Highly sensitive gas-phase explosive detection by luminescent microporous polymer networks

We propose microporous networks (MPNs) of a light emitting spiro-carbazole based polymer (PSpCz) as luminescent sensor for nitro-aromatic compounds. The MPNs used in this study can be easily synthesized on arbitrarily sized/shaped substrates by simple and low-cost electrochemical deposition. The resulting MPN afford an extremely high specific surface area of 1300 m2/g, more than three orders of magnitude higher than that of the thin films of the respective monomer. We demonstrate, that the luminescence of PSpCz is selectively quenched by nitro-aromatic analytes, e.g. nitrobenzene, 2,4-DNT and TNT. In striking contrast to a control sample based on non-porous spiro-carbazole, which does not show any luminescence quenching upon exposure to TNT at levels of 3 ppm and below, the microporous PSpCz shows a clearly detectable response even at TNT concentrations as low as 5 ppb, clearly demonstrating the advantage of microporous films as luminescent sensors for traces of explosive analytes. This level states the vapor pressure of TNT at room temperature.

Scientific RepoRts | 6:29118 | DOI: 10.1038/srep29118 of the monomer SpCz (0.88 m 2 /g). We demonstrate, that the luminescence of PSpCz is selectively quenched by nitro-aromatic analytes. In striking contrast to a control sample based on thermally evaporated (non-porous) SpCz, which does not show any luminescence quenching upon exposure to TNT at levels of 3 ppm and below, the microporous PSpCz shows a clearly detectable response even at TNT concentrations as low as 5 ppb. Figure 1a shows a schematic of the electrochemical polymerization of SpCz monomers to microporous films of PSpCz. The substrate is ITO coated glass which forms the working electrode. Further details of the electrochemical polymerization process can be found in the experimental section and in a previous report 10 . Please note, in a comparative study published earlier, PSpCz has shown the highest BET surface area (1300 m 2 /g) compared to other microporous carbazole-based compounds prepared in a similar fashion. The photo-luminescence (PL) spectrum of PSpCz in comparison to that of a thermally evaporated SpCz thin film is shown in Fig. 1b. As expected, the PL maximum of the polymerized network (λ = 472 nm) is substantially red-shifted compared to the monomer film (λ = 407 nm). A similar red-shift has been observed for other microporous polymer structures with respect to films of their corresponding monomers 12,15 . The absorption spectrum of the monomer SpCz shows a sharp onset at 378 nm (3.28 eV), while in the polymer the main absorption edge can be found red-shifted at 420 nm (2.95 eV) ( Figure S1). This red-shift is in agreement with the shift found in the PL spectra of SpCz and PSpCz. Moreover, the absorption spectrum of PSpCz exhibits a pronounced low energy tail extending up to about 550 nm, pointing to some significant disorder in the polymerized material. The HOMO levels for both PSpCz and SpCz have been determined by atmospheric pressure photo-electron spectroscopy to HOMO SpCz = − 5.83 eV and HOMO PSpCz = − 5.58 eV ( Figure S2). The HOMO level of the monomer SpCz is in agreement with previous reports for the HOMO levels of similar carbazole compounds, i.e. − 5.63… − 6.2 eV for 4,4′ -Bis(N-carbazolyl)-1,1-biphenyl (CBP) [16][17][18] or − 5.82 eV for 4,4′ ,4″ -tris(N-carbazolyl)triphenylamine (TCTA) 19 . Note, the cross linking in the PSpCz occurs via its carbazole units. The difference in the HOMO levels of the monomer and the polymerized film is therefore reasonable, as the HOMO of many carbazole-based compounds, e.g. CBP, has been shown to be predominantly governed by the carbazole units 17 . For an application as luminescence sensor the optical gap of the respective material is of paramount importance. It has been shown that the bandgap of organic materials determined by photo-emission may substantially deviate from the so-called optical gap (E opt ), which is relevant for the absorption/emission of photons 20,21 . E opt can be estimated from the short-wavelength onset of the PL spectrum. From the PL spectra of SpCz and PSpCz (Fig. 1b), E opt = 3.4 eV and E opt = 3.0 eV is derived, respectively. Thus, an estimate of the LUMO position based on the data for E opt results in positions of the LUMO levels as LUMO SpCz = − 2.43 eV and LUMO PSpCz = − 2.58 eV. This is somewhat deeper than the LUMO levels of CBP and TCTA, which have been determined by inverse photo emission or electrochemical methods to be on the order of − 2.2 eV [16][17][18][19] .

