**1. Introduction**

Triethylamine (TEA) is a colorless, transparent oily liquid with strong ammonia odor and was widely used as an organic solvent, raw material, polymerization inhibitor, preservative, catalyst and synthetic dye [1,2]. However, TEA is also a flammable, explosive and toxic volatile organic gas that can harm human health, such as eye and skin irritation, dyspnea, headache, nausea and even death [3–6]. Therefore, it is very important to develop a method for detecting and monitoring the concentration of TEA. Up to now, gas/liquid/solid chromatography, gel chromatography, ion mobility spectrometry, electrochemical analysis and colorimetry and other methods have been explored to monitor TEA gas. However, these methods require expensive equipment and complex detection processes, which hamper their widespread use in real life [6–12]. Thus, it is very necessary to develop a device that is simple to manufacture and easy to detect TEA.

Metal oxide semiconductor (MOS) based gas sensors have been widely investigated in recent years because of their advantages such as simple detection, simple preparation, low cost, high sensitivity and good real-time performance [13–15]. SnO2 is a wide band gap n-type MOS material with a band gap width of 3.62 eV at room temperature. It is widely used in various fields such as photocatalysts [16,17], solar cells [18,19], lithium-ion batteries [20,21] and gas sensors [22–24]. As a gas sensor, SnO2 is one of the most widely considered gas sensitive materials due to its better gas sensitivity to various organic and toxic gases. However, we also know that the traditional SnO2 based gas sensors has some obvious problems of low gas response, high optimum working temperature, poor selectivity and stability [25–27]. Hence, researchers use another MOS doped SnO2 to improve its gas sensitivity. For example, Zhai et al. [28] prepared Au-loaded ZnO/SnO2 heterostructure, this sensor not only reduces the optimum operating temperature of SnO2 but also has a higher response to TEA than pure SnO2. Yang et al. [24] synthesized porous SnO2/Zn2SnO4 composites and their response to 100 ppm of TEA was about 2–3 times higher than that of pure SnO2. Yan et al. [29] reported a kind of porous CeO2-SnO2 nanosheets, which exhibited excellent gas sensing properties toward ethanol compared with the pure SnO2. These works have confirmed that another MOSs doped SnO2 can indeed improve gas sensing properties. As a rare earth element, Ce not only has the highest abundance of elements but also has some special characteristics such as high oxygen storage capacity, rich oxygen vacancies and low redox potential [29–31], these characteristics make it an ideal candidate for gas sensing materials [32,33], such as Motaung et al. [34] synthesizing CeO2-SnO2 nanoparticles with a dramatic improvement in sensitive and selective to H2, Dan et al. [35] prepared novel Ce-In2O3 porous nanospheres for enhancing methanol gas-sensing performance. Xu et al. [12] reported that a cataluminescence gas sensor based on LaF3-CeO2 has better sensitivity and selectivity for TEA. However, as far as we know, there are few reports about the materials of CeO2-SnO2 used to detect TEA.

Herein, we synthesized the flower-like CeO2-SnO2 composites via a facile hydrothermal method. The gas sensing performance of the flower-like CeO2-SnO2 composites based sensors were tested. The experimental results indicate that the TEA gas sensing performance of the CeO2-SnO2 composites based sensors are significantly improved by the modification of a small amount of CeO2 compared with pure SnO2, especially in terms of sensitivity and selectivity. The improvement of gas sensing properties of the composite materials is mainly due to the formation of n-n heterojunction between CeO2 and SnO2.

