response and recovery time were defined as the time needed for the sensor to reach 90% of the total change of the resistance during adsorption and desorption, respectively.The WS-30A Gas Sensing Analysis System (Weisheng Tech Co, Ltd., Zhengzhou, China) was used to measure the performance of the gas sensor, and the static testing method was applied to examine the gas sensing performance.The gas sensors were placed into a closed test chamber (15 L in volume), and the target gases were injected into the test chamber through a rubber plug.The electrical properties changed once the sensor was exposed to the target gas.When the response reached a constant value, the sensors were exposed to air again, and the response returned to its original value.The response was defined as the ratio of R a to R g or R g to R a , where R a and R g are the resistances in the air and target gas, respectively.The response and recovery time were defined as the time needed for the sensor to reach 90% of the total change of the resistance during adsorption and desorption, respectively.2a.However, the difference can be found at the specific region (35-40 • ) of XRD patterns in Figure 2b.After Cu 2+ ions had been replaced by Pt 2+ ions in the CuO matrix, the lattice parameter of CuO crystal is changed because of the different ionic radius between Cu 2+ (0.73 Å) [28] and Pt 2+ (0.86 Å) [29].As a result, with the increase of the Pt dopant from 0.75 wt % to 1.50 wt %, the (−111) and (111) peaks were slightly shifted to a lower angle [30].What is more, we used the least square refinement method in the unit cell program to determine the lattice constants.2a.However, the difference can be found at the specific region (35-40°) of XRD patterns in Figure 2b.After Cu 2+ ions had been replaced by Pt 2+ ions in the CuO matrix, the lattice parameter of CuO crystal is changed because of the different ionic radius between Cu 2+ (0.73 Å) [28] and Pt 2+ (0.86 Å) [29].As a result, with the increase of the Pt dopant from 0.75 wt % to 1.50 wt %, the (−111) and ( 111) peaks were slightly shifted to a lower angle [30].What is more, we used the least square refinement method in the unit cell program to determine the lattice constants.Field emission scanning electron microscope (FESEM) images of pristine CuO and Pt-doped CuO with various doping concentrations are shown in the Figure 3.As shown in Figure 3a, the morphology of the pristine CuO was olive-like, exhibiting particle diameters ranging from 200 nm to 4 μm.Once the Pt 2+ ions were introduced, the morphology of the CuO structure transformed into being flower-like, as illustrated in Figure 3b,c.At a Pt doping concentration of 1.25 wt % (Figure 3c), a three-dimensional (3D) flower-like CuO with hierarchical nanostructures was obtained.The flowerlike CuO is ca.1.5 μm in diameter and homogeneously distributed, and the hierarchical flower-like CuO consists of nanosheets with sizes around 200 nm.The resulting growth-mechanism for CuO after Pt doping can be explained as follows: A copper nitrate solution provides Cu 2+ ions to form a blue precipitate of copper hydroxide Cu(OH)2.The color of the Cu(OH)2 precipitate gradually turned into dark brown, which was caused by the dehydration of Cu(OH)2 and finally formed the nano-CuO along with the increase of the water bath temperature.Once the Pt was introduced, it did not affect the precipitate color, but seemed to have a consecutive influence on the growth rate of the nanoflower CuO [31].The Pt ions tend to be adsorbed on the basal planes of the CuO, and hence modify the chemical surface properties of CuO, such as the relative surface free energy on the facets.It was also observed that the presence of higher amounts of Pt influenced the dehydration of Cu(OH)2 and Field emission scanning electron microscope (FESEM) images of pristine CuO and Pt-doped CuO with various doping concentrations are shown in the Figure 3.As shown in Figure 3a, the morphology of the pristine CuO was olive-like, exhibiting particle diameters ranging from 200 nm to 4 µm.Once the Pt 2+ ions were introduced, the morphology of the CuO structure transformed into being flower-like, as illustrated in Figure 3b,c.At a Pt doping concentration of 1.25 wt % (Figure 3c), a three-dimensional (3D) flower-like CuO with hierarchical nanostructures was obtained.The flower-like CuO is ca.1.5 µm in diameter and homogeneously distributed, and the hierarchical flower-like CuO consists of nanosheets with sizes around 200 nm.The resulting growth-mechanism for CuO after Pt doping can be explained as follows: A copper nitrate solution provides Cu 2+ ions to form a blue precipitate of copper hydroxide Cu(OH) 2 .The color of the Cu(OH) 2 precipitate gradually turned into dark brown, which was caused by the dehydration of Cu(OH) 2 and finally formed the nano-CuO along with the increase of the water bath temperature.Once the Pt was introduced, it did not affect the precipitate color, but seemed to have a consecutive influence on the growth rate of the nanoflower CuO [31].The Pt ions tend to be adsorbed on the basal planes of the CuO, and hence modify the chemical surface properties of CuO, such as the relative surface free energy on the facets.It was also observed that the presence of higher amounts of Pt influenced the dehydration of Cu(OH) 2 and retarded the phase transformation of the Cu(OH) 2 to CuO.In addition, the morpholoies of the CuO were not significantly altered after increasing the Pt doping concentration to 1.50 wt %, instead, the uniformity of the CuO declined again (Figure 3d).Thus, it was clearly demonstrated that the morphology of CuO can be adjusted by the Pt doping.

