In order to evaluate the quenching effect of fluorescent nanofibrous membranes with different loading amounts of ZnS towards nitroaromatic explosive vapors, time-dependent fluorescence emission spectra were performed. As shown in
Figure 4, the fluorescence intensity of PU-1, PU-2 and PU-3 increase significantly with the increase of loaded ZnS before exposure to TNT vapor, reaching about 1180, 1900 and 3000. Because the loading amounts of ZnS in PU-4 is close to that of PU-2, the fluorescence intensity of PU-4 is close to that of PU-2, reaching about 1700 due to the coating of polymer layers. And one can see that from
Figure 4 and
Figure S5, the fluorescence quenching rate gradually decreases from PU-1 to PU-3 due to the increase of loaded ZnS. It is worth pointing out that although PU-4 does not have the most loads of ZnS, the fluorescence quenching rate of PU-4 is the slowest, which should be attributed to the coating of ZnS by the polymer layer. The coating of polymer layer effectively prevents the contact of ZnS and TNT vapor, causing the slow quenching rate. Based on the quenching performances of the above fluorescent membranes, PU-1 is selected as an ideal fluorescent sensing membrane for the detection of nitroaromatic explosive vapors. Furthermore, the reproducibility of fluorescent nanofibrous membranes were evaluated. The fluorescence quenching efficiencies (presented as (1–I/I
o), where I
o is the initial fluorescence intensity in the absence of analyte, I is the fluorescence intensity in the presence of analytes) of the above four kinds of membranes towards TNT vapor as a function of time among three batches were comparable (
Figure S5). The results indicated that the fluorescent nanofibrous membranes show satisfactory reproducibility for application to the detection of explosives.
In order to realize the differential quenching of fluorescent membranes towards explosives, PU-1 membranes are modified by lysine, cysteine, trifluoroacetyl lysine and cysteine hydrochloride, recorded as PU-1L, PU-1C, PU-1TL and PU-1CH, respectively. After modification, mercapto groups of cysteine and cysteine hydrochloride tightly attached onto the surface of the QDs due to the excess of metal ions with respect to sulfide ions at the surface of the QDs [
26]. Similarly, lysine and trifluoroacetyl lysine also tightly attached onto the surface of the QDs due to the interaction between carboxyl groups of lysine and trifluoroacetyl lysine and hydroxyl groups at the surface of the QDs [
27]. When fluorescent membranes are exposed to the explosive vapors, the ligand should bind the explosive, and an electron-transfer mechanism between the QD and the electron-deficient explosive causes QD fluorescence quenching.
Figure 5 shows the variation of the quenching percentage as a function of the exposure time to TNT, DNT, PA and NB. The response curves of fluorescent membranes towards different analytes are displayed in
Figure S6 and S7. As for PU-1C, after about 12 min, the quenching percentage is ~43, 38, 56 and 36% (
Figure 5) for saturated TNT (4 ppb) [
26], DNT (180 ppb [
28] or 200 ppb [
26]), PA (0.0077 ppb) [
26,
28] and NB (3 × 10
5 ppb [
28] or 4 × 10
5 ppb [
26]) vapors at 25 °C, respectively. Because the vapor pressures of TNT and DNT are about 520 and 2.3~2.6 × 10
4-fold that of PA, respectively, the quenching percentage for PA is thus surprisingly larger than that expected from the relative vapor pressure of these analytes. In terms of molecular structure (
Figure S8), PA is a stronger acid than TNT, and a stronger acid-base pairing interaction thus occurs between PA and amino ligands, resulting in the formation of PA anions at the surface of amine-capped ZnS QDs [
26]. However, it is well known that the DNT molecules with two nitro groups are much weaker Lewis acids and electron acceptors than TNT molecules. This suggests that it is less likely to form a Mesienheimer complex with the amine for DNT by the relatively weak basic amine groups. Therefore, the enhanced sensitivity toward PA vapor originates from the extremely strong adsorption of PA species at the amino of QDs and the larger quenching efficiency due to the high electron-accepting ability. Moreover, the high surface-to-volume ratio of QDs [
29,
30] and the good permeability of nanofiber membranes [
5,
31,
32] are further advantageous to the enhancement of the interaction between nitroaromatic explosive vapors and the amino ligands, hel** maximize the quenching efficiency. The fluorescent membranes of PU-1CH show the differential responses towards explosives owing to the different molecular structures of ligands, after 12 min the quenching percentage was ~60, 42, 47 and 36% for saturated TNT, DNT, PA and NB vapors at 25 °C, respectively. Lysine has two amino groups, but the response of PU-1L is not as we expected it to be. The quenching percentage towards TNT, DNT and PA vapors does not increase simultaneously, only reaching 36 and 45% for DNT and PA, respectively. Trifluoroacetyl lysine has an amino group, an amino group and three substituted fluorine atoms and the quenching percentage of PU-1TL towards saturated TNT, DNT, PA and BN vapors at 25 °C is ~45, 61, 37 and 39%, respectively. It is possible that the high quenching percentage towards DNT should be attributed to the polarity interaction of DNT and trifluoroacetyl lysine brought by the fluorine atoms. It is worth noting that although the vapor pressure of NB is much higher than TNT and PA, the fluorescence quenching efficiency is still lower than TNT and PA. This should be attributed to the weaker electron-withdrawing ability of BN with only one electron-withdrawing nitro group, compared with TNT and PA with three electron-withdrawing nitro groups. As a result, the differential quenching of fluorescent membranes towards nitroaromatic explosives is basically achieved by the surface modification of QDs, which lays a good foundation for the recognizable detection of nitroaromatic explosives.