3.1. Microstructure Characterization of PdOx/SnO2 Composites
The XRD patterns of SnO
2, PdCl
2/SnO
2, and PdO
x/SnO
2 composites were indicated in
Figure 2. The recognizable diffraction peaks can be assigned to the tetragonal SnO
2 (JCPDS 41-1445). There is no recognizable peak of PdO
x or other crystalline phases appeared, probably due to the low crystallization and small content of the additives.
The crystalline sizes were measured by means of an X-ray line-broadening method, using the Scherrer equation:
D is the crystalline size in nanometers; λ is the wavelength of the radiation (1.54056Å for CuKα radiation); k is 0.94; βhkl is the peak width at half-maximum intensity, and θ is the peak position.
It can be seen that the peaks of Pd doped SnO
2 NPs with different annealing temperature are wider than the peak of no annealing. The crystallite size calculated by the Scherrer method is decreasing with annealing temperature increasing. In 1991, Yamazoe demonstrated that a reduction in crystal size would significantly increase the sensor’s performance [
20].
The microstructure of PdO
x/SnO
2-450 composite was investigated by TEM and STEM-EDS characterizations. The TEM image of PdO
x/SnO
2-450 is shown in
Figure 3, where the size of nanoparticles is 50 to 70 nm and the shape is a sphere. The particle size is consistent with that of SnO
2. Furthermore, the rough surface morphology of PdO
x/SnO
2-450 composite is obviously different from the smooth surface morphology of SnO
2 nanoparticles, indicating that the PdO
x nanoparticles are homogenously dispersed on the surfaces of SnO
2 nanoparticles.
The corresponding elemental mapping of Sn, O, and Pd reveals that the PdOx nanoparticles are uniformly dispersed on the surfaces of SnO2 nanoparticles.
3.2. Chemical States of PdOx Dispersed on the Surfaces of SnO2
X-ray Photoelectron Spectroscope proves a very effective method in identifying the elements of the materials, chemical states and electronic states of the elements [
21].
In
Figure 4a–d, we plotted evolution of the Pd 3d
5/2 spectrum in the dependence on the different annealing temperature. The Pd1s spectra of the XPS results were analyzed with more details. During annealing treatment, three characteristic peaks Cl-Pd-Cl (BE = 337.3 eV), Pd-O (BE = 336.6 eV), and O-Pd-O (BE = 338.1 eV) are observed and these peaks indicate the changes of chemical state of palladium. The asymmetric peak was belonged to two components with BE of 337.3 eV and 336.6 eV, with spectrums that are typical of palladium in the oxidation state of Pd
2+ and Pd
4+. And increasing the temperature to 450 °C resulted in the peak Pd (BE = 337.3 eV) which corresponds to the Cl-Pd-Cl structure having disappeared in the spectra. When the temperature is to 600 °C, we also observed further changes that the binding energy (BE) of Pd was shifted to lower values for smaller particles, see
Figure 4a, and the intensity ratio of peak O-Pd-O to peak Pd-O increased gradually. According to
Table 1, it is interesting in the series of spectra in
Figure 4a–d that pure Pd
0 or Pd
n+ situations are never obtained.
For more understanding of the chemical states of Pd, the O 1s spectra of the XPS results were analyzed with more details in
Figure 4e–h. The asymmetric O 1s peak was deconvoluted into three components with binding energies of O-Sn-O (BE = 530.9 eV), Pd-O (BE = 531.3 eV) and O-Pd-O (BE = 532.3 eV) [
23]. As the temperature increases to 120 °C, two new peaks BE = 530.9 eV and BE = 531.3 eV of O-Sn-O, and Pd-O, appear respectively. As the temperature reaches 250 °C, a peak BE = 532.3 eV is observed to be the structure of O-Pd-O. When the temperature is in the range of 250 °C to 600 °C, the intensity ratio of peak O-Pd-O and Pd-O is gradually increased.
Figure 4 illustrates the process of Pd surface change on Pd-SnO
2. O
2 diffuses to the surface of Pd and adsorbs on the surface of Pd to form an active adsorption state. Then, the oxygen atoms and Pd atoms on the surface of Pd undergo an orderly remodeling to form PdO. As the temperature increases, the oxygen atoms accumulated on the surface further react with PdO to form PdO
2. The chemical state transition of palladium and the order in which these peaks appear are as follows:
Figure 5 shows the variation of the Sn 3d (a) binding energy (BE) of the PdCl
2/SnO
2 nanoparticles with the different annealing temperature. As shown in
Figure 4b, the two BEs have the same tendency, which indicates that the BE of Sn 3d
3/2 and O 1s decreases with increasing temperature, the value of Sn is 1.92 eV, the value of O is 1.82 eV, and the oxygen in SnO
2 loses electrons. As a result, the Fermi level shifts and bends upward, forming a depletion layer.
