Electrospun ZnO–SnO2 heterojunction belts for hydrogen sensing

Chemiresistive sensors are promising devices for sensing hydrogen gas in a broad range of applications including fuel cells, hydrogen storage systems, petroleum refinement, and diagnosis of oil-insulated transformers. Herein, electrospun ZnO–SnO2 belts (BLs) were synthesized and applied as resistive-type sensing layers for hydrogen sensing. The ZnO–SnO2 BLs containing 20 mol% of Zn relative to Sn showed a response (Ra/Rg, Ra: resistance in air, Rg: resistance in target gas) of 6.7, fast response speed (3.6 s), and a distinguishable selectivity toward 5 ppm of hydrogen at 400 °C in the presence of formaldehyde, methane, ammonia, carbon monoxide, and carbon dioxide gases. The sensor displayed a repeatable response when subjected to 15 cycles of alternate air and 5 ppm hydrogen exposure. A unique hydrogen sensing performance of the BLs was attributed to their belt morphology, numerous surface pores, smaller crystal size, ZnO/SnO2 heterojunction, and ZnO metallization following hydrogen exposure. The present synthetic method paves the way for generating microstructures with smaller diffusion length that overcomes the shortcomings of non-porous and/or thick materials while providing a potential platform for reliable and enhanced hydrogen sensing.


Introduction
Hydrogen is an abundant and environmentally benign gas that is one of the most promising nextgeneration energy resources in various application areas, including industrial processing, fuel cells, petroleum refinement, and metallurgical processes. Although hydrogen can be produced in large amounts without creating hazardous byproducts during the energy generation process, its explosive nature and low flammable limit in air (& 4% by volume) limit its use [1,2]. Thus, fast, stable, and hydrogen-selective sensors are becoming essential to alert the formation of potentially explosive mixtures with air, hence preventing the risk of an explosion [3].
Many types of hydrogen (H 2 ) safety sensors based on electrochemical, chemiresistive, and combined principles are commercially available [4,5]. Apart from being susceptible to ambient conditions [6], electrochemical sensors are based on costly complicated systems that require regular calibration [7]. Chemiresistive H 2 sensors are simple in design and compatible with miniature electronic devices, and offer a wide range of operating temperature [8]. However, the need to achieve rapid response and recovery, stable operation, and low cross-sensitivity with interfering gases among commercial H 2 chemiresistive sensors has attracted the attention of many researchers in developing materials to achieve stringent performance targets [5,[9][10][11][12][13][14]. Palladium is the most used material for state-of-the-art chemiresistive H 2 sensing due to its high selectivity and room-temperature operation [13,[15][16][17]. Nevertheless, adsorption of O 2 and water vapor on palladium surface reduces active sites for adsorption of H 2 , leading to poor response/recovery characteristics and high limit of detection [13]. Recently, semiconducting metal oxides (SMOs) such as ZnO, SnO 2 , WO 3 , TiO 2 , In 2 O 3 and their composites have been synthesized and employed as alternative materials to palladium [18][19][20][21][22][23][24][25]. In particular, manipulating the structure, composition, morphology, and porosity of SMOs are critical for H 2 sensing [26,27].
Chemiresistive sensors based on ZnO-SnO 2 heterojunction exhibit enhanced sensitivity because of their ZnO/SnO 2 grain boundary [28,29], and synergistic effect ascribed to the strong interactions between the closely packed nano-units in the composite structure [30,31]. Since the early report of H 2 sensor based on ZnO/SnO 2 heterojunction by Huang et al. [32], there has been tremendous progress in the synthesis and H 2 [22]. Despite the strenuous effort to improve the hydrogen-sensing performance of ZnO-SnO 2 composites, little attention has been given to synthesizing heterostructures based on ZnO-SnO 2 BLs and their H 2 -sensing properties. The belt morphology could reduce the decay of Knudsen diffusion by providing a shorter diffusion length, improved reaction kinetics, and enhanced gas sensing performance [35]. Herein, ZnO-SnO 2 BLs containing n-n heterojunctions were synthesized via electrospinning and high-temperature calcination, and their H 2 sensing performance was evaluated. The H 2 -sensing mechanism of the BLs was also elucidated.

