A MEMS-based gas sensor for H₂S detection with enhanced performance using a Ni–In₂O₃/ZnO nano-composite

2.1 Sensing principle

Figure 1(a) displays the experimental arrangement for the MEMS-based H₂S gas sensor. In this study the sensor consists of a digital power supply (DP 100), an electrochemical workstation (CH Instruments), a MEMS sensor chip, a gas chamber, and the reacting gases (hydrogen sulfide and other reactant gases) to form a MEMS gas sensing system. One end of the MEMS chip is connected to the digital power supply (DP 100), and the other end is connected to the CHI760E Shanghai Chenhua electrochemical workstation (CH Instruments) for real-time monitoring of electrical signals. In Figure 2(a), the gas sensor chip is displayed in the Nexcope NE700 series microscope. The lens uses a HAYEAR 4K USB 3.0 camera. The size of the MEMS gas sensor chip is 1 × 1 mm. The gas chamber of the gas sensor is made of 2 mm thick acrylic material, and the internal volume is 10 × 5 × 5 cm³, as shown in Figure 2(b). The MEMS gas sensor device is shown in Figure 2(c). The internal structure of the chip used in this experiment has a support layer, a sensitive film, two sets of electrodes, one set of heating electrodes and one set of test electrodes. The sensitive film prepared in this experiment is spin-coated on the top layer, as shown in Figure 2(d).

Figure 1:

The gas sensor system. (a) MEMS sensor H₂S monitoring experimental device diagram, (b) gas sensing mechanism of Ni–In₂O₃/ZnO.

Figure 2:

MEMS sensor with (a) the MEMS chip, (b) the air chamber, (c) the gas sensor device, and (d) internal structure of the chip.

The fundamental structure of this sensor comprises a MEMS microheater and a resistor composed of Ni–In₂O₃/ZnO nano-multistage heterostructured nano-materials. The microheater is made using MEMS technology, capable of providing a constant temperature to boost both the sensitivity and the stability of the sensor. In₂O₃/ZnO ¹¹ heterostructured nano-materials are synthesized using the hydrothermal method, followed by Ni doping to finally produce Ni–In₂O₃/ZnO, which serves as the sensitive layer for the sensor. The sensitive material is uniformly spin-coated on the MEMS microheater and then placed in a gas chamber for gas experiments. When the target gas enters the sensor, it adsorbs onto or reacts with the Ni–In₂O₃/ZnO nano-sensitive film, altering the concentration and mobility of carriers (electrons or holes) on the surface or in the volume, thus changing the resistance. ¹³

The working principle ^{14 , 15} of resistive semiconductor sensors is as follows: upon air exposure, the surface of the Ni–In₂O₃/ZnO nano-wires becomes coated with oxygen atoms from the surrounding environment. This reaction reduces oxygen vacancies on the ZnO nano-wires, lowering the surface carrier concentration and increasing the resistance. The amount of oxygen adsorbed varies with the sensor’s operating temperature due to oxygen’s high electron affinity. At 300 °C, the oxygen adsorbed as O⁻, tends to withdraw electrons from the material’s surface, resulting in a charge depleted layer on the Ni–In₂O₃/ZnO surface (Equations (1) and (2)). ^{14 , 15} The depletion of electrons causes the barriers at the interfaces between nearby oxide particles to thicken, which subsequently enhances the resistance of the sensor material. As a result, in the presence of air, the sensor maintains an elevated resistance (Ra) state. When H₂S, a reducing agent, is introduced into the chamber, it chemically reacts with the oxygen ions bound to the sensor’s surface. Oxygen adsorbed via chemisorption releases trapped electrons into the material, causing surface electrons to re-enter their conduction bands (Equations (3) and (4)), ¹⁵ thereby reducing the sensor’s resistance Rg).