3.3. H2 Sensing Characteristics of PdOx/SnO2 Composites
To determine the optimum operating temperature, we tested the response of Pd-doped SnO
2 gas sensors at different operating temperatures. The response of the material to 100 ppm H
2 was tested at 25, 50, 80 and 100 °C and the results are shown in
Figure 6. As the operating temperature increases, the response of the Pd-doped SnO
2 first increases and reaches a maximum at 80 °C, whereas the temperature continues to increase, and the sensitivity is reduced. This is caused by the gas adsorption/desorption process on the sensor surface of Pd-doped SnO
2. When the working temperature rises to a certain temperature, the desorption rate of the gas increases, resulting in a decrease in sensitivity. At different annealing temperatures, there is a similar trend.
Normally, the reactivity between the target gas and adsorption oxygen needed certain activation energy, which was provided by increasing the work temperature. Since the Pd nanoparticles have a catalytic effect, the activation energy is lowered, thus the optimum working temperature is lowered to 80 °C. As the temperature increases, the diffusion rate of gas molecules increases, and it is easier to react with gas-sensitive materials. In addition, at high temperatures, due to a large decrease in adsorbed oxygen, the grain boundary barrier decreases and the conductance increases, so that the resistance change decreases when contacted with H2. The gas sensitivity performance is declining. In the experiments, the following sensing characteristic of Pd-doping SnO2 was tested at the optimal operating temperature (80 °C).
The sensor is annealed from 120 °C to 600 °C, showing a high corresponding sensitivity. The results of four reproducibility tests performed at the optimum operating temperature are shown in
Figure 7, indicating that the sensor has good repeatability and stability.
The response recovery time is also one of the important parameters for evaluating the gas sensor, as shown in
Figure 8. Gas sensing materials are often required to have fast response and recovery characteristics in applications. We can see that the response time is similar as 100 s at PdO
x/SnO
2-120, -250, -450, and the response time at PdO
x/SnO
2-600 is slower than the others. The recovery time is faster at higher temperature than those in lower temperature (from 120 °C to 450 °C). The recovery times were 340 s and 185 s for the sensor based on annealing temperature at 120 °C and 250 °C, respectively, and 140 s for the sensor based on annealing temperature at 450 °C. Moreover, the recovery time of annealing temperature at 600 °C is 290 s that is not better than lower temperature. It can be seen that when hydrogen is introduced, the resistance of the sensor is reduced and then the steady state is reached. The nanostructures provide a larger specific surface area that can speed up the adsorption and desorption processes, which shorten response and recovery time.
Figure 8 also demonstrates that the sensitivity of the PdO
x/SnO
2 at the different annealing temperature. The sensitivity of sensor at 450 °C is higher than the others. These features are produced by the influence of grain boundaries between the nanoparticles. In XRD, we know that the crystallite size is decreasing as the annealing temperature is increasing. By studying the effect of grain size on the sensitivity of the H
2 sensor, it is found that the larger the particle size of the nano-gas sensitive material, the lower the sensitivity.
Figure 9a shows that the response of PdO
x/SnO
2 to different concentrations of H
2 in air. The sensor chips were tested at an optimum temperature of 80 °C It can be found that the sensor show strong responses from 2.5 ppm to 100 ppm.
Figure 9 shows that as the H
2 concentration increases, the response sensitivity increases. When exposed to H
2, the mechanism is that hydrogen converted to atomic hydrogen at the material interface, causing the current signal to change.
The corresponding sensitivity can be obtained from the slope of the standard curve and the gas concentration. This result indicates that the sensitivity is related to the adsorption of H2 molecules on the surface of the gas sensing material.
The response stability of the PdO
x modified SnO
2 gas sensor to 100 ppm H
2 was tested at 80 °C with 15 cycles (
Figure 9c). As the number of times increases, the sensitivity decreases and gradually reaches a steady state. This may be due to the deposition of some contaminants on the sensor surface after the first few cycles. Obviously, Pd decorated SnO
2 has good stability and can be used in various practical applications.
