3.1. Gas Sensitive Material Properties
The morphology and chemical composition of the gas sensitive materials (i.e., PPy and Ag/PPy) were examined by TEM and XPS. TEM of the PPy NPs (
Figure 2a) revealed spherical shaped particles with an average size of 44 ± 10 nm, calculated for a population of 100 particles. XPS displayed characteristic C 1s (
Figure 2b) and N 1s (
Figure 2c) core level spectra. The components in the C 1s spectrum centred at 284.4, 285.0, 286.3, 288.1 and 288.9 eV are ascribed to the binding energies of the C–C bonds of β atoms, the C–C bonds of α atoms and the C–N, C–O and C=O bonds of PPy, respectively. Similarly, the three components in the N 1 s spectrum correspond to the =N– (398.0 eV), –N–H (400.0 eV) and =N–H
+ (402.3 eV) bonds of PPy. The presence of these components in the C 1s and N 1s core level spectra is consistent with our previous observations [
19] and corroborates the synthesis of PPy.
The synthesis of crystalline Ag NPs was also corroborated by HR-TEM.
Figure 2d displays the TEM images of the particles with mean diameter of approximately 17 ± 3 nm (calculated for a population of 30 particles) and lattice fringes with spacing of ~0.23 nm, consistent with the (111) planes of Ag face centered cubic phase (JCPDS number 04-0783) [
20]. High-resolution XPS analysis of the Ag NPs (
Figure 2e) revealed typical Ag 3d doublets separated by 6 eV. The deconvoluted Ag 3d doublet with a pair of components centered at 366.3 and 372.3 eV, and another pair centered at 367.4 and 373.2 eV indicates the presence of Ag
+ and Ag
0, respectively. This suggests the coexistence of silver oxide (Ag
2O) and metallic silver (Ag), in accordance with previous observations [
21].
Further SEM study of the PPy and Ag/PPy NPs guiding/sensitive layers after spin coating showed the integration of uniform sensing layers with a thickness of ~262 ± 10 nm on the L-SAW platforms. A typical cross section SEM image of the L-SAW sensors is displayed in
Figure 2f.
3.3. Gas Sensing Properties of the L-SAW Sensors
The sensing capability of the L-SAW sensors (PPy and Ag/PPy) was evaluated by exposing these devices simultaneously to various concentrations (between 0.5 and 5 ppm) of acetone, ethanol and toluene at room temperature. The calibration curves for acetone, ethanol and toluene are shown in
Figure 4a,c,e, respectively. These figures show the proportional increase of the sensor response with the increase of concentration for each gas without reaching the saturation point, which indicates the possibility to sense higher concentrations of these analytes. Generally, the results display enhanced responses for the Ag/PPy sensors compared to the PPy sensors for all tested gases, with frequency shifts of approximately 1.4 times more for the Ag/PPy sensors than for the PPy sensors. These results are in line with previous experimental research, which proved the enhancement of gas sensing properties (e.g., sensors response) by the modifications of a host gas sensitive material with metal catalysts (e.g., Ag, Au) due to a “spillover effect” [
14,
22].
The LOD of the Ag/PPy L-SAW sensors (defined as the concentration providing a signal-to-noise ratio of at least three [
23]) was estimated to be 3, 5 and 20 ppb for acetone, ethanol and toluene, respectively. These LOD are below the limits set for acetone, ethanol and toluene in different areas. For instance, the American Conference of Governmental Industrial Hygienists (ACGIH) sets the occupational threshold limit values (TLV) for acetone, ethanol and toluene to 250, 1000 and 20 ppm, respectively, considering 8-h time-weighted averages. Similarly, the concentration of VOCs in the food industry are above (typically by tens or hundreds of ppb) the LOD obtained for our sensors [
24]. Additionally, in the breath analysis field, the concentrations of acetone, ethanol and toluene in the exhaled breath (e.g., lung cancer patients register concentrations of 112–2654 ppb of acetone, 13–1520 ppb of ethanol and 9.3–21.3 ppb of toluene) [
4] are above the LOD obtained in this work.