Structure and Morphology of Sensing Materials
Appl.Sci.2018, 8, x 5 of 14 retarded the phase transformation of the Cu(OH)2 to CuO.In addition, the morpholoies of the CuO were not significantly altered after increasing the Pt doping concentration to 1.50 wt %, instead, the uniformity of the CuO declined again (Figure 3d).Thus, it was clearly demonstrated that the morphology of CuO can be adjusted by the Pt doping.The EDX element mapping was employed to evaluate the uniformity of element distribution in a 1.25Pt-CuO specimen.Figure 4 shows the distribution of the O, Cu, and Pt, respectively, and it is clearly demonstrated that all three elements are homogeneously distributed.Meanwhile, the concentrations of the 1.00Pt-CuO, 1.25Pt-CuO, and 1.50Pt-CuO specimens were also determined by inductively coupled plasma mass spectroscopy, the Pt concentrations were 0.76 at%, 0.85 at%, and 0.91 at%, respectively.The XPS spectral measurements were used to analyze the chemical states of the 1.25Pt-CuO sample.The full spectra of 1.25Pt-CuO are shown in Figure 5a, which proved the existence of Cu, Pt, and O.The appearance of the C 1s peaks can be ascribed to the adsorbed or bonded organic species [32].In Figure 5b, two peaks at 933.4 eV and 953.5c), the spectrum consists of two pairs of peaks at 74.7 and 76.5 eV and at 75.6 and 77.9 eV, which are assigned to Pt 0 and Pt 2+ , respectively [34].The above measurements clearly proved that the Pt 2+ ions were successfully doped into the CuO matrix, which is in good accordance with the results of EDX and ICP-AES.Figure 5d shows the high-resolution O 1s XPS profile of Pt-doped CuO, in which the two peaks at binding energies of 529.4 eV and 531.1 eV correspond to the typical surface lattice oxygen and the surface absorbed oxygen species (O2 − ads and O − ads) [35].The surface absorbed oxygen species play a positive role in the enhancement of gas sensing performance, as they are capable of reacting with test gas molecules [36].The EDX element mapping was employed to evaluate the uniformity of element distribution in a 1.25Pt-CuO specimen.Figure 4 shows the distribution of the O, Cu, and Pt, respectively, and it is clearly demonstrated that all three elements are homogeneously distributed.Meanwhile, the concentrations of the 1.00Pt-CuO, 1.25Pt-CuO, and 1.50Pt-CuO specimens were also determined by inductively coupled plasma mass spectroscopy, the Pt concentrations were 0.76 at%, 0.85 at%, and 0.91 at%, respectively.
The XPS spectral measurements were used to analyze the chemical states of the 1.25Pt-CuO sample.The full spectra of 1.25Pt-CuO are shown in Figure 5a, which proved the existence of Cu, Pt, and O.The appearance of the C 1s peaks can be ascribed to the adsorbed or bonded organic species [32].In Figure 5b, two peaks at 933.4 eV and 953.The BET surface areas of the pure CuO and 1.25Pt-CuO were investigated via a BET method using N 2 adsorption and desorption.As shown in Figure 6, it has been shown that the BET surface area of the 1.25Pt-CuO is 24.3 m 2 /g, which is 1.9 times larger than pure CuO (13.1 m 2 /g).However, none of such two samples showed a porous structure based on the BET results.The larger surface area of the 1.25Pt-CuO can be beneficial for the enhancement of gas sensing properties.
Appl.Sci.2018, 8, x 7 of 14 The BET surface areas of the pure CuO and 1.25Pt-CuO were investigated via a BET method using N2 adsorption and desorption.As shown in Figure 6, it has been shown that the BET surface area of the 1.25Pt-CuO is 24.3 m 2 /g, which is 1.9 times larger than pure CuO (13.1 m 2 /g).However, none of such two samples showed a porous structure based on the BET results.The larger surface area of the 1.25Pt-CuO can be beneficial for the enhancement of gas sensing properties.

