Zinc Oxide Loaded Cellulose-based Carbon Gas Sensor for Selective Detection of Ammonia

1. Introduction

Ammonia is a colorless, toxic, corrosive substance with a strong odor and a choking effect. Excess ammonia can pose a huge threat to human health and to cause environment pollution. Different from other toxic gases, ammonia has a low boiling point of -33.5 ℃, a low melting point of -77.75 ℃, a low density of 0.771 g·L^-1, a refractive index of 1.33, and a dipole moment of 1.42 d. These characteristics make ammonia an excellent gas sensor, and its sensors can be used in a variety of applications such as environmental monitoring, agriculture or medical diagnostics, and industrial waste dealing [1,2].

Zinc oxide is a metal oxide semiconductor material with wide band gap (3.37 eV), high bonding strength and large exciton binding energy (60 meV) at room temperature. Based on these properties, it is often used in gas sensors, chemical sensors, biosensors, cosmetics, energy storage, optical and electronic devices and other products [3,4,5,6,7]. The existing zinc oxide gas sensors show great gas sensing performance for carbon dioxide, ammonia and ethanol. The resistive gas sensor based on ZnO nanosheets was prepared by Srinivasulu et al. which has a large surface area and can quickly and highly detect carbon dioxide in the air [8]. The ZnO nanoflowers were prepared by Yu et al. which have great gas sensing properties for low concentrations of ammonia [9]. However, the original zinc oxide still has some shortcomings such as poor gas sensitivity, slow response and recovery time and poor selectivity. Therefore, many researchers have used by doping other elements or mixing zinc oxide with other metal oxides to prepare composite materials in order to improving gas sensing performance [10,11,12,13,14].

It is an effective way to improve the gas sensing performance of the sensor by introducing carbon material into the gas sensing material based on zinc oxide. Using activated carbon fiber as a template, Chao et al. synthesized ZnO/C nanoporous fibers by hydrothermal method. The materials showed great gas sensing performance to low concentrations of ethanol and acetone at optimum operating temperature [15]. The heterojunction ZnO/hollow porous carbon microtubule (CMT) prepared by Sun et al. from simple carbonized buttonwood fluff fibers showed a great gas sensing response to trace ammonia molecules [16]. Hu et al. prepared composite materials by generating zinc oxide particles on nitrogen-doped carbon sheets, and the heterogeneous structure improved the gas sensing ability of composite materials to ppb grade NO₂ [17]. Cellulose, as a natural polymer material abundant on the earth, has a large number of hydroxyl functional group, excellent biodegradability and stable physical and chemical properties [18]. And compared with other carbon sources, cellulose is more available in nature, which can reduce the cost of producing carbon materials significantly. In recent years, there have been many reports on the application of cellulose-based carbon (CBC) as carbon materials in the field of supercapacitors and photocatalytic degradation by taking advantage of its excellent properties such as porous structure, huge specific surface area, and ability to adsorb gas molecules and macromolecular pollutants [19,20,21,22,23]. However, there are few researches on carbonizing cellulose into biomass carbon materials into ZnO-based gas sensors.

In this work, zinc oxide was prepared by a simple chemical precipitation method, and then it was carbonized with microcrystalline cellulose at high temperature, and ZnO/CBC were prepared for ammonia detection. Compared with the original ZnO-based ammonia sensor, due to the construction of the heterogeneous structure between ZnO and CBC, the composite has a great gas sensing response to a certain concentration of ammonia at room temperature, and has a better ammonia selectivity. The use of cellulose as the precursor of biomass carbon materials and combined with metal oxides to form composite materials provides a new reference in the field of gas detection.

2. Materials and Methods

2.1. Materials

Zinc acetate dihydrate (Zn(CH₃COO)₂·2H₂O, 98%) was get from Anneji (Shanghai) Pharmaceutical Chemical Co., Ltd. Ammonia liquor (NH₃·H₂O, 26%-28%) was purchased from Sinopharm Chemical Reagent Co., Ltd. Microcrystalline cellulose (50 μm) was obtained from Aladdin Reagent (Shanghai) Co., Ltd. Deionized water was prepared in the laboratory.

