Characterization
A schematic diagram of the synthesis of Cu-SnO2/rGO is illustrated in Fig. 1. Firstly, the weak reducing property of L-ascorbic acid was utilized to reduce GO. Secondly, a mixed solvent consisting of oleylamine and oleic acid was employed as a surfactant to control the growth of SnO2 nanoparticles, Cu2+ and as-synthesized rGO were incorporated as dopants to react together. Finally, Cu-SnO2/rGO nanocomposites were prepared via high temperature annealing (400°C).
In Fig. 2a-c and Fig. S1, TEM and SEM images illustrate the morphology of pristine SnO2 CQDs, Cu-SnO2-2, and Cu-SnO2/rGO-2 (Fig. 2a-c), all of which are nanoparticles randomly shaped. The synthetic rGO is a lamellar structure with folds. The nanocrystal sizes of these three samples can be obtained by measuring the number of particles via the software Nano Measure. Thus, their average diameters of them are estimated to be 6.1, 6.2, and 5.7 nm, respectively. A smaller size of Cu-SnO2/rGO-2 may be attributed to the planar hydrophilic edges of rGO acting as surface ligands bound to Sn4+, resulting in the nucleation and growth of SnO2 nanocrystals limited21. It should be noticed that Cu-SnO2 grain size is slightly expanded compared to pristine SnO2 CQDs owing to the ionic radii of Cu 2p (0.87 Å) being larger than that of Sn 4b (0.83 Å) as Sn4+ is replaced by Cu2 + 22. HRTEM images of pristine SnO2 CQDs, Cu-SnO2-2, and Cu-SnO2/rGO-2 exhibit a high degree of crystallinity in Fig. 2d-f. The plane distances of (110) and (101) are estimated to be around 0.334 and 0.266 nm. As shown in Fig. 2g, the four well-defined diffraction rings of the selected area electron diffraction (SAED) pattern shown in each inset correspond to the (110), (101), (211) and (112) planes, respectively, confirming the tetragonal rutile structure of SnO223. EDS image of Cu-SnO2/rGO reveals that the nanocomposites are doped with copper elements at an atomic ratio of ~ 1%, as shown in Fig. 2h.
The three strongest peaks of XRD patterns of the three samples in Fig. 3a situated at 26.6°, 34.0°, and 51.8° (2θ) correspond to the (110), (101), and (211) crystallographic facets of tetragonal rutile SnO2 structure (JCPDS No. 41-1445), respectively24. No additional features associated with CuO/Cu2O, meanwhile, are observed. Nevertheless, as the amount of Cu doping increases, the positions of the (100) and (101) peaks of the samples shift towards lower 2θ values as shown in Fig. S1a. This indicates that Cu2+ replaces part of Sn4+ into the SnO2 lattice to form a solid solution phase, in agreement with the XPS results25. For the Cu-SnO2/rGO nanocomposites, no rGO diffraction peaks are observed (Fig. S1b), probably owing to the low rGO doping and relatively weak peak intensity so that the rGO cannot change the lattice structure, consistent with the HRTEM and SAED results. In addition, the characteristic peak intensity of SnO2 gradually increases with rGO doping rising, which indicates a continuous increase in SnO2 crystallinity. No (002) Bragg peak of rGO was observed.
Raman spectra are utilized to illustrate the reduction of GO and the synthesis of Cu-SnO2/rGO. As show in Fig. 3b, the Raman peaks of pristine SnO2 CQDs and Cu-SnO2/rGO-2 at 474, 632, and 778 cm− 1 correspond to the vibrational models of Eg, A1g, and B2g of tetragonal rutile SnO2, respectively. GO, rGO, and Cu-SnO2/rGO-2 have two intense peaks related to the typical graphene characteristic peaks situated at 1351 and 1589 cm− 1 corresponding to the D and G bands26. The intensity ratio of D to G bands (ID/IG) related to the number of functional groups in rGO are 1.01, 1.21, and 1.26 for GO, rGO, and Cu-SnO2/rGO-2, respectively. A higher value testifies to the partial modification of surface oxygen-containing functional groups.