Gas Sensing Characteristics
To investigate the optimum operating temperature and additive amount of the Pt-doped CuO based sensor for H2S, we examined the responses of the four sensors to 10 ppm H2S gas from room temperature to 260 °C. Figure 7a illustrates that the response of all the doped sensors increased once the operating temperature changed from room temperatures to 40 °C, while the undoped sensor has its maximum response at 70 °C.Furthermore, all the Pt-doped sensors exhibit a much higher response than the pristine one.The 1.25Pt-CuO sensor showed the highest response of 135.1 to 10 ppm of H2S gas at 40 °C, which was 13.1 times higher than that of the pure CuO sensor (Rg/Ra = 10.3) at the same temperature.At a high operating temperature (above 240 °C), the pure CuO and all Pt-doped sensors showed a small response (1.6-2.7).Therefore, the optimized doping concentration is 1.25 wt % Pt for CuO and the operating temperature is 40 °C.Figure 7b shows the dynamic response curves of the gas sensors, based on the pure CuO and Pt-doped CuO to 10 ppm H2S at 40 °C.It proves that the 1.25Pt-CuO gas sensor has the highest response.However, it is noteworthy that neither the pure CuO sensor nor Pt-doped CuO sensor could be completely recovered to the baseline.

Gas Sensing Characteristics
To investigate the optimum operating temperature and additive amount of the Pt-doped CuO based sensor for H 2 S, we examined the responses of the four sensors to 10 ppm H 2 S gas from room temperature to 260 • C. Figure 7a illustrates that the response of all the doped sensors increased once the operating temperature changed from room temperatures to 40 • C, while the undoped sensor has its maximum response at 70 • C. Furthermore, all the Pt-doped sensors exhibit a much higher response than the pristine one.The 1.25Pt-CuO sensor showed the highest response of 135.1 to 10 ppm of H 2 S gas at 40 • C, which was 13.1 times higher than that of the pure CuO sensor (R g /R a = 10.3) at the same temperature.At a high operating temperature (above 240 • C), the pure CuO and all Pt-doped sensors showed a small response (1.6-2.7).Therefore, the optimized doping concentration is 1.25 wt % Pt for CuO and the operating temperature is 40 • C. Figure 7b shows the dynamic response curves of the gas sensors, based on the pure CuO and Pt-doped CuO to 10 ppm H 2 S at 40 • C. It proves that the 1.25Pt-CuO gas sensor has the highest response.However, it is noteworthy that neither the pure CuO sensor nor Pt-doped CuO sensor could be completely recovered to the baseline.
Figure 8 shows the sensing behavior of the sensor, based on 1.25 wt % Pt-doped CuO towards 10 ppm H 2 S gas at 40 • C. As it is shown in Figure 8a, the resistance of the 1.25Pt-CuO sensor is 0.1 kΩ when exposed to air.After the sensor was exposed to H 2 S gas, the response resistance increases to 13.51 kΩ, thus the response of the 1.25Pt-CuO sensor to 10 ppm H 2 S is as high as 135.1 (R g/ R a ).Although the Pt-doped CuO sensors were highly sensitive to detect H 2 S gas, the extremely long recovery time was the bottleneck for real practice.Figure 8b shows the dynamic response-recovery curves of the sensor; it took 37 s to reach 90% of the resistance variation.Figure 8b also shows that it took more than 3000 s to recover about 5% of the maximum value after removal of the H 2 S. In fact, the response could not be completely recovered to the baseline, and it is the common issue for gas sensors operating at a low temperature.According to Chaturvedi et al. [37] because of molecular affinity, Cu and S could easily form a helical bond at a low operating temperature.Therefore, H 2 S could be present in the form of molecular adsorption on the surface, and the adsorption is intermediate between physisorption and chemisorption.
temperature.At a high operating temperature (above 240 °C), the pure CuO and all Pt-doped sensors showed a small response (1.6-2.7).Therefore, the optimized doping concentration is 1.25 wt % Pt for CuO and the operating temperature is 40 °C.Figure 7b shows the dynamic response curves of the gas sensors, based on the pure CuO and Pt-doped CuO to 10 ppm H2S at 40 °C.It proves that the 1.25Pt-CuO gas sensor has the highest response.However, it is noteworthy that neither the pure CuO sensor nor Pt-doped CuO sensor could be completely recovered to the baseline.Figure 8 shows the sensing behavior of the sensor, based on 1.25 wt % Pt-doped CuO towards 10 ppm H2S gas at 40 °C.As it is shown in Figure 8a, the resistance of the 1.25Pt-CuO sensor is 0.1 kΩ when exposed to air.After the sensor was exposed to H2S gas, the response resistance increases to 13.51 kΩ, thus the response of the 1.25Pt-CuO sensor to 10 ppm H2S is as high as 135.1 (Rg/Ra).Although the Pt-doped CuO sensors were highly sensitive to detect H2S gas, the extremely long recovery time was the bottleneck for real practice.Figure 8b shows the dynamic response-recovery curves of the sensor; it took 37 s to reach 90% of the resistance variation.Figure 8b also shows that it took more than 3000 s to recover about 5% of the maximum value after removal of the H2S.In fact, the response could not be completely recovered to the baseline, and it is the common issue for gas sensors operating at a low temperature.According to Chaturvedi et al. [37] because of molecular affinity, Cu and S could easily form a helical bond at a low operating temperature.Therefore, H2S could be present in the form of molecular adsorption on the surface, and the adsorption is intermediate between physisorption and chemisorption.In this work, we proposed a solution to solve the problem that the gas sensors could not be completely recovered at a low operating temperature.The principle of this scheme is based on the Joule heating effect, which accelerates the detachment of the H2S molecules from the CuO surface by applying a short electric current pulse.As shown in Figure 9a, Vc is the test voltage; Rl is a load resistor, and the voltage drop on Rl is denoted as Vout; Vh is the heating voltage applied on the the heating wire; and Vm is the 4.6 V electric current pulse.Figure 9b shows the dynamic response curves of the 1.25Pt-CuO sensor to 1 ppm H2S gas at 40 °C.Once the H2S was removed and the sensor was exposed to air, the response could not be recovered to its original baseline.After applying a 4.6 V (the temperature was 300 °C) voltage pulse, the 1.25Pt-CuO