Synthesis of ZnO-SnO 2 BLs
The ZnO-SnO 2 BLs were prepared via electrospinning from a solution containing Zn(CH 3 COO) 2 Á2H 2 O, SnCl 2 Á2H 2 O, and polyvinylpyrrolidone, followed by heat treatment in air. 0.3 g of SnCl 2 Á2H 2 O and a varying amount of Zn(CH 3 COO) 2 Á2H 2 O (corresponding to 10, 20, and 30 mol% of Zn relative to Sn) were dissolved in 2.7 g of a cosolvent of N, Ndimethylformamide (DMF) and ethanol at a weight ratio of 1:1. After stirring the mixture for 1 h, 0.3 g of polyvinylpyrrolidone was added to the resulting solution. The mixture was continuously stirred at room temperature (350 rpm) for 1 h to achieve a homogeneous solution. The final solution was poured into a 12-mL plastic syringe for electrospinning. A DC voltage of 8 kV was applied between the stainless steel needle (25-gauge) and the collector, both separated by a gap of 17 cm. The electrospun BLs were collected as a non-woven fibrous mat on the collector. Afterward, the as-spun BLs were calcined at 650°C for 2 h in air at a ramping rate of 5°C/min from room temperature. For simplicity, the calcined BLs containing Zn at 10, 20, and 30 mol% were denoted as 10% ZnO-SnO 2 BLs, 20% ZnO-SnO 2 BLs, and 30% ZnO-SnO 2 BLs, respectively.

Material characterization
Characterization of the surface morphology was conducted by scanning electron microscopy (SEM) analysis (XL-30 SFEG, Philips). The microstructure of the calcined ZnO-SnO 2 BLs was investigated by transmission electron microscopy (TEM) [FETEM, Tecnai G2 F30 STwin, FEI] equipment, integrated with energy dispersive X-ray spectrometry (EDS) tool. The crystal structure of ZnO-SnO 2 BLs was characterized by X-ray diffraction XRD (D/MAX-2500 series, Rigaku) using Cu Ka (k = 1.5406 Å ) radiation. X-ray photoelectron spectroscopy (Sigma Probe, Thermo VG Scientific) analysis was performed to investigate the chemical binding states of the elements present in ZnO-SnO 2 BLs. The thermal decomposition of as-spun BLs during calcination was studied by thermal gravimetric (TG) and differential scanning calorimetry (DSC) analyzer (Thermo ONIX Gaslab 300) in a temperature range of 30-900°C at a heating rate of 5°C/min in air.

Gas-sensing measurements
The gas sensing characteristics were measured by homemade equipment as illustrated and described elsewhere [36]. Al 2 O 3 substrates with an area of 2.5 9 2.5 mm and thickness of 0.2 mm were used and consisted of a platinum (Pt) microheater on the backside, and two interdigitated parallel gold (Au) electrodes on the front side. The paste was prepared by dispersing 2.2 mg of the sensing materials in 60 lL of ethanol, followed by drop-coating 1 lL of the obtained paste to the substrates and drying on a hot plate for 6 min at 60°C. Before hydrogen exposure, all the sensors were stabilized in dry air at each operating temperature. Various analyte molecules (H 2 , HCHO, CH 4 , NH 3 , CO, and CO 2 ) were exposed to ZnO-SnO 2 BLs with concentrations ranging from 1 to 5 ppm at a 10 min gas on/off interval. The response (R a /R g ) of sensors was determined from the resistance of ZnO-SnO 2 BLs in air and target gas. The resistance changes were determined using a data acquisition system (34,972 A, Agilent). The operating temperature was controlled via Pt microheaters at the backside of the alumina substrate using a direct current power supply (E3647A, Agilent). The relative humidity (RH) was measured using a humidity sensor (605H1, Testo Inc.).