Results and Discussion
AFM images of nominally 50 nm thick films of thermally evaporated SpCz and electrochemically deposited microporous PSpCz are shown in the supporting information ( Figure S3). The roughness is 2.6 nm (rms) for the SpCz layer and it increases significantly to 14 nm (rms) for the microporous PSpCz. In an attempt to estimate the pore size in the MPN films, we assumed a model substance with spherical pores, which for simplicity are arranged in a regular simple cubic structure ( Figure S4). With a BET surface of 1300 m 2 /g and an assumption of a bulk mass density of the non-porous polymer of 1.2 g/cm 3 , a typical density for a wide range of organic thin films 22 , an upper limit for the pore diameter of 4 nm can be derived. Given the simplicity of the estimate, this result is in reasonable agreement with previous reports of pore diameters on the order of 2 nm for carbazole based electropolymerized MPNs 12,23 .
In order to assess the sensor properties of the microporous PSpCz layers, they have been exposed to vapors of various analytes. For this purpose a setup as that shown in Fig. 2a has been used. Specifically, the analyte is positioned in a separate chamber (analyte chamber), which is connected to the chamber where the PSpCz coated substrate is located (sample chamber). The entire system can be heated. Details about the measurement setup can be found in the experimental section. In an orienting experiment, which was aimed to check for the selectivity of the PSpCz as luminscence sensor, we used the analytes shown in Fig. 2b. As an example, Fig. 2d shows the decay of the PL spectrum of PSpCz upon exposure to 100 ppm of The quenching efficiency of PSpCz upon exposure to 100 ppm of various analytes is shown in Fig. 2e. Please note the logarithmic representation of the vertical axis. Obviously, a significant quenching efficiency > 1 is found in the case of electron deficient nitro-aromatic analytes (NB, DNT) with a LUMO level below that of PSpCz (LUMO PSpCz = − 2.58 eV), while for the other analytes (with significantly higher LUMO levels) only a negligible effect on the PL is observed.
The mechanism of PL quenching, which is schematically shown in Fig. 2c, is therefore in line with an electron transfer quenching mechanism proposed earlier 14 . As a result, the higher quenching efficiency of DNT compared to NB can be attributed to the significantly deeper LUMO level of DNT compared to NB, providing more driving force for the electron transfer from the LUMO of PSpCz to the LUMO of DNT. Note, as we were not able to provide 100 ppm of TNT due to its low vapour pressure, we did not include TNT in this comparison. The response of PSpCz to TNT will be discussed in detail below. As shown in Figure S5, the PL quenching is reversible by removing the analyte from the MPN. Please note, most gaseous species which are potentially present in our atmosphere, e.g. O 2 (E A = 0.45 eV), O 3 (2.1 eV), NO 2 (2.27 eV), SO (1.12 eV), SO 2 (1.1 eV), H 2 O (0.9-1.3 eV) etc., show electron affinities in the range of 0.5-2.3 eV 25 , which would not be expected to lead to an electron transfer from the LUMO of the PSpCz (E A = 2.58 eV).