#### **2. Results and Discussion**

#### *2.1. Sample Characterization*

Figure 1 displays XRD patterns of the pure SnO2 (SC-0), 3, 5 and 7 wt.% CeO2 decorated SnO2 (SC-3, SC-5, SC-7) samples. It can be seen from Figure 1 that the XRD peaks are sharp and coincide with those of the tetragonal rutile of SnO2 in the space group P42/mnm with lattice constants ofa=b = 4.738 Å and c = 3.187 Å (JCPDS file No. 41−1445). Apart from, all samples showing the same crystal planes of (110), (101), (200), (211), (220), (310), (301) and (321), respectively. However, no reflections characteristic of CeO2 was observed even at the maximum CeO2 content of 7 wt.%, which may be mainly due to less CeO2 loading. Nevertheless, one can also see from the figure that all the diffraction peak becomes sharper as the content of Ce increases. According to Debye-Scherrer principle as shown in Equation (1), where K (K = 0.89) is the Scherrer constant, D is the crystallite size, λ (λ = 0.15406 nm) is the X-ray wavelength, B is the half-height width of the diffraction peak of the measured sample, *θ* is the diffraction angle. The calculated average particle sizes of SC-0, SC-3, SC-5 and SC-7 were 11.2350, 14.0154, 14.4425, 14.633, respectively. The combined diffraction peaks became sharper, indicating that CeO2 was indeed loaded onto SnO2, since the n-n heterojunction is formed between CeO2 and SnO2 and the average particle size of the composite becomes larger as the CeO2dopant increases.

$$\mathbf{D} = \frac{\mathbf{K}\lambda}{\mathbf{B}\cos\theta} \tag{1}$$

**Figure 1.** XRD patterns of the samples.

The morphology and structure of pure SnO2 (Figure 2a,b) and SC-5 nanoflowers (Figure 2c,d) were observed by SEM and the presence of Ce dopant in the SC-5 nanoflowers (Figure 2e–g) was detected by EDS as illustrated in Figure 2. In Figure 2a,b, the SnO2 sheets are gathered together to look like a spider web and have no flower-like structure. While, in Figure 2c,d, one can be clearly seen that SC-5 has a flower-like structure with an average diameter of about 1 μm and many nanoparticles are relatively evenly dispersed on the surface of the flower-like structure. By EDS detection as shown in Figure 2e–g, it was proved that only three elements of Sn, O and Ce were found in the composite, which proved that the sample had a relatively high purity.

**Figure 2.** Field-emission scanning electron microscopy (FESEM) images of pure SnO2 (**<sup>a</sup>**,**b**) and SC-5 nanocomposite (**<sup>c</sup>**,**d**) and the EDS images (**e**-h) of the SC-5 sample.

It is well known that the presence of a heterojunction affects the band gap width of a material. Hence, in order to further illustrate the formation of n-n heterojunctions, we investigated the band gap energies of the synthesized SnO2 and SC-5 by UV-vis absorption spectra. As can be seen in Figure 3, the absorption peak of the red SC-5 sample moved significantly upward compared to the pure SnO2 curve. The relationship diagram between *(αhv)<sup>2</sup>* and photon energy *hv* (illustration in the upper right corner of Figure 3) is obtained according to the formula *(αhv)<sup>2</sup> = A (hv* − *Eg)*, where α is the absorption index, *h* is the Planck constant, *v* is the light frequency, *Eg* is the semiconductor bandgap width, *A* is a constant associated with the material. It can be seen from the illustration that the band gap energy values of pure SnO2 and SC-5 are 3.64 eV and 3.56 eV, respectively. The data show that the doping of CeO2 narrows the band gap energy of the composite, which further proves that an n-n heterojunction is formed between composite nanomaterials.

**Figure 3.** UV–vis absorption spectra of the synthesized SC-0 and SC-5 samples. The upper right corner inset is the relationship lines of (*αhv*)<sup>2</sup> and *hv*.

#### *2.2. Gas Sensing Performance*

In order to study the gas sensing property of the as-synthesized samples to TEA, a series of examinations on the pure SnO2 and CeO2-SnO2 composites were performed. Because the working temperature has grea<sup>t</sup> impact on the response of the gas sensor, the response of four sensors to 200 ppm TEA at different temperatures was first studied, as shown in Figure 4. One can see from Figure 4 that the four curves show similar variation tendency, which is first increase and then decrease as the working temperature increases. And the response of all the sensors reached their top value at the working temperature of 310 ◦C. This may be due to the amount of chemisorbed oxygen ions on the surface of the sensor has reached a sufficient amount to react with TEA and the effective reaction on the surface of MOS causes an eminent change in resistance [36]. Moreover, the response of the sensors based on SC-0, SC-3, SC-5 and SC-7 are 65.77, 218.12, 252.21, 156.38, respectively. It can also be clearly seen from the Figure 4 that all composite sensors have higher response to TEA than pure SnO2 and the sensor based on SC-5 show higher response value than the response of the other two composite sensors. The TEA gas-sensitive properties of the SC-5 composite prepared in this work and the materials reported in other literatures [4,28,37–41] are shown in Table 1. Although the sensitivity of the CeO2-SnO2 sensor is higher than that of the reported results, the optimum operating temperature has not been improved very well. Therefore, we still need to do more in-depth research on reducing the working temperature, such as introducing photoexcitation equipment [42] or using p-type semiconductor doping modification [43].