For sensor fabrication, the Ni–In₂O₃/ZnO heterojunction is specifically utilized to detect hydrogen sulfide (H₂S) as shown in Figure 1(b). When gas molecules react with the oxygen bonded to the heterojunction, electrons previously held on the surface are freed and return to the conduction bands. This leads to a reduction in particle boundary barriers, a shallower electron depletion zone, and a swift drop in the sensor’s overall resistance. ¹⁵ When the reducing gas is present, the resistance (Rg) greatly decreases, significantly enhancing the sensing response.

3.1 Preparation of the sensing materials In₂O₃ and ZnO

To prepare In₂O₃ nano-materials, a solution was created by gradually adding 0.5 g of indium nitrate, 0.3 g of sodium citrate, and 0.1 g of urea into 40 mL of purified water, ensuring constant agitation with a magnetic stirrer. After complete dissolution, a clear and homogeneous solution was obtained, which was then transferred to a 100 mL Teflon-lined autoclave for hydrothermal treatment at 200 °C for 20 h. The solid product was extracted by centrifugation, rinsed with ethanol several times, and dried thoroughly. Subsequently, the material was heated at 500 °C for 3 h with a controlled temperature increase of 3 K min⁻¹, producing high-quality In₂O₃ nano-material. ¹⁵

For the preparation of ZnO, 7.44 g (0.025 mol) of zinc nitrate hexahydrate and 2.94 g (0.0132 mol) of sodium citrate were dissolved in 20 mL of distilled water with magnetic stirring (4 rpm) for 100 min to achieve a clear and homogeneous solution. Aqueous sodium hydroxide solution was introduced drop by drop, maintaining stirring until the solution’s pH was adjusted to 13. The mixture was then placed into a Teflon-lined autoclave and heated at 150 °C for 12 h. After cooling to ambient temperature, the solid material was isolated through centrifugation, rinsed four times with ethanol to eliminate remaining impurities, and dried at 80 °C for 5 h. The dried product was annealed at 200 °C for 3 h, with a controlled temperature increase of 5 K min⁻¹, yielding ZnO powder characterized by excellent crystallinity and purity. ¹⁶

3.2 Preparation of the sensing material Ni–In₂O₃/ZnO

All chemicals used in this study were of analytical grade. 0.5 mmol of indium nitrate In(NO₃)₃ · xH₂O and 1 mmol of zinc nitrate Zn(NO₃)₂ · 6H₂O were dissolved in a mixed solvent of 30 mL glycerol and 50 mL isopropanol (3:5 v/v) under vigorous stirring for 30 min. Then, the mixture was transferred into a 100 mL PTFE-lined reactor and subjected to the hydrothermal reaction at 180 °C for a duration of 6 h. The reaction mixture was taken out of the autoclave and allowed to air-cool to ambient temperature. The precipitate was centrifuged and rinsed three times with deionized water and pure ethanol. Finally, the material was dried at 80 °C for 1 h to acquire the In/Zn-glycerate precursor which was calcined at 350 °C at a heating rate of 2 K min⁻¹ for 3 h to form the In₂O₃/ZnO heterostructure. ¹⁷ 1 mmol (74.69 mg) of the NiO was ground it together with In₂O₃/ZnO in a natural agate mortar for 3 h. After sieving through a 300-mesh screen, the mixed powder is placed into an alumina crucible and heated at 800 °C for 2 h to synthesize the Ni–In₂O₃/ZnO nano-material.

3.3 Characterization

The crystal structure analysis was conducted using a Bruker D8 ADVANCE X-ray diffractometer, set at 40 mA tube current and 40 kV tube voltage, with Cu and Co targets emitting wavelengths of 1.5406 and 1.79026 Å, respectively. The elemental composition of the nano-particles was examined through X-ray photoelectron spectroscopy (XPS) using a Thermo Escalab 250XI from Thermo Fisher Scientific. The system has a monochromatic AlKα source (hν = 1,486.6 eV) at 150 W, with a 650 µm beam spot, 14.8 kV voltage, and a 1.6 A current. To correct for charge, the C1s peak of surface carbon contamination at 284.8 eV was used. The sample surfaces were studied using a HITACHI S-4800 field emission SEM to investigate their morphological characteristics. Elemental analysis was conducted with the HORIBA EMAX mics2 system, utilizing acceleration voltages of 3 or 5 kV and achieving a resolution of 50 nm. The broad scan was performed using a 100 eV pass energy and a 1 eV step size, whereas the fine scan applied a 20 eV pass energy with a 0.1 eV step size. All tests were conducted at ambient temperature.