2.2. Synthesis of ZnO/CBC

ZnO was prepared by a simple chemical deposition method. Add 0.25 g zinc acetate dihydrate and 0.5 g microcrystalline cellulose to 30 mL deionized water, add ammonia water under magnetic stirring in an oil bath at 90 °C, make the molar ratio of Zn²⁺ to OH^- 1:15. Continued magnetically stirring at 90 ℃ for 1h. After the reaction, the solution was washed and centrifuged for several times until the pH was neutral, and then dried in a vacuum drying oven at 60 °C to obtain a white powder. The white powder was placed in a tube furnace at 600 °C and calcined by N₂ for 2 hours. The black powder obtained was recorded as ZnO/CBC-60% (the ratio of zinc acetate dihydrate to microcrystalline cellulose). According to the same experimental methods, ZnO/CBC-33%, ZnO/CBC-50%, ZnO/CBC-66%, ZnO/CBC-71% and CBC with different zinc oxide contents were prepared.

2.3. Characterization

The morphology and size of the samples were characterized by field emission scanning electron microscopy (SEM, ZEISS Sigma 300) and transmission electron microscopy (TEM, FEI Tecnai F30). The crystal structure of the sample was characterized by X-ray diffraction (XRD, Bruker D2 PHASER) of copper target radiation at ambient temperatures with 2θ values of 10 to 80 degrees and scanning rates of 5°/mins. The chemical composition and elements of the samples were characterized by X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-Alpha). Raman spectra were recorded using a Raman spectrometer (Raman, Horiba LabRAM HR Evolution) with a laser wavelength of 532 nm. In situ DRIFT spectroscopy was performed using Bruker VERTEX 80v infrared spectrometer. ZnO/CBC-60% was exposed to ammonia and pure nitrogen at room temperature. The Mott-Schottky curves of the samples were measured with an electrochemical workstation (CHI660E model), 0.5 M sodium sulfate solution is configured, Ag/AgCl is the reference electrode, platinum plate is the counter electrode.

2.4. Fabrication of gas sensing device

In order to perform electrical and sensing performance, a certain amount of samples were taken on an agate mortar and a small amount of deionized water was added to grind the samples to a sticky state. The grinding liquid was evenly coated on the interdigital electrode of the ceramic substrate and dried in a drying oven at 100 ℃. Gas sensing measurements are carried out using the self-made device shown in Figure 1. The static liquid-gas distribution method is used to generate ammonia gas environments with different concentrations. The calculation formula is shown as follows [24,25]：

$$C=\frac{22.4\times \mathrm{\varnothing }\times \rho \times p\times V_1}{M\times V_2}\times 1000$$

(1)

where C (ppm) stand for target gas concentration, Ø stand for required gas volume fraction, ρ (g mL⁻¹) stand for density of the liquid, p stand for purity of the liquid, V₁ (μL) stand for volume of liquid, V₂ (L) stand for volume of the chamber, and M (g mol⁻¹) stand for molecular weight of the liquid.

The change in resistance of the sensor was measured by a multimeter. The response of the sensor was defined as:

$$S=\frac{R_g-R_a}{R_a}\times 100\mathrm{\%}$$

(2)

where R_a is the resistance of the sensor in the air, and R_g is the resistance of the sensor after it passes into the target gas.