XPS survey spectrum of Cu-SnO2/rGO-2 indicates the appearance of Sn, Cu, O, and C peaks without other impurity peaks as shown in Fig. 3c. The C 1s peak at 284.8 eV can be attributed to the surface critical carbon of the XPS instrument. The spectra of pristine SnO2 CQDs only confirmed the presence of Sn, O, and C which can be recognized Cu is successfully doped. The binding energies of 495.67 and 487.27 eV in Fig. 3d correspond to the Sn 3d3/2 and Sn 3d5/2 of Sn4+, indicating the formation of SnO227. The excellent symmetric shape of these peaks excludes the existence of metal tin. The disparity in binding energy (0.16 eV) is probably because of the Sn-O interaction after doping with Cu. The high-resolution spectra of O 1s peaks of the three samples (Fig. 3e) can be split into three Gaussian peaks, attributed to the three chemical states of O. The O 1s peaks adjacent to 530.5 eV are attributed to the O2− in the SnO2 crystal lattice named as lattice oxygen (Olat)28. The peak nearby 531.0 eV is Ov concerned with the oxygen-related vacancies in SnO2 crystallographic structures29. The peak at around 532.1 eV is related to Oads and can be described as oxygen specie(s) adsorbed by the materials30. These three types of oxygen species are of great importance in gas sensing properties and will be investigated separately in the Sensing Mechanism Section. As for Cu-SnO2-2 and Cu-SnO2/rGO-2, the high-resolution spectra of Cu 2p shows four peaks (Fig. 3f). The bending energies of 952.5 and 933.2 eV fall into Cu 2p1/2 and Cu 2p3/2, respectively, indicating the presence of Cu2+ and Cu+ states. The two shaking satellite peaks of Cu 2p nearby 962.3 and 942.9 eV correspond to the CuO phase, respectively14.
N2 sorption isotherms of pristine SnO2 CQDs, Cu-SnO2-2, and Cu-SnO2/rGO-2 are presented in Fig. 3g-i. The distinct hysteresis loops of the three samples indicate the presence of mesopores31. The doping of rGO endows Cu-SnO2/rGO-2 with larger average pore size (13.1 nm) and increased Brunauer-Emmett-Teller (BET) surface area (90.7 m2g− 1), compared with pristine SnO2 CQDs (11.3 nm; 85.4 m2g− 1) and Cu-SnO2-2 (11.6 nm; 82.1 m2g− 1). The larger average pore size facilitates the transport of H2S molecules between the ex- and internal regions to swift response/recovery even at low temperatures; meanwhile, the higher BET surface area contributes to more gas absorption and active sites. Noticeably, the Cu-doped SnO2 has a slightly lower specific surface area than the pristine SnO2 CQDs probably for the originally doped Cu occupies some channels of the SnO2.
Gas-sensing performance
The sensing performance of pristine SnO2 CQDs, Cu-SnO2, and Cu-SnO2/rGO was systematically evaluated. The real-time resistance was monitored to identify the optimal operating temperature by exposing the SnO2-based sensors doped different concentrations of Cu and rGO to 10 ppm H2S at a different temperature, as shown in Fig. 4a. The sensing response (Ra/Rg) toward 10 ppm H2S for pristine SnO2 CQDs witnessed a gradual rise as temperature arose, and the highest response was attained up to 4.4 at 280℃. With various Cu doping, the response values all peaked at the identical temperature (160℃), and it reached the highest of 1636.8 for Cu-SnO2-2. Furthermore, the dopant of rGO further reduced the operating temperature down to 120℃, and there was a peak in sensing response of 1415.7 for Cu-SnO2/rGO-2. This phenomenon may be attributed to the rGO and Cu-SnO2 forming a p-n heterojunction, thereby reducing the activation energy required for the chemical reaction between the semiconductor and the gas molecules. Hereafter, 280, 160, and 120℃ were chosen as the optimum operating temperature to evaluate the sensing properties of pristine SnO2 CQDs, Cu-SnO2-2, and Cu-SnO2/rGO-2 toward H2S, respectively. Cross-response of different gases has been a widespread barrier for MOS sensors. To assess the gas selectivity of the pristine SnO2 CQDs, Cu-SnO2-2, and Cu-SnO2/rGO-2, these sensors were measured at their operating temperatures toward 50 ppm of HCHO and C7H8 (toluene) and 100 ppm of H2, C4H10 (n-butane), and CO. As shown in Fig. 4b, the responses of these sensors to the above gases were all less than 2, much lower than the response values to H2S; meanwhile, the selectivity of Cu-SnO2-2 and Cu-SnO2/rGO-2 was significantly higher than that of pristine SnO2 CQDs. The dynamic response-recovery transient of pristine SnO2 CQDs (in blue), Cu-SnO2-2 (in green), and Cu-SnO2/rGO-2 (in orange) regarding H2S exposure/release cycles of different concentrations (1, 2, 5, 10, 15, 20, 30, 50, 100, and 150 ppm) in Fig. 4c-e. It can be seen that the response of the sensors rises sharply with the increase in H2S concentration. The sensor of Cu-SnO2/rGO-2 whose slope of the linear fit is around 156.5 ppm− 1 shows enhanced sensitivity of over 1900 times in comparison with pristine SnO2 CQDs (Fig. 4f). Meanwhile, Cu-SnO2/rGO-2 sensor features a linear response (R2 = 0.991) compared with Cu-SnO2-2 (R2 = 0.921), which tends to saturate value at a relatively large concentration (50 ppm). In summary, the Cu-SnO2/rGO-based sensor presents superb sensing performance compared with pristine SnO2 CQDs and Cu-SnO2-2 in operating temperature, linearity, and selectivity.