Morphology and structural characterization
The produced ZnO-SnO 2 heterostructures exhibited a belt-type morphology, as shown in the SEM image in Fig. 1a. The belt morphology resulted from the differences in solvent evaporation rates and higher affinity of additive polymers toward one of the cosolvent components [37,38]. The present work used a co-solvent comprising ethanol (fast evaporating) and DMF (slow evaporating) solvents. During electrospinning, Zn 2? and Sn 4? ions migrate with the evaporating co-solvent, and the migration intensifies with the stretching of the electrospun fibers by the applied electric field. Fast evaporation of ethanol relative to DMF creates a dry layer on the fiber surface, unstable to maintain the integrity of the cylindrical morphology. The formed ribbon-like structure gradually transforms into a belt when the entrapped DMF evaporates. SEM images shown in Fig. 1a (Fig. 1f). The lattice fringes of 0.282 nm in the HRTEM image matched well with the crystallographic (100) planes of hexagonal wurtzite (zincite) ZnO, whereas the fringes spaced at 0.266 and 0.345 nm matched the (101) and (110) planes of the tetragonal cassiterite structure of SnO 2 , respectively [22]. Energy-dispersive X-ray spectroscopy (EDS) analysis showed that the calcined samples were composed of homogeneously distributed Zn, Sn, and O elements throughout the belt structure, with no noticeable impurities (Fig. 1g). Thermogravimetric and differential scanning calorimetry (TG/DSC) analyses were conducted by heating the as-spun BLs in air from room temperature to 800°C at a heating rate of 5°C/min (Fig. 2a).
Weight loss below 150°C was ascribed to the evaporation of unbound moisture and loss of water molecules from the precursors contained in the BLs. The exothermic peak from 180 to 320°C resulted from the thermal decomposition of polyvinylpyrrolidone and anhydrous zinc acetate in forming ZnO [39]. The observed exothermic peak from 320 to 480°C resulted from the thermal decomposition of stannous chloride during the formation of the SnO 2 nanocrystals [40,41]. In a temperature range of 480-510°C, the weight loss was due to the decomposition of organic residues in the BLs. Beyond 550°C, the weight loss remained stable, suggesting a complete removal of organic residues from the precursors and polyvinylpyrrolidone. Microstructural XRD characterization was conducted over 2h = 20 -80°to examine the phases of ZnO-SnO 2 BLs. As shown in Fig. 2b, 10% ZnO-SnO 2 BLs and 20% ZnO-SnO 2 BLs indicated diffraction peaks ascribed to cassiterite SnO 2 (JCPDS No. 41-4145). The absence of ZnO peaks was due to a smaller amount of ZnO. The XRD patterns of 30% ZnO-SnO 2 BLs were indexed to a mixture of hexagonal wurtzite ZnO (JCPDS no. 36-1451) and tetragonal cassiterite SnO 2 (JCPDS no. 41-1445). Diffraction patterns ascribed to (100) and (101) of zincite ZnO were detected in 30% ZnO-SnO 2 BLs, with no traces of Zn 2 SnO 3 or Zn 2-SnO 4 ternary phases. The absence of these phases indicated that ZnO and SnO 2 phases grew in isolation [29]. The SnO 2 crystallite sizes were estimated using the Scherrer equation (D = 0.9k/bcosh, where D is the average crystallite size, k is the X-ray wavelength equal to 0.15406 nm, b is the Bragg angle, and h is the full width at half maximum of the X-ray diffraction liner) based on the (100), (101) and (211) planes. The average crystallite sizes were 14.3, 9.2, and 11.5 nm for 10% ZnO-SnO 2 BLs, 20% ZnO-SnO 2 BLs, and 30% ZnO-SnO 2 BLs, respectively. The results showed that the crystallite size decreased with the amount of ZnO. Crystallite sizes of ZnO were larger than for SnO 2 due to the crystallization of ZnO before SnO 2 , and the associated inhibition of the SnO 2 crystal growth [42].
XPS provided information on the chemical composition and binding states of ZnO-SnO 2 BLs. The spectrum of Zn 2p was fitted by two peaks at 1021.84 and 1044.86 eV for Zn 2p 3/2 and Zn 2p 1/2 of ZnO, respectively (Fig. 3a) [22]. The splitting of the Sn 3d XPS spectrum into symmetric Sn 3d 3/2 and Sn 3d 5/2 spectral peaks was evident at 494.84 and 486.45 eV, respectively, and was assignable to the doublet signature of Sn 4? in SnO 2 (Fig. 3b) [43]. As shown in