sensor returned to its original state within 15 s.Reproducibility is an important indicator to evaluate the performance of the sensor.Figure 9b also shows the dynamic response curves of the 1.25Pt-CuO sensor to 10 ppm H2S gas.In the five-cycle continuous test process, the sensors based on the Pt-doped CuO exhibit good stability, and their gassensing performance, such as sensitivity, and response/recovery speed did not show obvious resistance attenuation.In addition, such a 'recovery heating' process had tiny effect on the response of the sensor, which made the gas sensor highly reusable.Therefore, by applying a short electric current pulse, we have developed high performance H2S gas sensors based on the Pt-doped CuO with a low operating temperature.In this work, we proposed a solution to solve the problem that the gas sensors could not be completely recovered at a low operating temperature.The principle of this scheme is based on the Joule heating effect, which accelerates the detachment of the H 2 S molecules from the CuO surface by applying a short electric current pulse.As shown in Figure 9a, V c is the test voltage; R l is a load resistor, and the voltage drop on R l is denoted as V out ; V h is the heating voltage applied on the the heating wire; and V m is the 4.6 V electric current pulse.Figure 9b shows the dynamic response curves of the 1.25Pt-CuO sensor to 1 ppm H 2 S gas at 40 • C. Once the H 2 S was removed and the sensor was exposed to air, the response could not be recovered to its original baseline.After applying a 4.6 V (the temperature was 300 • C) voltage pulse, the 1.25Pt-CuO sensor returned to its original state within 15 s.Reproducibility is an important indicator to evaluate the performance of the sensor.Figure 9b also shows the dynamic response curves of the 1.25Pt-CuO sensor to 10 ppm H 2 S gas.In the five-cycle continuous test process, the sensors based on the Pt-doped CuO exhibit good stability, and their gas-sensing performance, such as sensitivity, and response/recovery speed did not show obvious resistance attenuation.In addition, such a 'recovery heating' process had tiny effect on the response of the sensor, which made the gas sensor highly reusable.Therefore, by applying a short electric current pulse, we have developed high performance H 2 S gas sensors based on the Pt-doped CuO with a low operating temperature.Selectivity is one of the key parameters of the gas sensors.To explore the selectivity of the sensors, the dynamic response curves of the 1.25Pt-CuO based sensor to 1 ppm H2S and 50 ppm of other gases were measured at 40 °C.The results are shown in Figure 10a.Compared with the responses to H2S gas, the responses to other reducing gases were negligible.Figure 10b shows the responses of the 1.25Pt-CuO sensor to various gas samples, including 50 ppm C2H5OH, NH3, CH3COCH3, NO2, Cl2, H2, NO, and 1 ppmH2S, operated at 40 °C.The sensitivity of such sensors are 32.9 and 2.4 to 1 ppm H2S and 50 ppm NO2, respectively, while the responses to other gases ranged between 1.1 and 1.2.Therefore, 1.25 wt % Pt doping led to highest selectivity and sensitivity for the detection of H2S at low operating temperature (i.e., 40 °C). Figure 11a displays the dynamic response-recovery transient of the sensor based on 1.25Pt-CuO, under the H2S concentration ranging from 0.1 to 25 ppm, at the operating temperature of 40 °C.It is apparent that as the concentration of the H2S gas increases, the response curve also displays a stepwise distribution.Figure 11b shows the response of the gas sensor to different concentrations of H2S gas.It can be observed that the gas response increased as the concentration increased.Even when the H2S concentration is as low as 300 ppb, the response of the sensor is still 3.2.According to the value of response (Rg/Ra > 2), as the standard for a valid response [38], it can be said that the gas sensor based on 1.25Pt-CuO has a detection limit of H2S gas as low as 300 ppb. Figure 12 illustrates  Selectivity is one of the key parameters of the gas sensors.To explore the selectivity of the sensors, the dynamic response curves of the 1.25Pt-CuO based sensor to 1 ppm H2S and 50 ppm of other gases were measured at 40 °C.The results are shown in Figure 10a.Compared with the responses to H2S gas, the responses to other reducing gases were negligible.Figure 10b shows the responses of the 1.25Pt-CuO sensor to various gas samples, including 50 ppm C2H5OH, NH3, CH3COCH3, NO2, Cl2, H2, NO, and 1 ppmH2S, operated at 40 °C.The sensitivity of such sensors are 32.9 and 2.4 to 1 ppm H2S and 50 ppm NO2, respectively, while the responses to other gases ranged between 1.1 and 1.2.Therefore, 1.25 wt % Pt doping led to highest selectivity and sensitivity for the detection of H2S at low operating temperature (i.e., 40 °C). Figure 11a displays the dynamic response-recovery transient of the sensor based on 1.25Pt-CuO, under the H2S concentration ranging from 0.1 to 25 ppm, at the operating temperature of 40 °C.It is apparent that as the concentration of the H2S gas increases, the response curve also displays a stepwise distribution.Figure 11b shows the response of the gas sensor to different concentrations of H2S gas.It can be observed that the gas response increased as the concentration increased.Even when the H2S concentration is as low as 300 ppb, the response of the sensor is still 3.2.According to the value of response (Rg/Ra > 2), as the standard for a valid response [38], it can be said that the gas sensor based on 1.25Pt-CuO has a detection limit of H2S gas as low as 300 ppb. Figure 12 illustrates  It can be observed that the gas response increased as the concentration increased.Even when the H 2 S concentration is as low as 300 ppb, the response of the sensor is still 3.2.According to the value of response (R g /R a > 2), as the standard for a valid response [38], it can be said that the gas sensor based on 1.25Pt-CuO has a detection limit of H 2 S gas as low as 300 ppb. Figure 12 illustrates the long-term stability of the gas sensor based on 1.25Pt-CuO.The results reveal that the response of the sensor changes by less than 10%, indicating that our sensor is stable and reliable.
Appl.Sci.2018, 8, x 10 of 14 the long-term stability of the gas sensor based on 1.25Pt-CuO.The results reveal that the response of the sensor changes by less than 10%, indicating that our sensor is stable and reliable.

