3.1. Morphological, Structural, and Compositional Studies
Figures 1 (a), (c), and (e) are TEM views of the mesoporous CuOx samples annealed at 250, 350, and 450 °C, respectively. High-magnification TEM images
of the aforementioned samples are presented in Figures 1 (b), (d), and (f), respectively. In general, the particle size of the samples annealed at higher temperatures
was slightly larger. For example, for the sample annealed at 250 °C, the approximate
size of the NPs was 7–9 nm, while the sizes of the samples annealed at 350 and 450
°C were 10–12 nm and 12–14 nm, respectively. Figures 2 (a)–(c) show HR-TEM views of the mesoporous CuOx powders annealed at 250, 350, and 450 °C, respectively.
For all the samples, the spacings between the parallel fringes were 0.252, 0.232,
and 0.274 nm, which matched the spacings of the (002), (111), and (110) crystalline
planes of the CuO phase, respectively[37-39].
Figures 3 (a)–(c) show the TEM–EDS elemental mapping of the mesoporous CuOx annealed at 250 °C. Both oxygen and Cu elements are uniformly distributed throughout
the samples. The corresponding TEM-EDS spectrum, illustrated in Figure 3 (d), shows peaks related to both O and Cu. The proportions of these elements were 11.22
and 88.78 wt%, respectively. Figures 3 (e)–(g) present the uniform distribution of O and Cu, as shown in the TEM-EDS analysis of
the sample annealed at 350 °C. The corresponding TEM-EDS spectrum of this sample shows
peaks for O (9.04 wt%) and Cu (90.96 wt%), as observed in Figure 3 (h). Figures 3 (i)–(k) present the TEM-EDS elemental mapping analysis of the mesoporous CuOx sample annealed at 450 °C, highlighting the even distribution of both O and Cu. In
addition, Figure 3 (l) shows the TEM-EDS spectrum, where peaks corresponding to O (7.92 wt%) and Cu (92.08
wt%) are present again. Overall, no impurity elements were observed in any of the
samples, which reflects the high purity of the materials and the effectiveness of
the synthesis procedure. In addition, the compositional analyses of all the samples
yielded similar results, again reinforcing the suitability of the employed synthesis
route.
Figure 4 (a) shows the XRD patterns of the mesoporous CuOx NPs at various temperatures. All the patterns have peaks that can be indexed to the
monoclinic phase of CuO, matching JCPDS File No. 80-1268 [40]. However, with increasing annealing temperature, the intensity of the peaks increased,
reflecting the improved crystallinity of CuOx at higher temperatures. Figure 4 (b) discloses the FTIR spectra of the mesoporous CuOx samples. For the sample annealed at 250 °C, the oxalate peaks at 1610, 1360, and
1317 cm-1 corresponding to surfactant–precursor interactions [36] almost disappeared, as did the nitrate ion (from copper nitrate precursor) peak
at 1418 cm-1 [36]. In fact, for the samples annealed at higher temperatures, no peaks related to surface
groups were observed; this indicates that all organic materials on the surface of
the samples were burned during annealing. Figure 4 (c) presents the N2 adsorption–desorption curves of the three mesoporous CuOx samples, and their calculated BET surfaces are compared in Figure 4 (d). For the CuOx annealed at 250, 350, and 450 °C, the BET surface areas were calculated to be 161.85,
39.88, and 22.52 m2/g, respectively. Therefore, the surface area decreased with increasing annealing
temperature because of the growth of particles at higher temperatures.
The deconvoluted XPS Cu 2p core-level spectra of the mesoporous CuOx annealed at 250, 350, and 450 °C are presented in Figures 5(a), (c), and (e), respectively. In all cases, peaks related to Cu2O and CuO appeared in the XPS spectra. For the sample annealed at 250 °C, the main
peak was related to Cu2O, whereas with increasing annealing temperature, the CuO peaks became dominant. Because
XPS can be used to detect elements at a depth of 10 nm, both Cu2O and CuO phases appear to have been present on the surface of samples. However, the
sample annealed at 250 °C contained more Cu2O than CuO, whereas the other two samples contained more CuO than Cu2O. The deconvoluted O 1s core-level spectra of the CuOx annealed at 250, 350, and 450 °C are presented in Figures 5 (b), (d), and (f), respectively. In all cases, the O 1s peak can be fitted to three separate peaks
at binding energies of 529.6, 531.3, and 533 eV, which are considered to be related
to lattice oxygen, adsorbed oxygen species, and oxygen vacancies, respectively [41].
3.2. Gas-Sensing Studies
For NO2 gas (2–10 ppm) at 200 °C, Figures 6 (a), (c), and (d) display the transient resistance graphs of the gas sensors fabricated using the CuOx samples annealed at 250, 350, and 450 °C. The sensors annealed at 350 and 450 °C
showed no response to the NO2 gas.
However, the sensor annealed at 250 °C could detect the NO2 gas; as shown in Figure 6 (b), its response values (R) for 2, 4, 6, 8, and 10 ppm of NO2 gas were 1.101, 1.219, 1.31, 1.354, and 1.907, respectively. In addition, none of
the sensors exhibited a meaningful response to H2S gas (Figure 7).
Therefore, considering that the sensor annealed at 250 °C could detect NO2 gas while exhibiting no response to H2S gas, it achieved high selectivity to NO2 gas in the presence of H2S.
3.3. Gas-Sensing Mechanism
When gas sensors are exposed to air, oxygen is initially adsorbed on their surfaces.
Owing to its high electron affinity, the oxygen then attracts the electrons from the
conduction band of the sensing layer. The related reactions are as follows [42]:
Therefore, a hole accumulation layer (HAL) was initially formed on the surface of
the mesoporous CuOx NPs constituting the sensors fabricated in this study. After subsequent exposure
to NO2 gas, the NPs were oxidized further due to the powerful oxidizing nature of the gas
[43]. In fact, additional electrons were transferred from the sensor to the NO2 gas; hence, the width of the HAL increased, leading to a reduction in the sensor
resistance. The relevant equations are as follows [44]:
The sensor fabricated using the mesoporous CuOx annealed at 250 °C showed the strongest response to the NO2 gas. Because the operating temperatures of all the gas sensors were identical, this
factor can be excluded when comparing the behaviors of the three sensors. Thus, the
enhanced response of the sensor annealed at 250 °C is assumed to be related to its
higher surface area (161.85 m2/g) relative to the other two sensors. Because a higher surface area affords more
adsorption sites, a larger amount of gas can be adsorbed on the sensor surface.
In addition, based on the XPS results, the Cu2O and CuO phases co-existed in the annealed samples. Therefore, in regions where these
phases were in close contact, the electrons likely moved from CuO, which has a work
function of 5.3 eV [45], to Cu2O, which has a work function of 4.6 eV [45], to equate the Fermi levels on both sides of the phase boundary. In an air atmosphere,
this causes band bending and the formation of potential barriers at the phase boundary.
In a NO2 atmosphere, more electrons are captured by the NO2 gas, leading to a change in the height of the potential barriers. This finally leads
to modulation of the sensor resistance and contribute to the sensing signal.