3.4 MEMS gas sensing characteristics

The sensitive material was mixed with deionized water (DI) in a 4:1 ratio by weight and ground in an agate mortar to produce a paste. Using a micropipette, the mixture was placed onto the micro-hotplate, air-dried at ambient temperature, and subsequently aged in a vacuum oven. The aging temperature was stabilized at 350 °C for 2 days. The sensing component was subjected to a 500 °C heat treatment for 2 h, removing the bound water and forming a compact oxide layer on the gas sensor’s surface, ultimately producing the MEMS sensor. Once cooled, a CNC power supply was used to power the electrodes, ensuring that the micro-hotplate reached the required reaction temperature. To evaluate the sensor’s electrical behavior, an instrument for gas sensing analysis (model CHI760E, manufactured by Shanghai Chenhua) was employed. During the test, the testing device was placed in a fume hood. The target gas, combined with dry air, was injected into the chamber and left to diffuse, ensuring complete exposure to the sensitive surface of the chip over a set period. The CHI760E software was then opened to begin testing the I–T curve. Response and recovery times refer to the intervals required for the sensor to attain 90 % of the complete current shift during the phases of gas adsorption and desorption. The operating temperature of the sensor was 300 °C.

4.1 Crystal structure

Figure 3 presents the XRD patterns of In₂O₃, ZnO, In₂O₃/ZnO, and Ni–In₂O₃/ZnO, along with their simulated reference patterns. The diffraction peaks of In₂O₃ correspond to a cubic bixbyite structure (space group Ia 3 ‾ , #206), ¹⁸ while ZnO exhibits the characteristic reflections of the hexagonal wurtzite phase (space group P63mc, #186). ¹⁹ All simulated patterns were generated using VESTA software with CIF data from the Materials Project and CuKα radiation (λ = 1.5406 Å).

Figure 3:

The XRD patterns of (a) In₂O₃, (b) ZnO, (c) In₂O₃/ZnO and (d) Ni–In₂O₃/ZnO.

The In₂O₃/ZnO composite displays combined diffraction peaks from both In₂O₃ and ZnO, with no evidence of new phases, indicating good phase compatibility. In the Ni–In₂O₃/ZnO sample, additional peaks were observed, corresponding to the cubic NiO phase (space group Fm 3 ‾ m, #225), ²⁰ suggesting that some of the Ni²⁺ ions introduced during synthesis formed a separate crystalline NiO phase ²¹ rather than fully incorporating into the host lattice.

These results confirm the coexistence of In₂O₃, ZnO, and NiO in the doped composite. No unexpected or unidentified peaks were detected, supporting the phase purity and structural integrity of the synthesized materials.

4.2 Compositions

Figure 4(a) presents the XPS spectra of In₂O₃, ZnO and the In₂O₃/ZnO composite, while Figure 4(b) shows the In 3d peaks of In₂O₃ and In₂O₃/ZnO, Figure 4(c) the Zn 2p peaks of ZnO and In₂O₃/ZnO, and Figure 4(d) the O 1s peaks of In₂O₃, ZnO and In₂O₃/ZnO.

Figure 4:

The XPS spectrum. (a) The XPS full-scan images of In₂O₃, ZnO and In₂O₃/ZnO, (b) the 3d regions of In₂O₃ and In₂O₃/ZnO, (c) the Zn 2p regions of ZnO and In₂O₃/ZnO, (d) the O 1s regions of In₂O₃, ZnO and In₂O₃/ZnO.