The Sensing Mechanism
The gas-sensing mechanism of the p-type semiconducting oxides has been previously reported, and it can be summarized as follows: Adsorption/desorption of gas molecules on the surface of semiconductors and reactions cause the change of resistance [11,24].When the p-type CuO semiconductor is in contact with air, oxygen molecules are adsorbed on the surface of CuO and are ionized by electrons from the valence band, thereby generating negatively charged chemisorbed oxygen species (O 2− , O − , and O2 − ), the processes are shown in Equations ( 1)-(4).The types of absorbed oxygen species depend on the operating temperature.Oxygen is absorbed on the surface as O2 − ions below 147 °C or as O − ions between 147-397 °C.Above 397 °C, O 2− ions occur [39].According to the above conclusions, the major oxygen adsorption species on the surface of CuO is O2 − when the operating temperature is 40 °C.O2 (gas) → O2 (ads) (1)

The Sensing Mechanism
The gas-sensing mechanism of the p-type semiconducting oxides has been previously reported, and it can be summarized as follows: Adsorption/desorption of gas molecules on the surface of semiconductors and reactions cause the change of resistance [11,24].When the p-type CuO semiconductor is in contact with air, oxygen molecules are adsorbed on the surface of CuO and are ionized by electrons from the valence band, thereby generating negatively charged chemisorbed oxygen species (O 2− , O − , and O2 − ), the processes are shown in Equations ( 1)-(4).The types of absorbed oxygen species depend on the operating temperature.Oxygen is absorbed on the surface as O2 − ions below 147 °C or as O − ions between 147-397 °C.Above 397 °C, O 2− ions occur [39].According to the above conclusions, the major oxygen adsorption species on the surface of CuO is O2 − when the operating temperature is 40 °C.O2 (gas) → O2 (ads) (1)

