Introduction

Fluorescent chemosensory materials are one option for the detection of high explosives [1, 2]. They usually show good sensitivity toward volatile nitroaromatic explosives, such as 2,4,6-trinitrotoluene (TNT), and have the potential to operate in low-cost portable devices. Other explosive analytes, which include the military and civilian high explosives—HMX, RDX, and PETN, have higher lying excited states and are generally more difficult to sense. They have been detected with blue-emitting fluorescent polymers [3]. Polymeric luminescent materials containing conjugated organic frameworks take advantage of an amplified fluorescence quenching mechanism to increase detection sensitivity [4]. A key factor, among others, in the design of these polymeric materials is the inclusion of binding sites or porous pockets for effective sensor-to-analyte hydrophobic interactions. This can optimize analyte–polymer interactions and reduce energy barriers required for the excited-state electron-transfer process, thereby increasing the probability and efficiency of analyte detection. Swager et al. have developed rigid pentiptycene polymers that are porous in their solid-state thin films [5]. Porous films allow for ready access of highly volatile nitroaromatic explosives into the polymer structure. Adsorption is facilitated by hydrophobic π–π interactions. To further improve detection, other functionalities were employed to increase analyte binding [6]. The high sensitivity of detection has led to application of conjugated organic polymers in a commercial instrument for the detection of landmines by ICX Technologies.

We recently synthesized a conjugated silafluorene-vinylene polymer (1) and its model trimer complex (2) for explosives detection applications (Fig. 1a) [7]. In their thin films, these materials are highly luminescent in the UV-blue region of the spectrum. By targeting explosive particulates using a surface-detection method, these polymers were able to image a range of high explosives at picogram levels of sensitivity, including explosives of extreme low volatility, such as RDX and PETN. Trace detection of explosive particles is potentially important for forensic applications, such as locating areas where high explosives may have been assembled into improvised bombs. Currently, canine detection is the most widely used method for such applications, but portable trace detection approaches could be used in such applications. The silafluorene-containing polymer simultaneously detects multiple classes of explosives including nitroaromatic (TNT), nitramine (RDX), and nitrate esters (PETN) [7–10]. The relative sensitivity for detection correlates well with the energy of the lowest unoccupied molecular orbital (LUMO) of many explosive analytes [7]. The electrostatic potential (Mulliken charges) distribution calculated for the silafluorene trimer 2 reveals a positive charge density centered on the silicon atoms (Fig. 1b). This, in combination with the sensitive explosives detection results obtained for a wide range of silacycle polymers [7–15], suggested that specific sensor-to-analyte interactions might occur between these polymers and the targeted explosives.

Fig. 1
figure 1

a Chemical structure of poly(silafluorene-vinylene) (1) and methylsilafluorene-vinylene trimer (2); b Electrostatic potentials for 2 calculated using DFT at the B3LYP/6-31G* level of theory. Colors range from −0.668 to +0.668 with green denoting the electron-deficient regions and red denoting the electron-rich regions

Full size image

The main difference between 1 and conjugated organic polymers used for explosives detection is the incorporation of a silacycle moiety into the polymer backbone. The silicon atoms are part of a silacyclopentadiene unit which is responsible for the unique σ*–π/π* conjugation that defines their reduced bandgap energies [16, 17]. These silacycles in dimeric and trimeric model compounds for the polymers have been characterized by X-ray crystallography and show strained internal angles near 90° at the silicon atom [7, 10, 18]. Silacycles have previously found success as effective Lewis acid catalysts [19]. When exposed to ring strain, the silicon center may expand its valence shell to act as a Lewis-base binding site via a process labeled “strain release” Lewis acidity [20–22]. This raises the possibility that the hard Lewis acid silicon center may actively bind hard weakly basic oxygen atoms in the nitro or nitrate groups of organic high explosives. It is important to understand all the factors which control detection sensitivity. In the development of a spray-on reagent for explosives detection, efficiency, cost of synthesis, and minimizing the amount of polymer reagent needed are all important practical considerations.

Experimental

General

Caution: TNT is a high explosive and should be handled only in small quantities. All synthetic manipulations were carried out under an atmosphere of dry argon gas using standard Schlenk techniques. Dry solvents were purchased from Aldrich Chemical Co. Inc. and used after purification with an MBraun Auto Solvent Purification System. Spectroscopic grade toluene from Fisher Scientific was used for the fluorescence measurements. Reagents were purchased from Aldrich Chemical Co. and used as received. DNT were purchased from Aldrich Chemical Co. and recrystallized from methanol. TNT was prepared from DNT and recrystallized from toluene [15]. UV–Vis spectra were obtained with the use of a Hewlett-Packard 8452A diode array spectrometer.