In Fig. 3 the quenching efficiency upon variation of the analyte concentration of nitrobenzene is studied. The quenched PL spectra are shown in Fig. 3a. The resulting quenching efficiency (I 0 /I − 1) for a concentration varied over several orders of magnitude is shown in Fig. 3b. Here, we have either used the integrated intensity of the full PL spectra or of a selected spectral region (455-480 nm) around the spectral maximum. Obviously, a somewhat higher quenching efficiency is determined if only the spectral region around the PL maximum is considered. This result may be explained by the fact that excitons belonging to the low-energy tail of the PSpCz spectrum (λ > 550 nm ⇔ hν < 2.25 eV) may not experience enough driving force for dissociation by electron transfer from the PSpCz to the LUMO of the NB located at − 2.915 eV. Note, the energy difference between HOMO PSpCz = − 5.58 eV and that of a photon with hν = 2.25 eV is about − 3.3 eV, which would be below the LUMO of nitrobenzene. Thus, for a transfer from PSpCz to NB, electrons in this case would have to go uphill in energy. As a result, the PL spectra in the thoroughly quenched state of the PSpCz show a residual spectral feature in the low-energy region. Note, some significant signature of low energy states can be seen in the absorption spectrum of PSpCz ( Figure S1).
Interestingly, as shown in the log-log plot of Fig. 3b, in the regime of low concentrations of nitrobenzene ([NB]), a non-linear behaviour for the quenching efficiency is found with (I 0 /I − 1) ∼ [NB] x , where x varies between 1.8-2.3. It has to be noted that in the ideal Stern-Volmer model (I 0 /I − 1) is proportional to the concentration of the quencher (i.e. x = 1). The prerequisites for the linearity are the equal accessibility of the quenching molecules to all luminescent parts of the sensor and the prevalence of only one quenching mechanism. The origin of the non-linearity in our case is not fully clarified, and has to be the subject of further studies. In general, non-linear effects have been encountered in case of a mix of static and dynamic interaction of chromophore and analyte 26 .
At elevated concentrations of the analyte ([NB] > 100 ppm), the quenching efficiency levels off and saturates in a range of 10-100. Again the evaluation of the spectral region around the PL maximum shows a higher saturation quenching efficiency, possibly due to the reasons discussed above. The origin of this saturation behavior could be NB and DNT are both molecules with a relatively high vapor pressure (0.25 mbar and 1.9 × 10 −4 mbar) even at room temperature 27,28 . Opposed to that, TNT owing to is low vapor pressure is substantially more challenging to detect in the gas phase. To explore the detection limits of our PSpCz MPN layers for nitro-aromatic vapors, we studied its response to TNT.
For the control of the vapor pressure of the TNT analyte, we have used the heating capability of our measurement setup, as shown in Fig. 2a. The relation of TNT vapor pressure and temperature has been estimated according to the Clausius-Clapeyron-equation with parameters as reported in the literature, i.e. log 10 (P) = A − B/T, where P is the pressure in Torr and T is the temperature in Kelvin (A = 14.74, B = 5960) 28 .
For comparison, we have studied the PL response of a SpCz thin film prepared by thermal evaporation and that of a microporous PSpCz layer. Both were nominally 50 nm thick, but while the thermally evaporated layer is rather dense with a BET surface area of only 0.88 m 2 /g, the microporous network in the PSpCz affords a surface area of 1300 m 2 /g. Therefore, we assume that the interaction of TNT and SpCz happens predominantly at the surface of the SpCz layer. At a TNT concentration of 3 ppm, no detectable PL quenching effect in SpCz can be observed even within 30 min of exposure (Fig. 4a). This indicates that the integrated intensity of the quenched PL at the surface of the SpCz film relative to the non-quenched PL of the volume of the layer is below the detection limit of the setup. In strong contrast, the microporous PSpCz clearly shows a detectable PL quenching even at three orders of magnitude lower TNT concentrations of 5 ppb, roughly corresponding to the vapor pressure of TNT at room temperature (22 °C). This detection level impressively demonstrates the importance of the microporous morphology, which facilitates the accessibility of the TNT molecules to a substantially larger amount of chromophore units compared to the case of the thermally evaporated (non-porous) thin film of SpCz. Figure 4b shows that even on a time scale of seconds, a clearly detectable quenching efficiency can be found for a non-optimized PSpCz sample exposed to 5 ppb of TNT. The limits of various electron rich chromophore systems for the detection of TNT and other nitro-aromatic compounds has been recently reviewed 29 . The ability to detect levels of TNT in the gas phase equivalent to the vapor pressure at room temperature, puts our non-optimized microporous PSpCz on the same level with the most sensitive luminescence sensors for nitro-aromatics. We want to note that the detection limits of techniques like ion mobility spectroscopy or time of flight mass spectroscopy may reach the sub-ppt level 30,31 , but at the same time these are significantly more complicated and expensive tools. For comparison, a collection of other sensing techniques can be found in Table S1 (supporting information) and in the comprehensive review by Caygill et al. 2 .