Figure 5(a) presents the XPS spectrum of the Ni–In₂O₃/ZnO composite. The analysis of the spectrum confirms that the composite contains the elements nickel (Ni), indium (In), zinc (Zn), and oxygen (O). The C 1s peak ²² is attributed to organic carbon contamination or trace amounts of glycerol that are not fully decomposed in the composite (Figure 5(b)). Binding energies of 854.40 eV and 873.11 eV are assigned to Ni 2p _3/2 and Ni 2p _1/2 ²³ (Figure 5(c)). The low doping levels result in weak peak intensities, accompanied by satellite peaks at 861.36 and 879.16 eV. As demonstrated in Figure 5(d), the fitted Zn 2p peaks reveal binding energies of 1,044.3 and 1,022.2 eV, which align with Zn 2p _1/2 and Zn 2p _3/2, respectively. Based on previous studies, ²⁴ the In 3d peaks, shown in Figure 5(e), are identified as follows: 444.60 eV represents the In 3d _5/2 peak, while 452.18 eV is attributed to In 3d _3/2 peak. To determine the exact proportion of chemisorbed oxygen, the O 1s peak ²⁵ was analyzed through XPS testing. As shown in Figure 5(f), the deconvolution of the O 1s peak reveals three distinct peaks at binding energies of 529.8 eV (O_Ⅲ), 530.5 eV (O_Ⅱ), and 531.8 eV (O_Ⅰ). The O_Ⅰ peak corresponds to lattice oxygen, while the O_Ⅱ peak is linked to adsorbed oxygen. The binding energy at O_Ⅱ (Oc) is typically attributed to chemisorbed or dissociated oxygen. ²⁶ In resistive gas sensors, oxygen chemisorption on the material surface is critical, as it governs both the redox interactions between gas molecules and oxygen species, along with the gas adsorption and desorption mechanisms.

Figure 5:

The XPS spectrum of Ni–In₂O₃/ZnO. (a) XPS full-scan images of Ni–In₂O₃/ZnO, (b) the C 1s region, (c) the Ni 2p region, (d) the Zn 2p region, (e) the Sn 3d _5/2 region, (f) the O 1s region.

4.3 Morphology

The SEM images of indium oxide are shown in Figure 6(a)–(c). From Figure 6(c), it can be seen that the crystals of indium oxide are in a cube-like shape. ¹⁸ The images of zinc oxide are presented in Figure 6(d)–(f), and Figure 6(f) shows that zinc oxide is present in the form of flakes. The images of indium oxide/zinc oxide are shown in Figure 6(g)–(i), where Figure 6(i) illustrates that the granular objects are attached to the flaky material. The SEM images of the sensor material, shown in Figure 6(j)–(l), show the Ni–In₂O₃/ZnO morphology at magnifications of 20×, 50×, and 100×. At 100× magnification, the surface of the sensor material is clearly visible. The microscope images reveal irregular small droplet shapes with inconsistent sizes, in a loose and porous structure. The size of the smallest irregular spherical grains is mainly in the range of 100–250 nm in diameter. The grain boundaries are distinct, and the surface is relatively smooth. The presence of gaps between the blocks facilitates gas adsorption and desorption. ²⁷

Figure 6:

Figures (a), (b) and (c) show the FE-SEM images of In₂O₃, figures (d), (e) and (f) the FE-SEM images of ZnO, figures (g), (h) and (i) the FE-SEM images of the In₂O₃/ZnO sensing film, and figures (j), (k) and (l) the FE-SEM images of Ni–In₂O₃/ZnO.

Elemental distribution maps of the Ni–In₂O₃/ZnO material are illustrated in Figure 7(a)–(e), showing that the synthesized Ni–In₂O₃/ZnO consists solely of the intended elements, with no impurities present. Figure 4(f) shows the EDS spectrum of Ni–In₂O₃/ZnO.