We investigated the gas response characteristics of the sensor to different concentrations of PdO
x loaded SnO
2 at its optimum operating temperature. Sensors of 0.75 wt%, 1.25 wt%, 2.5 wt%, 5 wt%, and 7.5 wt% PdCl
2 supported SnO
2 were prepared and annealed at 450 °C. The results are shown in
Figure 10, where the change in sensitivity indicates that the corresponding performance is highly correlated with the Pd loading amount. We can see that the 5.0% by weight load sensor more sensitive to hydrogen than other concentrations of PdCl
2. Because of more PdO
x is on the surface of SnO
2, the stronger reaction captures electrons and generates an electron depletion layer on the surface of SnO
2. The higher loading of Pd leads to its full coverage on the surface of SnO
2, which reduces the contact between gas and SnO
2. The effect of PdO
x greatly exceeds the influence of SnO
2 bulk material, so that the performance of SnO
2 itself can’t be fully reflected, and therefore the sensitivity decreases.
Figure 11 shows the experimental results of the effect of humidity on the H
2 sensitivity. We have performed sensor response tests of Pd/SnO
2-450 under relative humidity from 20% to 70% conditions. The sensitivity changed significantly in different humid environments. It can be showed that Pd/SnO
2-450 has less sensitivity in low and high relative humidity at an operating temperature of 80 °C, while the corresponding sensitivity to 100 ppm H
2 is highest at 40% humidity. This is that water vapor increases the electron concentration of Pd/SnO
2-450, thereby increasing oxygen absorption. When the relative humidity is further increased, the sensitivity starts to decrease, which is caused by the synergistic competition between H
2 and water vapor.
The selectivity of the gas sensor is another important parameter for assessing the sensing capabilities of gas sensing materials. The gas selection performance of the material was obtained by testing the sensitivity of PdO
x-doped SnO
2 to 100 ppm different kinds of gases. The testing gases include hydrogen (H
2), ethanol (EtOH), carbon monoxide (CO) and methane (CH
4). As can be seen from
Figure 12, the Pd-doped SnO
2 gas sensor exhibits significantly higher response sensitivity to H
2 and a smaller change to other gases. Therefore, we believe that the gas sensor has good selectivity for H
2.
3.4. Sensing Mechanism
The sensing mechanism of semiconductor material sensors is based on semiconductor oxygen vacancy reactions. In air, due to the difference in temperature, molecular oxygen may exist in the form of O
2−, O
− or O
22−, etc., and oxygen forms an electron depletion layer by depleting electrons on the surface of the material to lower the electrical conductivity. When the palladium atom acts as a catalyst, its function is to dissociate oxygen on the surface of SnO
2, thus increasing the overflow of oxygen from the surface of SnO
2 to obtain a greater amount of electrons and thereby increasing sensor sensitivity [
24,
25,
26]. In addition, in order to increase the conductivity, the palladium atom can be divided into two processes. First, hydrogen molecules dissociate into hydrogen atoms on the surface of Pd, and then hydrogen atoms adsorbed at moderate temperatures act as donors, providing additional electrons to the material and inducing a cumulative layer.
With increasing the annealing temperature, different chemical states of Pd doped SnO
2 complex materials were obtained (
Figure 13). The Pd
2+ additive on SnO
2 substrate is reduced to a Pd metal, which plays an important sensitivity role in hydrogen and leads to change of conductivity. But partial PdCl
2 of the sensor on a low annealing temperature was hydrolyzed to Pd(OH)
2 in wet air and it is generation of [PdCl
4]
2− in different pH. From the potential of the reduction reaction, we know that Pd(OH)
2 and [PdCl
4]
2− are more difficult to reduce to catalytic Pd metal than Pd
2+. So the sensitivity of the sensors at a low annealing temperature is worse than the high annealing temperature.
Pd2+ + 2e→Pd | φ = 0.915 V. |
Pd(OH)2 + 2e→Pd + 2OH− | φ = 0.07 V |
[PdCl4]2− + 2e→Pd+ 4Cl− | φ = 0.621 V |
It is deduced that Pd4+ is the key factor to promote the sensitivity. Basing on the valence electron layer structure of platinum group metals, the stability of high-valent Pd(Ⅳ) is lower than Pd(Ⅱ). So the conversion of Pd4+ is easier. On the high annealing temperature, the proportion of PdO2 is increasing. This is one of the reasons why sensor chips at high annealing temperatures have better gas sensitivity. In addition, while increasing the annealing temperature, the sensitivity performance was stronger because of the unpreventable decrease of the grain sizes. In our study, when the annealing temperature is higher than 600 °C, the gas sensitivity is reduced in contrast. The detailed mechanism of different Pd chemical states doping SnO2 affecting the performance of the sensor still needs further investigation.