Figure 5 presents the response/recovery curves of the three sensors to 2 ppm H2S. The Cu-SnO2/rGO-based sensor not only works at the lowest temperature but also exhibits the shortest tres (31 s), which can be attributed to the larger specific surface area and the heterojunction between Cu-SnO2 and rGO that reduces the activation energy of gas sensing and thus accelerates the reaction between H2S and chemisorbed oxygen. In addition, the baseline resistance of Cu-SnO2/rGO-2 increases compared to Cu-SnO2-2 and pristine SnO2 CQDs. This may be owed to the heterojunctions between the ternary heterojunction further promoting the adsorption and decomposition of O2, forming a higher concentration of chemisorbed oxygen on its surface, resulting in an increase in the thickness of the electric depletion layer.
Figure 6a illustrates that the sensor based on Cu-SnO2/rGO-2 attains an average of 1.26 for three sequential responses to 50 ppb H2S. Hence, we conclude that the LOD of this sensor is less than 50 ppb. As shown in Fig. 6b, the response of the sensor to 20 ppm H2S presents similar transients and all the resistance can be recovered to the initial value among consecutive four cycles, confirming its outstanding repeatability. Figure 6c displays the effect of the Cu-SnO2/rGO-2 sensor as the relative humidity varies from 55 to 90%. It is found that the response of the sensor changes slightly as the ambient humidity gain, which substantiates that it is minimally affected by humidity. In addition, the sensitivity of the sensor based on Cu-SnO2/rGO-2 turns out almost constant after 28 days as demonstrated in Fig. 6d, indicating its good-term stability. We have summarized the performance of the Cu-SnO2/rGO sensor and compared it with other SnO2-based H2S sensors reported in recent literature in Table S1 for the sake of easier comparison. The Cu-SnO2/rGO-2 sensor prepared in this work shows high sensitivity, low detection limits, and fast recovery at relatively low operating temperatures, which indicates that the prepared Cu-SnO2/rGO-2 sensor has a broad development prospect and application in H2S detection.
Sensing mechanism
The sensing mechanism of the SnO2 (an N-type MOS) sensor concerns gas adsorption and surface-related redox reactions32. A schematic of the gas sensing mechanism is illustrated in Fig. 7a. For the as-synthesized Cu-SnO2/rGO in this work, in an ambient atmosphere, oxygen molecules are adsorbed on the surface of the material and ionized into oxygen-negative ions by trapping electrons in the conduction band. As per the test conditions (120°C), the surface oxygen species are primarily O2− and O− 33. The loss of electrons causes the forming of an electron depletion layer in the surface region, while a potential barrier builds between adjacent grains impeding the flow of electrons at the grain boundaries, which manifests itself macroscopically as an increase in resistance. When the sensor is exposed to H2S, oxygen ions react with H2S and deliver electrons to the Cu-SnO2/rGO surface. As a result, the electron depletion layer narrows, the barrier between adjacent grains is reduced and the resistance decreases. The exact reactions are shown below34.
The sensor based on Cu-SnO2/rGO operates at the lowest temperature and has enhanced sensitivity in comparison to pristine SnO2 CQDs and Cu-SnO2 for the following reasons. First of all, Cu-SnO2/rGO forms a special P-N-P ternary heterostructure. The energy band structure is shown in Fig. 7b. In the atmosphere, electrons transfer from SnO2 to rGO and CuO, whereas holes transfer in the opposite direction until the Fermi energy level reaches equilibrium. The electron depletion layer at the heterojunction interface is hence further expanded compared to pristine SnO2, corresponding to an increase in baseline resistance. In H2S, H2S reacts with the surface negative oxygen species and electrons enter the conduction band of SnO2. At the same time, some electrons will also enter the conduction band of rGO and CuO, which manifests itself at the macroscopic level as an enhanced conductivity. In addition, the presence of heterojunctions also contributes to the catalytic activity by providing more adsorption reaction sites35. Secondly, according to the three oxygen species occupancy ratios derived from the O 1s spectra (Table S2), the Cu-SnO2/rGO surface has increased adsorbed oxygen and oxygen vacancies. The abundant chemisorbed oxygen promotes the adsorption and reaction of the reduced gas; the oxygen vacancies contribute to the increasing charge density near the valence and conduction bands, narrowing the band gap of SnO2 (Fig. S2) and facilitating the adsorption and activation of the target gas. Moreover, the mesoporous structure provides an effective diffusion channel for the gas, and the larger BET surface area provides more active sites for foreign oxygen molecules, which enables gaseous molecules’ penetration into and interaction with interior grains. Last but not least, Cu-SnO2/rGO grain size (5.7 nm) is the smallest and close to 2LD (3 nm at 120℃), which maximizes the effect of varying electron depletion layer thickness on its overall resistance.