GasSensing results
The gas-sensing performance of ZnO-SnO 2 BLs was investigated at a temperature range of 350-450°C. For the 10% ZnO-SnO 2 BLs and 20% ZnO-SnO 2 BLs, the maximum response (R a /R g ) at 5 ppm H 2 gas was observed at a temperature of 400°C. The response of the 10% ZnO-SnO 2 BLs, 20% ZnO-SnO 2 BLs, and 30% ZnO-SnO 2 BLs to 5 ppm of H 2 at 400°C was 5.2,  (Fig. 4a). The 30% ZnO-SnO 2 BLs did not respond to H 2 at 350°C because their baseline resistance exceeded the capacity of the sensing instrument. The results indicated that optimal ZnO loading was 20 mol% for enhanced sensing at 400°C in dry condition. Before sensing tests, ZnO-SnO 2 BLs were exposed to air for 3.5 h to stabilize the baseline resistances at each operating temperature. The baseline resistances in air at 400°C are shown in Fig. 4b. The resistances stabilized approximately within 2.5 h of exposure to air. The baseline resistances of 10% ZnO-SnO 2 BLs, 20% ZnO-SnO 2 BLs, and 30% ZnO-SnO 2 BLs were 6.15, 6.4, and 7.18 MX, respectively. The increase in the baseline resistance of the ZnO-SnO 2 BLs with the amount of ZnO was attributed to intrinsic defects and low carrier concentration in ZnO [45]. The base resistances of the ZnO-SnO 2 BLs decreased upon exposure to H 2 and increased upon exposure to air, suggesting that the ZnO-SnO 2 BLs exhibited an n-type property. The H 2 sensing transients of 10% ZnO-SnO 2 BLs, 20% ZnO-SnO 2 BLs, and 30% ZnO-SnO 2 BLs at 400°C in dry conditions are shown in Fig. 4c. The H 2 response decreased monotonically with the decrease in H 2 concentration. The response and recovery times were determined from the graph of normalized sensor resistance at DR/DR max = 0.9 and DR /DR max = 0.1, where DR is the instantaneous change in resistance relative to the base resistance, and DR max is the difference between the sensor base resistance in air and the minimum resistance upon H 2 exposure. The 20% ZnO-SnO 2 BLs sensor exhibited the fastest response (3.6 s) upon exposure to 5 ppm of H 2 , compared to 10% ZnO-SnO 2 BLs (7.6 s) and 30% ZnO-SnO 2 BLs (6.3 s), as shown in Fig. 4d. The recovery times were 83.1, 94.2, and 56.7 s for 10% ZnO-SnO 2 BLs, 20% ZnO-SnO 2 BLs, and 30% ZnO-SnO 2 BLs, respectively (Fig. 4e).
The determination of sensor selectivity is critical for their potential application. The response of the 20% ZnO-SnO 2 BLs to HCHO, CH 4 , NH 3 , CO, and CO 2 was also determined at 400°C to ascertain the cross-sensitivity of the sensor. Figure 4f shows the response of the sensor to 5 ppm of H 2 and interfering gases at 400°C. The response of the 20% ZnO-SnO 2 BLs (R a /R g = 6.7) to H 2 was higher than HCHO, CH 4 , NH 3 , CO, and CO 2 . A 1.3-and 1.47-fold decrease in response of 20% ZnO-SnO 2 BLs toward 5 ppm H 2 was observed when the humidity increased to 55 and 80%, respectively, at 400°C (Fig. 5a). The results showed a small difference in response at 55 and 80% RH. The difference can be explained by the desorption of moisture from ZnO-SnO 2 BLs surfaces at 400°C, in agreement with reported findings [40]. The 20% ZnO-SnO 2 BLs showed a repeatability of response for 15 cycles of H 2 exposure (5 ppm) at 400°C in dry condition (Fig. 5b), suggesting that the sensor can achieve stable H 2 sensing measurements. A comparison of the H 2 sensing performance of ZnO-SnO 2 based materials reported by various authors is provided in Table 1. H 2 -sensing performance of ZnO-SnO 2 BLs can be explained by the ZnO-SnO 2 heterojunctions, surface electron depletion, and grain boundary mechanisms. In air, the oxygen molecules chemisorb in the vacant oxygen sites of ZnO and SnO 2 grain surfaces because of their high electron affinity (Eq. 1). The chemisorption of oxygen enlarges the electron-depleted region of ZnO and SnO 2 grains, increasing the resistance and bending of the energy band. The smaller crystallite size (9.2 nm) of SnO 2 in 20% ZnO-SnO 2 BLs implies that the crystallites can be entirely depleted of electrons when exposed to air. H 2 reacts with the chemisorbed oxygen to decrease the electron-depleted region and the inter-grain resistance to electron flow (Eq. 2) [46]. At the ZnO-SnO 2 interface, electrons can transfer from SnO 2 (work function U = 4.9 eV) to ZnO (work function U = 5.2 eV) to equilibrate the Fermi levels, followed by charge redistribution and band bending at the interface [47]. The resulting potential barrier prevents further transport of charge carriers across the ZnO-SnO 2 interface. The consumption of electrons at the n-n heterojunction and in the electron accumulation (ZnO) and depleted (SnO 2 ) regions by H 2 increases the response. The metallization ZnO by H 2 further reduces the resistance at the ZnO-SnO 2 interface and offers selectivity to H 2 [34]. Metallization of ZnO is caused by back-donation of electrons from neighboring O atoms to the Zn 4s states after adsorption of H 2 on ZnO [48].

Conclusions
In summary, this work demonstrated the fabrication of ZnO-SnO 2 BLs and their application as H 2 sensing layers. The reported strategy allows for the design of ZnO-SnO 2 heterogeneous composite structures in a belt morphology, which offer shorter diffusion lengths as well as enhanced reactions kinetics and sensitivity toward target gases compared to the conventional cylindrical fibers. The belt morphology, nano-and micropores, small crystallites, ZnO-SnO 2 heterojunctions, and metallization of ZnO were key determinants for high H 2 sensing performance, attaining a response of up to 6.7, and fast response (3.6 s) upon exposure to 5 ppm at 400°C. This work serves as a platform for developing reliable and fastresponding H 2 sensors based on ZnO-SnO 2 heterostructures.

Author contributions
All work including study conception and design, data collection, analysis, interpretation, drafting, and revision of the manuscript was done by the author.

Funding
The author declares that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability
Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Declarations
Competing Interests The author declares no relevant financial and non-financial interests to disclose.