**Figure 4.** The response of the synthesized samples to 200 ppm TEA at different operating temperatures.


**Table 1.** TEA sensing performance comparison between this study and other reported results.

Sensitivity, selectivity and stability are also three important properties for evaluating sensor quality. Figure 5a exhibits the response of the four gas sensors to different TEA concentrations in the range of 20–2000 ppm at 310 ◦C. Obviously, the response of the four gas sensors increases with the increase of TEA concentration. It can also be seen that the response is almost linearly related to TEA concentration and the slope of the curve increase rapidly as TEA concentration increases in the concentration range of 20–200 ppm (inset of Figure 5a). Above 200 ppm, the responses increase slowly as the gas concentration increases, indicating that the adsorption of TEA by the sensor gradually becomes saturated. In addition, the SC-5 sensor exhibits higher response than that based on SC-0, SC-3 and SC-5 at different TEA concentrations. Figure 5b shows the dynamic response-recovery curves of the SC-0 and SC-5 sensors to TEA in the concentration range of 20–2000 ppm at 310 ◦C. As can be seen from Figure 5b that as the TEA concentration increases from 20 to 2000 ppm, the response amplitude of the two sensors gradually increases. The SC-5 composite sensor has a much higher response to the same TEA concentration than the pure SnO2 sensor, indicating improved gas sensitivity of the composite. Moreover, after several times of gas injection, the output voltage of the sensor in air can still return to the original value, which means that the sensor has better repeatability. Figure 5c displays the results of selective testing of the SC-0 and SC-5 sensors for five different 200 ppm reducing gases, including formaldehyde, methanol, acetone, methane and triethylamine. As can be seen in Figure 5c, the flower-like SC-5 sensor has a significantly higher response to all detected reducing gases than the response of the SC-0 sensor, further demonstrating that the SC-5 sensor has better sensitivity to reducing gases. Moreover, The SC-5 sensor responds to 200 ppm of triethylamine up to 252.2, which is 10.9, 32, 40.6 and 117.2 times higher than formaldehyde, methanol, acetone and methane, respectively, which means that the SC-5 sensor has good selectivity for triethylamine. From the perspective of the

practical application of the sensor, long-term stability is also an important factor in evaluating the property of the sensor. Figure 5d displays the durable response of the SC-5 sensor to 200 ppm TEA after storing for 30 days. As shown by the curve, the response remains almost constant and remains at around 250. It can be concluded that the SC-5 sensor had better stability and can be a promising candidate for TEA sensor.

**Figure 5.** (**a**) Response of the sensors to different TEA concentrations at 310 ◦C (the inset shows the response curve in the range of 20–200 ppm), (**b**) dynamic response-recover curves of the sensors to different TEA concentrations at 310 ◦C, (**c**) responses of the SC-0 and SC-5 sensors to five gases of 200 ppm, (**d**) long-term stability measurements of the SC-5 sensor to 200 ppm TEA at 310 ◦C.

#### *2.3. Gas Sensing Mechanism*

As is known, the gas sensing mechanism of n-type MOS is based on the resistance change of the sensor by the adsorption and desorption reaction of oxygen molecules on the material surface with the gas to be detected [39]. When the undoped SnO2 sensor is exposed to the air, as shown in Figure 6a, the oxygen molecule (O2) are physically adsorbed on the surface of SnO2 material and then convert from physisorption to chemisorption. Chemisorption oxygen molecules capture electrons from the SnO2 conduction band to form oxygen anions O<sup>ܤ</sup>−) O<sup>−</sup>2, O<sup>−</sup>, <sup>O</sup>2−), which leads to the formation of electron depletion layer (EDL) on the surface of sensing material, the conduction channel in flower-like SnO2 is narrowed, the carrier concentration and conductivity are lowered and the resistance rises (Ra). When TEA vapor is injected, the oxygen anion O<sup>ܤ</sup>− formed on the surface of the SnO2 material reacts with the TEA and releases the trapped electrons back into the conduction band of the SnO2 sensing material. Consequence, the EDL becomes narrow, the conduction channel becomes wider and the conductivity of the sensor is enhanced, thus reducing the resistance of the sensor (Rg).