3. Results

3.1. Characterization of ZnO/CBC

The XRD patterns of ZnO/CBC and CBC with different ZnO precursor contents are shown in Figure 2. ZnO/CBC showed sharp diffraction peaks at 2θ = 31.8°, 34.4°, 36.3°, 47.5°, 56.6°, 62.9°, 66.4°, 67.9°, 69.1°, 72.6° and 77.0°, which were in accordance with (100), (002), (101), (102), (110), (103), (200), (112), (201), (004) and (202) crystal planes, respectively, The strongest peak occurred on the (101) crystal plane. All diffraction peaks correspond to the standard hexagonal wurtzite structure (JCPDS-99-0111) [26,27]. No diffraction peaks from other phases or impurities were observed. These results indicated that pure ZnO structures were formed through precipitation. Since the amorphous carbon was prepared and its crystallinity is low, wide peaks with low signal strength appeared near 22 ° and 42 ° [28,29,30]. The diffraction peaked strength is weaker than ZnO, while there could not be shown in the XRD pattern of ZnO/CBC.

Figure 3 shows the SEM images of ZnO/CBC and pure carbon with different ZnO precursor contents. It can be seen from Figure 3b–e that ZnO has been successfully loaded on CBC. ZnO exhibited a flower-like structure when the precursor content was small. With the ZnO precursor content increasing, ZnO gradually changed into a rod-like structure, at mean while the aspect ratio also becoming larger. This was due to the increase in ammonia solution, which promoted oxidation. Zinc preferentially grew on the C axis [31]. Figure 3a shows that the surface of CBC was relatively smooth after high temperature carbonization. However, after loading ZnO, the surface of CBC becomes rough. EDS was used to conduct elemental analysis of ZnO/CBC-60%. It can be seen from Figure S1 that C, O and Zn have been evenly distributed on ZnO/CBC-60%, and the atomic ratios of C, O and Zn were 77.50%, 11.84% and 10.66%, respectively. Since aluminum foil was used as the substrate, the Al signal appeared.

In order to further understood the structure of ZnO/CBC-60%, TEM characterization was performed. As shown in Figure 4a,b, it can be clearly found that there were different lattice fringes at the boundary of CBC and ZnO. The lattice fringes near 0.136 and 0.282 nm correspond to the (201) and (100) crystal faces of ZnO, respectively [11,32]. The lattice fringes near 0.202 and 0.350 nm correspond to the (101) and (002) crystal faces of graphite, respectively [33]. The lattice fringes of ZnO and CBC were interwoven, which also confirmed the construction of heterojunctions between ZnO and CBC.

As shown in the Raman spectrum in Figure 5a, it can be observed that peak D and peak G of typical carbon materials were located at 1340 cm^-1 and 1584 cm^-1, respectively, where peak D represents sp³ hybrid carbon of disordered or defective carbon, while peak G corresponds to sp² hybrid carbon of graphite structure [34]. The relative strength ratio (I_D/I_G) of peak D and peak G indicates the degree of graphitization of the material. The I_D/I_G values of pure carbon and ZnO/CBC-60% were 0.90 and 0.79, respectively, which were much higher than the I_D/I_G values of typical graphite, indicating that the prepared carbon materials were not highly crystalline and disordered materials exist. This was consistent with the observation of XRD. Moreover, the addition of ZnO improved the degree of disorder and defects in the structure of carbon materials [35]. The peaks at 80 cm^-1 and 427 cm^-1 in ZnO/CBC-60% correspond to $E_2^{low}$ and $E_2^{high}$, respectively. $E_2^{low}$ was affected by Zn²⁺ gap defect, and $E_2^{high}$ was affected by O^2- vacancy [36]. The simultaneous appearance of characteristic peaks of ZnO and carbon materials confirmed the successful preparation of ZnO/CBC. At the same time, it can be observed from Figure 6b that the D-peaks and G-peaks of ZnO/CBC-60% had a certain wave number displacement relative to pure carbon materials. It can be concluded that the heterojunction constructed by ZnO and cellulose-based carbon induces charge transfer between them [17].