The dynamic response of the PPy and Ag/PPy L-SAW sensors exposed to various concentrations of acetone, ethanol and toluene are presented in
Figure 4b,d,f, respectively. One can notice from these results that the Ag/PPy and PPy sensors displayed stable and reversible responses. In order to reach the steady state of the response and determine the response and recovery time, the sensors were exposed to a gas concentration of 5 ppm for 5 min and then the gas was purged with dry synthetic air for 10 min (see the inset responses in
Figure 4b,d,f). Overall, the response and recovery time for the Ag/PPy sensors was approximately 10 s faster than that for the PPy sensors. This could be attributed the catalytic properties of chemically active Ag NPs, which accelerate the gas-solid interactions at the guiding/sensitive film. These results also reveal faster response and recovery time to acetone than to ethanol or toluene for both the PPy and Ag/PPy sensors. The response time for the Ag/PPy sensors to acetone and ethanol, for instance, was below 2.5 min, while the response time to toluene was below 3.5 min. In addition, the Ag/PPy sensors required less than 4 min to recover after acetone exposure and less than 5.5 min after ethanol and toluene exposure.
Results in
Figure 4b,d,f also give evidence of the negative frequency shift of the response to acetone, ethanol and toluene. This decrease in the resonant frequency of the L-SAW sensors may indicate that the mass and acoustoelectric loading effects outweigh the elastic effect [
25]. The mass loading effect and therefore the decrease of the resonant frequency of the L-SAW sensors may be connected to the change in the mass of the PPy and Ag/PPy guiding/sensitive layer caused by the sorption of gas analytes (i.e., acetone, ethanol and toluene). The acoustoelectric effect, in contrast, may be connected to the adsorption of the gas analytes at the guiding/sensitive layer and the increase of the conductivity of PPy [
26], which is also characterized by a decrease in the resonant frequency of the sensor. The contribution of the elastic effect, which is generally observed by an increase in the resonant frequency, is ruled out in this particular case [
25].
As the Ag/PPy L-SAW sensors showed improved sensing properties over the PPy sensors, further analysis was performed on these sensors.
Figure 5 shows the sensitivity of the Ag/PPy sensors calculated as the ratio between the change in frequency response of the sensor and the change in the gas concentration. These results show the sensitivity to various gases, including acetone (910 Hz/ppm), ethanol (742 Hz/ppm), carbon monoxide (458 Hz/ppm), hydrogen (396 Hz/ppm) and toluene (340 Hz/ppm). Overall, the sensors displayed good stability keeping a constant operating frequency and reproducible responses with standard errors below 5% after testing all gases in dry ambient. One can notice in
Figure 5 that the Ag/PPy sensors exhibit a higher sensitivity to acetone than to other analytes such as ethanol and toluene, which register a cross-sensitivity respect to acetone of 81% and 37%, respectively. Similarly, the cross-sensitivity of other gases such as carbon monoxide and hydrogen with respect to acetone is found to be 50% and 43%, respectively. In summary, these results indicate a relatively low interference among the tested analytes.
The performance of the Ag/PPy L-SAW gas sensors investigated in this work were summarized and compared with other SAW gas sensors reported in the literature (
Table 1). Results indicate that our sensors exhibited significantly higher sensitivity as well as lower LOD to acetone, ethanol and toluene in comparison to other SAW systems. The lowest tested concentrations of those VOCs detected via Ag/PPy L-SAW sensors were crucially lower than the theoretical LOD of state-of-the art sensors (
Table 1). This demonstrates the capability of Ag/PPy based Love wave sensors to detect ppb levels of the target VOCs.
To evaluate the effect of humidity on the sensing response, the Ag/PPy sensors were tested to acetone at RH of 10% and 30%.
Figure 6 compares the frequency shift of the sensor to 5 ppm of acetone in dry and humid conditions with 10% and 30% RH. The tests show that the sensor response decreases by a factor of 2 and 7 when the atmosphere changes from dry to 10% and 30% RH, respectively. Humidity tests are generally not reported for state-of-the art SAW sensors; therefore, these characteristics have not been included in
Table 1. The loss of response in the Ag/PPy sensors in a humid atmosphere may be caused by the water vapor sorption into the polymer layer, which fills the free volume fraction in the polymer and reduces the gas permeability [
31]. Additionally, after the humidity test, we registered a decrease in the sensor response in dry ambient of about 35% compared to the responses obtained in the initial operation hours (notice that the same sensors were exposed to all target gases in a dry and humid environment accumulating an operation time of 100 h). The irreversible loss of response in the material may be caused by the humidity rather than the testing time, as during the tests in dry ambient the response registered low dispersion as described above. This is a common issue in polymer based gas sensors that needs further technological solutions such as the use of humidity filters or dehydration elements [
32] in order to exploit these sensors in future consumer devices.