The Sensing Mechanism
The gas-sensing mechanism of the p-type semiconducting oxides has been previously reported, and it can be summarized as follows: Adsorption/desorption of gas molecules on the surface of semiconductors and reactions cause the change of resistance [11,24].When the p-type CuO semiconductor is in contact with air, oxygen molecules are adsorbed on the surface of Once the sensor is exposed to an H 2 S (reducing gas) atmosphere operating at 40 • C, the H 2 S molecules are oxidized by O 2− to form SO 2 and H 2 O, while releasing electrons [40], and the sensing reaction is as shown in Equation ( 5).
The released electrons during this process will recombine with the holes on the surface of CuO and cause the decrease of hole concentration, which will lead to an increased resistance of the gas sensor.According to some previous reports [41,42], a CuS layer might form on the surface of the CuO, when the sensor is exposed to high H 2 S concentrations, the spontaneous chemical reaction is as shown in Equation ( 6), as follows: In order to confirm the formation of CuS, we have carried out XPS studies on a 10 ppm H 2 S exposed 1.25Pt-CuO sensor.The recorded Cu 2p and S 2p spectra are shown in Figure 13.When the sensor was exposed to H 2 S gas, the Cu 2p spectrum confirmed the presence of CuO and CuS, and the corresponding peak of 169.2 eV in the S 2p spectrum also confirmed the presence of CuS [11].
Once the sensor is exposed to an H2S (reducing gas) atmosphere operating at 40 °C, the H2S molecules are oxidized by O2− to form SO2 and H2O, while releasing electrons [40], and the sensing reaction is as shown in Equation (5).H2S (g) + 3O2 − (ads) → H2O (g) + SO2 (g) + 6e − (5) The released electrons during this process will recombine with the holes on the surface of CuO and cause the decrease of hole concentration, which will lead to an increased resistance of the gas sensor.According to some previous reports [41,42], a CuS layer might form on the surface of the CuO, when the sensor is exposed to high H2S concentrations, the spontaneous chemical reaction is as shown in Equation ( 6), as follows: CuO (s) + H2S (g) → CuS (s) + H2O (g) (6) In order to confirm the formation of CuS, we have carried out XPS studies on a 10 ppm H2S exposed 1.25Pt-CuO sensor.The recorded Cu 2p and S 2p spectra are shown in Figure 13.When the sensor was exposed to H2S gas, the Cu 2p spectrum confirmed the presence of CuO and CuS, and the corresponding peak of 169.2 eV in the S 2p spectrum also confirmed the presence of CuS [11].Once H2S is removed by introducing air, the CuS is unstable, and is able to be slowly oxidized by oxygen in the air, as shown in Equation (7).As a result, recovery of the resistance will occur.CuS (s) + O2 (g) → 2CuO (s) + 2S (s) (7) The influence of Pt dopant on the gas sensing properties of the sensor can be attributed to the spillover effect [43].Various chemisorbed oxygen species and gas molecules were able to spill over onto the surface CuO support, as the O2 and H2S gas molecules are absorbed in the abundant sites provided by the Pt dopant.Thus, it is reasonable to believe that the better gas sensing properties of the Pt-doped CuO is mainly caused by the ability of the Pt dopant to accelerate the electron exchange between the test gas and the sensor, rather than being directly involved in the reaction.On the other hand, compared with pure CuO, 1.25Pt-doped CuO nanoflowers have a larger BET surface area, which can provide more adsorption sites for gas adsorption.However, as the doping concentration increased to 1.50 wt %, a constipation disorder occurs and the surface density increases, resulting in Once H 2 S is removed by introducing air, the CuS is unstable, and is able to be slowly oxidized by oxygen in the air, as shown in Equation (7).As a result, recovery of the resistance will occur.CuS (s) + O 2 (g) → 2CuO (s) + 2S (s) (7) The influence of Pt dopant on the gas sensing properties of the sensor can be attributed to the spillover effect [43].Various chemisorbed oxygen species and gas molecules were able to spill over onto the surface CuO support, as the O 2 and H 2 S gas molecules are absorbed in the abundant sites provided by the Pt dopant.Thus, it is reasonable to believe that the better gas sensing properties of the Pt-doped CuO is mainly caused by the ability of the Pt dopant to accelerate the electron exchange between the test gas and the sensor, rather than being directly involved in the reaction.On the other hand, compared with pure CuO, 1.25Pt-doped CuO nanoflowers have a larger BET surface area, which can provide more adsorption sites for gas adsorption.However, as the doping concentration increased to 1.50 wt %, a constipation disorder occurs and the surface density increases, resulting in Fermi level changes and the decrease of sensor response [44].In summary, the doping of Pt is the main reason for the enhanced performance of CuO-based sensors.

Conclusions
H 2 S sensors based on Pt-doped CuO nano-flowers were developed.The pristine and Pt-doped CuO nano-flowers were fabricated by a simple heating route using a water bath, and their gas-sensing performances toward H 2 S were thoroughly investigated.The Pt-doped CuO sensor exhibited a significant improvement in its H 2 S performance compared with the sensors based on pristine CuO.The enhanced sensing performances of Pt doped CuO would be most likely ascribed to the spillover effect.At an operating temperature of 40 • C, the optimized Pt-doped CuO sensor yielded a response of 135.1 and 32.9 upon exposure to H 2 S gas of 10 and 1 ppm, respectively.To solve the issue of a long recovery time at a low operating temperature, a 4.6 V (300 • C) heater voltage pulse was selected, and the recovery time could be shortened to 15 s.

Figure 1 .
Figure 1.(a) Schematic illustration of the gas sensor and (b) photographs of the gas sensor.Figure 1.(a) Schematic illustration of the gas sensor and (b) photographs of the gas sensor.

Figure 1 .
Figure 1.(a) Schematic illustration of the gas sensor and (b) photographs of the gas sensor.Figure 1.(a) Schematic illustration of the gas sensor and (b) photographs of the gas sensor.