In order to verify the elemental composition and surface chemical information of ZnO/CBC-60%, the materials were characterized by XPS. Figure 6a shows that ZnO/CBC-60% were mainly composed of Zn, C and O elements. Figure 6b was the high-resolution spectrum of Zn^2p. In the figure, the peaks of Zn^2p1/2 and Zn^2p3/2 were 1044.0 eV and 1021.1 eV respectively, and the binding energy distance before the two peaks was 22.9 eV, which proved the existence of divalent zinc ions[37]. Figure 6c shows the high resolution spectrum of O 1s. O 1s can be deconvolved into three peaks centered on 529.5 eV, 531.2 eV and 532.2 eV. The peak at 529.5 eV is attributed to the O^2- ion in the Zn-O bond in the ZnO wurtzite structure, and the peak at 531.2 eV is attributed to the Zn-O-C bond and the adsorbed hydroxyl group or water. The peak at 532.2 eV is attributed to the carbonate (C-O /C=O) species [38]. As shown in Figure 6d, three peaks appeared at 283.6 eV, 285.7 eV and 288.3 eV respectively, which belonged to the Zn-C bond, Zn-O-C bond and C=O bond respectively, indicating that the performance of ZnO/CBC-60% was different from that of pure ZnO [39].

Mott Schottky curve was used to determine the semiconductor type of the material to better explain the sensing mechanism of the material. Figure S2 shows the Mott Schottky curves of CBC and ZnO/CBC-60% at different frequencies measured at room temperature. It can be seen from the figure that the slopes of all curves are negative, indicating that CBC was a p-type semiconductor, and the prepared ZnO/CBC-60% also exhibited p-type semiconductor properties.

3.2. Results of Sensing Tests

Firstly, the gas sensing response of ZnO/CBC with different contents of CBC and ZnO precursors to 200 ppm ammonia at room temperature of 60% relative humidity was tested. When the sensor was in an air environment, the resistance of the sensor was at a stable value (R_a). When exposed to ammonia, the resistance of the sensor increased and reached a maximum value (R_g). When the sensor is exposed to air again, the resistance of the sensor will gradually return to the initial value (R_a). The resistance of CBC and ZnO/CBC increased gradually in ammonia environment and decreased gradually in air, which showed p-type semiconductor behavior. After testing the sensor, Figure 7a shows that when the ZnO precursor content was 60%, the sensor's gas sensing response to 200 ppm ammonia was the highest, reaching 27%. Therefore, zinc acetate/cellulose (wt%) =150 wt% was selected as the optimal ratio for preparing the required sensing material. At the same time, the ZnO/CBC-60% sensor was placed in the environment of methanol, isopropyl alcohol, ethanol and formaldehyde at 200 ppm at room temperature to test the gas sensing performance of the sensor to these gases, in order to verify the selectivity of the sensor to ammonia. It can be seen from Figure 7b that the gas sensing response of the sensor to ammonia is dozens of times that of other gases, indicating that the prepared ZnO/CBC-60% sensor had good selectivity to ammonia. This may be attributed to the fact that the heterostructure constructed between ZnO and CBC provided more active sites for the selective adsorption of ammonia molecules.

In order to further test the gas sensing performance of ZnO/CBC-60% for ammonia. Sensor was put in ammonia environment with five concentrations of 25 ppm, 50 ppm, 100 ppm, 150 ppm and 200 ppm at room temperature for gas sensing test. The gas sensing response of the sensor to ammonia with a concentration of 25-200 ppm at room temperature is shown in figure 8a. As the ammonia concentration increased, the gas sensing response of the sensor increased, which was in line with the expected goals of the experiment. Stability is one of the key parameters of gas sensing. We tested the stability of ZnO/CBC-60% sensor by placing the sensor in 200 ppm ammonia environment for five rounds of gas sensing tests. As shown in figure 8b, after five sensing cycled, the gas sensing performance of ZnO/CBC-60% sensor had not weakened, indicating that the sensor had good repeatability.