**Figure 6.** TEA sensing mechanisms diagram of (**a**) pure SnO2 and (**b**) CeO2/SnO2 nanostructure.

The gas sensitivity of the CeO2-SnO2 sensor be improved may be mainly due to two factors. Firstly, when CeO2 nanoparticles are supported on SnO2 sheets, the electrons with a low work function flow from SnO2 to CeO2 with a higher work function due to different work functions of SnO2 and CeO2. Thus, an n-n heterojunction is formed between the junctions of SnO2 and CeO2, which further provides the surface of the SnO2 sheets more active sites, so that the electron depletion layer of the composite undergoes a relatively large change depending on the atmosphere of the gas as shown in Figure 6b, thereby improving the gas sensitivity of SnO2. When the composite sensor exposed to the air, more oxygen molecules will capture electrons from the conduction band of composite nanomaterials, which reduces the carrier concentration and the thickness of the EDL is further increased compared with the pure SnO2 sensor, so the resistance of the CeO2-SnO2 sensor is greatly increased. When the composite sensor is exposed to TEA gas, more oxygen anions O<sup>ܤ</sup>− react with TEA and electrons are released back into the conduction band of the composite nanomaterials, concentration of free electrons in the conduction band increases and the electron depletion layer becomes thinner than the pure SnO2, resulting in a significant drop in sensor resistance. Thereby, the gas sensitivity of the sensor was improved. On the other hand, the electronic properties of CeO2 are also used to explain the sensitization mechanism [44]. Since CeO2 is a strong acceptor of electrons, it induces an electron depletion layer at the interface with the host semiconductor SnO2. By reacting with TEA, CeO2 is induced to release electrons back into the semiconductor conduction band. Especially on the surface of CeO2, the redox cycle of Ce4+/Ce3+ can be realized quickly and repeatedly under certain conditions, which makes oxygen vacancies easy to produce and diffuse [31,33]. Therefore, the doping of CeO2 can make the gas sensing material extract more oxygen more quickly, thereby enhancing the gas sensing performance. CeO2-SnO2 sensor has high selectivity for TEA in reducing gases such as methanol, formaldehyde, acetone and methane, which may be attributed to the electron-donating effect [4,45]. However, so far, there is no clear explanation for the sensitization mechanism, further research is still needed.

#### **3. Materials and Methods**

#### *3.1. Sample Preparation*

All of the chemical reagents was analytical grade (All of the chemical reagents was provided by Aladdin, Shanghai, China) and used without further purification in experiments, including Stannous chloride dihydrate (SnCl2·2H2O), Trisodium citrate dihydrate (Na3C6H5O7·2H2O), Sodium hydroxide (NaOH), Cerium (III) nitrate hexahydrate (Ce(NO3)3·6H2O) and absolute ethanol. Deionizer water was used throughout the experiments. In a typical process of synthesize CeO2-SnO2 composites, 3.6 g SnCl2·2H2O and 11.76 g Na3C6H5O7·2H2O were dissolved into 40 mL of distilled water with stirring. Subsequently, 0.182 g (3 wt.%) Ce(NO3)3·6H2O added into the above mixture with ultrasonic treatment for 20 min. Then 40 mL of NaOH solution was dropped into the above mixture under continuous magnetic stirring. Following, the above homogenous solution was transferred to a 100 mL of stainless autoclave lined with a Teflon vessel and heated to 180 ◦C for 12 h. The reaction system was then cooled naturally to room temperature after reaction. This pale yellow precipitate was collected by centrifugation and washed several times with distilled water and ethyl alcohol absolute and then dried at 60 ◦C for 12 h. Through varying the amount of Ce(NO3)3·6H2O in the synthesis process, the CeO2-SnO2 composites with 0 wt.%, 3 wt.%, 5 wt.% and 7 wt.% CeO2 decorated SnO2 were prepared and denoted as SC-0, SC-3, SC-5, SC-7, respectively.