Figure 2
Figure 2 plots the XRD patterns of pure CuO structures and Pt-doped CuO structures.The patterns of pure CuO are specified as monoclinic CuO (JCPDS No. 72-0629), and reveal a series of well-defined diffraction peaks at 2θ of 32.53 • , 35.56 • , 38.75 • , 48.75 • , 53.50 • , 58.36 • , 61.57• , 66.28 • , and 67.97 • , corresponding to the (110), (−111), (111), (−202), (020), (202), (−113), (−311), and (113) planes of monoclinic CuO.All XRD patterns look very similar at first sight, as shown in Figure 2a.However, the difference can be found at the specific region (35-40 • ) of XRD patterns in Figure2b.After Cu 2+ ions had been replaced by Pt 2+ ions in the CuO matrix, the lattice parameter of CuO crystal is changed because of the different ionic radius between Cu 2+ (0.73 Å)[28] and Pt 2+ (0.86 Å)[29].As a result, with the increase of the Pt dopant from 0.75 wt % to 1.50 wt %, the (−111) and (111) peaks were slightly shifted to a lower angle[30].What is more, we used the least square refinement method in the unit cell program to determine the lattice constants.The obtained parameters of the 1.25 wt % Pt-doped CuO are α = 4.6837 Å, b = 3.4226 Å, and c = 5.1288 Å, and the unit cell volume is 82.2169Å 3 .The lattice constants of the doped sample are the monoclinic data of undoped CuO (α = 4.6504 Å, b = 3.3903 Å, c = 5.0997 Å, and unit cell volume = 80.4032 Å 3 ) obtained under the same conditions.It can be seen that the unit cell expands with the Pt 2+ ion doping in CuO.This clearly proved that the Pt 2+ ions were successfully replaced by Cu 2+ in the CuO matrix.
Figure 2 plots the XRD patterns of pure CuO structures and Pt-doped CuO structures.The patterns of pure CuO are specified as monoclinic CuO (JCPDS No. 72-0629), and reveal a series of well-defined diffraction peaks at 2θ of 32.53 • , 35.56 • , 38.75 • , 48.75 • , 53.50 • , 58.36 • , 61.57• , 66.28 • , and 67.97 • , corresponding to the (110), (−111), (111), (−202), (020), (202), (−113), (−311), and (113) planes of monoclinic CuO.All XRD patterns look very similar at first sight, as shown in Figure 2a.However, the difference can be found at the specific region (35-40 • ) of XRD patterns in Figure2b.After Cu 2+ ions had been replaced by Pt 2+ ions in the CuO matrix, the lattice parameter of CuO crystal is changed because of the different ionic radius between Cu 2+ (0.73 Å)[28] and Pt 2+ (0.86 Å)[29].As a result, with the increase of the Pt dopant from 0.75 wt % to 1.50 wt %, the (−111) and (111) peaks were slightly shifted to a lower angle[30].What is more, we used the least square refinement method in the unit cell program to determine the lattice constants.The obtained parameters of the 1.25 wt % Pt-doped CuO are α = 4.6837 Å, b = 3.4226 Å, and c = 5.1288 Å, and the unit cell volume is 82.2169Å 3 .The lattice constants of the doped sample are the monoclinic data of undoped CuO (α = 4.6504 Å, b = 3.3903 Å, c = 5.0997 Å, and unit cell volume = 80.4032 Å 3 ) obtained under the same conditions.It can be seen that the unit cell expands with the Pt 2+ ion doping in CuO.This clearly proved that the Pt 2+ ions were successfully replaced by Cu 2+ in the CuO matrix.
Figure 2 plots the XRD patterns of pure CuO structures and Pt-doped CuO structures.The patterns of pure CuO are specified as monoclinic CuO (JCPDS No. 72-0629), and reveal a series of well-defined diffraction peaks at 2θ of 32.53°, 35.56°, 38.75°, 48.75°, 53.50°, 58.36°, 61.57°, 66.28°, and 67.97°, corresponding to the (110), (−111), (111), (−202), (020), (202), (−113), (−311), and (113) planes of monoclinic CuO.All XRD patterns look very similar at first sight, as shown in Figure2a.However, the difference can be found at the specific region (35-40°) of XRD patterns in Figure2b.After Cu 2+ ions had been replaced by Pt 2+ ions in the CuO matrix, the lattice parameter of CuO crystal is changed because of the different ionic radius between Cu 2+ (0.73 Å)[28] and Pt 2+ (0.86 Å)[29].As a result, with the increase of the Pt dopant from 0.75 wt % to 1.50 wt %, the (−111) and (111) peaks were slightly shifted to a lower angle[30].What is more, we used the least square refinement method in the unit cell program to determine the lattice constants.The obtained parameters of the 1.25 wt % Pt-doped CuO are α = 4.6837 Å, b = 3.4226 Å, and c = 5.1288 Å, and the unit cell volume is 82.2169Å 3 .The lattice constants of the doped sample are the monoclinic data of undoped CuO (α = 4.6504 Å, b = 3.3903 Å, c = 5.0997 Å, and unit cell volume = 80.4032 Å 3 ) obtained under the same conditions.It can be seen that the unit cell expands with the Pt 2+ ion doping in CuO.This clearly proved that the Pt 2+ ions were successfully replaced by Cu 2+ in the CuO matrix.
2 eV, corresponding to Cu 2p3/2 and Cu 2p1/2, denote the existence of Cu 2+ in the form of CuO.In addition, the satellite peaks of Cu 2p3/2 and Cu 2p1/2, denoted as peaks S1 and S2, respectively, are characteristic for a partially filled d-orbital (3 d9 in the case of Cu 2+ ) [33].The Pt 4f spectra of 1.25Pt-CuO are shown in Figure 5c.After the Pt 4f curve fitting (Figure