Future optimization regarding quenching efficiency (further lowered detection limit) and faster response should be possible by designing MPNs with deeper LUMO levels, optimized pore sizes and optimum thickness of the MNP layer.

Conclusions
In summary, we have shown microporous layers of electrochemically crosslinked spiro-carbazole (PSpCz) as highly sensitive luminescence sensors for vapor-phase traces of nitro-aromatic compounds. Their high BET surface area of 1300 m 2 /g, which is more than three orders of magnitude higher than that of a thermally evaporated thin film of the monomer SpCz (0.88 m 2 /g), provides facile access of analyte molecules. The high LUMO level of PSpCz of − 2.58 eV provides a high driving force for electron transfer to electron-deficient nitro-aromatic compounds, like DNT and TNT. In striking contrast to a control sample based on thermally evaporated (non-porous) SpCz, which does not show any luminescence quenching upon exposure to TNT at levels below 3 ppm, the microporous PSpCz shows a clearly detectable response even at TNT concentrations as low as 5 ppb, clearly demonstrating the advantage of microporous films as luminescent sensors for traces of explosive analytes.
As control samples thin films of the monomer SpCz were thermally evaporated in high-vacuum (10 −7 mbar). The layer thickness was 50 nm as controlled by a stylus profiler. For a reasonable comparison with the PSpCz samples, we have used glass/ITO substrates for the thermally evaporated SpCz thin films.

Materials characterization.
Krypton sorption isotherms were recorded on a BEL Japan Inc. Belsorp-max system at 77 K. Samples were dried on a Belprep-vac II at 140 °C and ~2 Pa overnight prior to the gas sorption measurements. Optical absorption was measured using a JASCO V-670 UV-VIS spectrometer. The ionization energy of SpCz and PSpCz was determined by atmospheric pressure photoelectron spectroscopy using a Riken Keiki AC-2.
For determining transmission and reflection spectra a Deuterium Halogen lamp (DH-2000-BAL, OceanOptics) and a spectrometer with a range from 186 nm to 1041 nm (USB 2000 + XR1-ES) were used.
The photoluminescence (PL) of the samples was measured in the setup shown in Fig. 2a. For excitation, a diode pumped solid state laser (λ = 355 nm, 11 mW/cm 2 ) or a UV laser diode (λ = 405 nm, 35 mW/cm 2 ) was used. The PL signal was coupled into a monochromator and detected by a cooled charge coupled device camera (Princeton Instruments).
Luminescence quenching experiments. The apparatus used in the quenching experiments is a home-built system of two stainless steel chambers connected by a needle valve. The temperature in the chambers can be controlled separately by heaters. One of the chambers contains the analyte, while the other chamber contains the substrate with the sensor film. This chamber is equipped with an optical port for PL measurement. For the high vapour pressure analytes, their concentrations in the sample were adjusted by the needle valve between the two chambers monitored by a vacuum gauge. For analytes with limited vapour pressure, e.g. TNT, the vapour pressure in the experiment has been controlled by the temperature of the analyte chamber. Note, the temperature of the sample chamber was always kept somewhat higher than that of the analyte chamber to avoid condensation.