Figure 7:

The SEM image. (a) The HAADF image of Ni–In₂O₃/Zn, the mapping images of (b) O, (c) Zn, (d) Ni and (e) In. Figure (f) shows the EDS spectrum of Ni–In₂O₃/ZnO.

4.4 Gas sensing

The gas sensor prepared in this work was tested with four different sensing materials under 100 ppm concentration of hydrogen sulfide gas. As shown in Figure 8, the response of Ni–In₂O₃/ZnO to hydrogen sulfide gas is the highest. Therefore, this material is chosen as the sensing film for this study.

Figure 8:

The gas sensor testing of different materials under 100 ppm hydrogen sulfide conditions.

At an operating temperature of 300 °C, the Ni–In₂O₃/ZnO nano-materials were tested under a broad range of hydrogen sulfide gas concentrations. The tested gas concentrations were 0, 20, 50, 80, and 100 ppm. In Figure 9(a), the sensor’s reaction patterns to differing levels of hydrogen sulfide gas concentrations are illustrated. An increase in test gas concentration leads to a corresponding rise in the response output of the MEMS sensor covered with the sensitive material. The MEMS gas sensor achieved the best sensitivity to H₂S at 3.4 ± 0.12 μA ppm⁻¹, indicating a linearity of 0.90201, as shown in Figure 9(b).

Figure 9:

The gas sensor testing of (a) the gas sensitivity response curve of the MEMS sensor for different concentrations of H₂S, (b) the linear fitting image, (c) the selectivity of the Ni–In₂O₃/ZnO MEMS senor towards H₂S, CO₂, CO, Cl₂, SF₆, ethanol and air, (d) the response and recovery time of the Ni–In₂O₃/ZnO micro-hotplate sensor, (e) the reproducibility of the Ni–In₂O₃/ZnO MEMS sensor for H₂S at 50 ppm, (f) the stability of the gas sensor.

By performing a linear analysis of the experimental data using Origin, the standard deviation (δ) was obtained, and the detection threshold (LOD) of the gas sensor was computed, with K being the slope of the linear fit. The LOD of the sensor is, ²⁸

The calculated limit of detection for the MEMS gas sensor can reach 0.57 ppm.

To examine how the coated MEMS sensor responds to different gases with cross-sensitivity, gases with equal concentrations of H₂S, CO₂, CO, Cl₂, SF₆, ethanol, and air were prepared and introduced into the gas chamber. Figure 9(c) presents the gas selectivity chart for the Ni–In₂O₃/ZnO sensor. It demonstrates minimal sensitivity when exposed to the tested gases such as CO₂, CO, Cl₂, SF₆, ethanol, and air at the same concentration, indicating that Ni–In₂O₃/ZnO exhibits specific selectivity for H₂S gas.

The response-recovery behavior of the sensor is depicted in Figure 9(d). Measurements were conducted at 0.1 s intervals, and random time points were selected for measuring the response time. Repeated measurements of the sensor’s response and recovery times were conducted to remove inaccuracies caused by manual computations. With a response time of approximately 4.4 s and a recovery time of 5.2 s, the sensor exhibits a quick reaction, considerably faster than the sensors referenced in Table 1. ^{29 , 30 , 31 , 32 , 33 , 34 , 35}

Figure 9(e) shows the reproducibility of sensors with different parameters in 100 ppm H₂S gas. For a single sensor, the relative standard deviation of the reproducibility, or sensitivity error should not exceed ±5 %. Since the reproducibility error of the Ni–In₂O₃/ZnO MEMS sensor meets this standard, it indicates that the sensor has relatively good reproducibility. The stability of the MEMS gas sensor was evaluated over a span of 30 days. Every 5 days, the sensor’s response value in a 50 ppm hydrogen sulfide gas environment was measured. As seen in Figure 9(f), the results reveal minor variations in response, confirming the consistent performance of the Ni–In₂O₃/ZnO gas sensor over both short and extended periods, which underscores its suitability for real-world market usage.