Figure 5 .
Figure 5. (a) X-ray photoelectron spectroscopy (XPS) spectrum of the 1.25 wt % Pt-doped CuO; (b-d) high-resolution spectra of Cu, Pt, and O peaks, respectively, (red line, blue line, purple line and green line are fitting curves).

Figure 5 .
Figure 5. (a) X-ray photoelectron spectroscopy (XPS) spectrum of the 1.25 wt % Pt-doped CuO; (b-d) high-resolution spectra of Cu, Pt, and O peaks, respectively, (red line, blue line, purple line and green line are fitting curves).

Figure 5 .
Figure 5. (a) X-ray photoelectron spectroscopy (XPS) spectrum of the 1.25 wt % Pt-doped CuO; (b-d) high-resolution spectra of Cu, Pt, and O peaks, respectively, (red line, blue line, purple line and green line are fitting curves).

Figure 8 .
Figure 8.(a) The resistance variation of the 1.25Pt-CuO sensor to 10 ppm H2S at 40 °C; (b) the corresponding sensing transients of the 1.25Pt-CuO sensor.The inset of (b) is the magnification of sensing transients between 300 s to 700 s.

Figure 8 .
Figure 8.(a) The resistance variation of the 1.25Pt-CuO sensor to 10 ppm H 2 S at 40 • C; (b) the corresponding sensing transients of the 1.25Pt-CuO sensor.The inset of (b) is the magnification of sensing transients between 300 s to 700 s.

Figure 9 .
Figure 9. (a) Sensing diagram of the gas sensor with a 4.6 V electric modulation voltage; (b) the recovery cycles for 10 ppm H2S of the 1.25Pt-CuO sensor at 40 °C, followed by a 4.6 V electric modulation voltage.

Figure 9 .
Figure 9. (a) Sensing diagram of the gas sensor with a 4.6 V electric modulation voltage; (b) the recovery cycles for 10 ppm H 2 S of the 1.25Pt-CuO sensor at 40 • C, followed by a 4.6 V electric modulation voltage.

Figure 9 .
Figure 9. (a) Sensing diagram of the gas sensor with a 4.6 V electric modulation voltage; (b) the recovery cycles for 10 ppm H2S of the 1.25Pt-CuO sensor at 40 °C, followed by a 4.6 V electric modulation voltage.

Figure 10 .
Figure 10.(a) Dynamic response curves of 1.25Pt-CuO and 0.00Pt-CuO based sensor to 1 ppm H 2 S and 50 ppm various interfering gases at 40 • C; (b) gas response (R a /R g or R g /R a ) of 1.25Pt-CuO to 50 ppm different gases and 1 ppm H 2 S gas, operated at 40 • C.

Figure
Figure 11a displays the dynamic response-recovery transient of the sensor based on 1.25Pt-CuO, under the H 2 S concentration ranging from 0.1 to 25 ppm, at the operating temperature of 40• C. It is apparent that as the concentration of the H 2 S gas increases, the response curve also displays a stepwise distribution.Figure11bshows the response of the gas sensor to different concentrations of H 2 S gas.

Figure 11 .
Figure 11.(a) Dynamic response transient of the gas sensor based on 1.25Pt-CuO and 1.00Pt-CuO to different concentrations of H2S gas at 40 °C; (b) gas response of 1.25Pt-CuO and 1.00Pt-CuO sensor to different concentrations of H2S at 40 °C.

Figure 11 .
Figure 11.(a) Dynamic response transient of the gas sensor based on 1.25Pt-CuO and 1.00Pt-CuO to different concentrations of H 2 S gas at 40 • C; (b) gas response of 1.25Pt-CuO and 1.00Pt-CuO sensor to concentrations of H 2 S at 40 • C.

Figure 11 .
Figure 11.(a) Dynamic response transient of the gas sensor based on 1.25Pt-CuO and 1.00Pt-CuO to different concentrations of H2S gas at 40 °C; (b) gas response of 1.25Pt-CuO and 1.00Pt-CuO sensor to different concentrations of H2S at 40 °C.