Figure 8. (a) Response of ZnO/CBC-60% to ammonia with concentration of 25-200 ppm, (b) Repeatability evaluation of ZnO/CBC-60% for 200 ppm ammonia at RT.

Figure 8. (a) Response of ZnO/CBC-60% to ammonia with concentration of 25-200 ppm, (b) Repeatability evaluation of ZnO/CBC-60% for 200 ppm ammonia at RT.

Relative humidity is an important factor affecting the performance of gas sensor [40]. as shown in Figure 9a, with the increase of relative humidity, the gas sensitivity of ZnO/CBC-60% sensor to ammonia was correspondingly weakened, which may be due to the fact that H₂O molecules in the test environment preempted the active sites, so that NH₃ molecules could not fully react with the materials, resulting in the decrease of gas sensing response[18].

The ZnO/CBC-60% sensor was put in an ammonia concentration environment of 200 ppm for one week. As can be seen from Figure 9b, the gas sensing response of the sensor to ammonia had not changed greatly, which indicated that the sensor had good stability.

4. Discussion

In order to prove that the gas sensing response of ZnO/CBC-60% was the result of the reaction with NH₃ molecules, the material was analyzed by in-situ DRIFT spectroscopy. As shown in Figure 10a, after introducing the mixed gas of NH₃ and N₂, it can be observed that adsorption peaks of coordinated NH₃ species corresponding to Lewis sites appeared near 929, 964, 1619 and 3334 cm^-1 [41,42,43], and these peaks gradually became stronger with the increase of adsorption time. Figure 10b shows that the peak value was obviously weakened after N₂ purging. These phenomena show that the adsorption and desorption of ammonia occurred on the surface of ZnO/CBC-60%, which explained the gas sensing response of ZnO/CBC-60% to ammonia.

Figure 11 shows the interface energy band diagram of ZnO/CBC sensor in air and ammonia. It can be seen from Mott Schottky curve that CBC was a p-type semiconductor, while ZnO was a typical n-type semiconductor, The combination of the two created a p-n heterojunction. Due to the difference in concentration of electrons and holes in the two materials, the free electrons ZnO conduction band will diffuse to CBC, and the holes in CBC will diffuse to ZnO until the Fermi level reached a new equilibrium [44,45]. This process made the energy band bend, forming a narrow depletion layer, forming an electron accumulation layer on the CBC side and a hole accumulation layer on the ZnO side. When ZnO/CBC sensor was exposed to ammonia, NH₃ molecules will capture hole protons from CBC, and the hole accumulation layer will provide more adsorption sites for ammonia adsorption. In addition, NH₃ molecules will release some electrons to ZnO/CBC, which will neutralize the holes and further widen the depletion layer on the CBC side, resulting in an obvious increase [46], in the resistance of the sensor, which was consistent with the phenomenon in gas sensing test.

5. Conclusions

In this experiment, ZnO was loaded on CBC, and the gas sensing response of ZnO/CBC loaded with different ZnO contents to ammonia at room temperature was studied by various the content of ZnO precursors. The composites were characterized by SEM, TEM, XRD, Raman and XPS. The results showed that ZnO had been successfully loaded on CBC. Subsequently, the gas sensing performance showed that the ZnO/CBC with 60% ZnO precursor content had better gas sensing response to 200 ppm ammonia at room temperature, reaching 27%. In addition, through five consecutive rounds of gas sensing response tests in 200 ppm ammonia environment, it was found that ZnO/CBC-60% exhibited good stability. At the same time, by comparing the gas sensing performance of several VOC gases, it was found that the material for ammonia had well selectivity. In-situ DRIFT spectroscopy proved that the sensing material did react with ammonia, and the ZnO/CBC-60% could provide more active sites for ammonia molecules because of the construction of heterojunction between ZnO and CBC. Therefore, ZnO/CBC composites had improved sensing performance. This study provides a new idea for the design and synthesis of biomass carbon-metal oxide composites.