Silicon NMR measurements

NMR data were collected using a Varian Mercury Plus spectrometer and a 9.4-T superconducting magnet (399.911 MHz for 1H and 79.450 MHz for 29Si NMR). The solution-state 29Si NMR spectra were recorded using an internal tetramethylsilane (TMS) reference, separated from the sample matrix using a tapered glass insert (Wilmad part WGS-5BL.) Chemical shifts were measured relative to the TMS reference (0 ppm) in all cases. 29Si NMR spectra were acquired using proton decoupling during the acquisition period only. The pulse delay was set 5 s and a pulse time of 7.5 μs was used (π/2 was measured as 14 µs for 2 in benzene-d 6.) T 1 was not measured for any of the 29Si sites. Typical acquisition times were 3 h. 1 Hz of exponential line broadening was applied to the data during processing. 1H T 1 measurements were performed on samples using an inversion-recovery pulse sequence and VNMR-native processing algorithms. Data for T 1 analysis were taken at 20 °C in benzene-d 6.The 29Si NMR data were fitted using the Lorentz equation in Origin8 (purchased from OriginLab):

$$ y = y_0 + \left[ {\left( {{{2A} \mathord{\left/ {\vphantom {{2A} \pi }} \right. } \pi }} \right)\;\left( {{w \mathord{\left/ {\vphantom {w {\left( {4\left( {x--x_{\text{c}} } \right)^2 + w^2 } \right.}}} \right. } {\left( {4\left( {x--x_{\text{c}} } \right)^2 + w^2 } \right.}}} \right)} \right] $$

Table 1 lists changes in 29Si NMR chemical shifts (Δ) for both the central and peripheral silicon resonances of 2 in the presence of the Lewis-base analytes. The changes in shifts are reported from single Lorentzian fits to the NMR data using the Simplex algorithms available in Origin 8. Errors are reported at the 95% confidence level (2σ). 29Si NMR spectra of the vinylene trimer (2) in various solvents may be found in the ESI.

Table 1 29Si NMR results
Full size table

Results and discussion

The nitro groups on the periphery of common explosive materials, such as TNT, have the potential to act as Lewis-base donors. It may be hypothesized that the silicon centers in the silole- and silafluorene-based polymers bind the nitro group of explosives, through the lone pairs of the nitro groups. In trace-explosive particle imaging, a dilute solution of polymer is sprayed on the sampling surface. Dissolution and mixing of explosive and polymer occurs as the film dries. Analyte–polymer acid–base interactions can provide a bridging bond through which electron transfer occurs from the excited state of the polymer to the explosive analyte. This helps explain the low detection limits observed (nanogram to picogram) and the general effectiveness of all studied silole and silafluorene polymers in detecting organic nitro, nitrate ester, and nitramine explosives using the spray-on-polymer detection method [7–15].

29Si NMR spectroscopy is a technique ideally suited to identify the possible substrate/analyte interactions through the use of NMR observable parameters, such as chemical shift and relaxation time [23–27]. The model trimer complex (2), which proved successful as a chemosensory material for a wide range of explosives in our previous study, provides a well-characterized structural fragment of polymer 1 [10]. A series of Lewis-base analytes were chosen to span a range of basicities. The analytes include dimethylsulfoxide (DMSO), nitrobenzene (NB), nitromethane (NM), acetonitrile (ACN), acetone (ACE), tetrahydrofuran (THF) and 2,4,6-trinitrotoluene (TNT). The chemical structure and Lewis basicities for these analytes, referenced to BF3 as the Lewis acid [28], are found in Table 1. Nitromethane was used as a model for aliphatic-based nitro/nitrate explosives such as RDX and PETN. Nitrobenzene and TNT represent aromatic based nitro explosives.

The 29Si NMR of 2 reveals two peaks that correspond to the peripheral silicon centers and the single central silicon center of 2 (see ESI). These peaks are up-field from the TMS resonance at −9.50 ppm and −18.9 ppm, respectively. Figure 2 shows the 29Si NMR spectra acquired in benzene-d 6 expanded around the resonance observed for the peripheral silicon centers of 2 with and without the addition of analytes. All spectra were acquired using 1H decoupling during acquisition only and were referenced to TMS (tetramethylsilane) at 0 ppm. TMS, itself a weak Lewis acid, was contained in an internal capillary to prevent any possible interaction with the basic analytes. The observed shifts are down-field and lie within a range of ~1 ppm. This finding is consistent with previous 29Si NMR studies analyzing Lewis-base displacement in rigid-rod silane polymers [29].

Fig. 2
figure 2

29Si NMR chemical shifts for the outer Si atoms of 2 at 20 °C when exposed to 33 equivalents of a no analyte; b NB; c TNT; d; ACE; e NM; f DMSO; g THF; h ACN. Spectra were taken in benzene-d 6 solvent

Full size image

Distinct shifts (Table 1) occur in the 29Si NMR resonances for both the silicon centers of 2 when exposed to TNT. This supports the hypothesis that nitro- and nitrate-containing organic explosives can bind to these silicon-containing polymers. This weak donor–acceptor interaction could provide an efficient pathway for electron-transfer quenching of the fluorescent excited state of silacycle containing polymers.

While a general correlation is observed between Lewis basicity and the Si chemical shift with added NB, ACE, DMSO, and THF (see ESI), the data obtained for NM and ACN indicate that the correlation may be complicated by additional effects. ACN is the least-sterically restricted base and may have better access to the Si centers. This could explain the largest shift difference (Δ) of 0.797(1) ppm noticed for this analyte. In addition, ACN is the only Lewis base in this study with nitrogen donor lone pairs.

Evidence of Lewis acid–base interactions in these strained silacycle systems can also be seen through analyte titration. Figure 3 shows the chemical shifts for the peripheral silicon of 2 when exposed to increasing equivalents of ACN. The 29Si NMR resonance shifts continually down-field on addition of ACN until a threshold is reached at around 33 added equivalents. An association constant (K a) of 0.12 M−1 was calculated for the binding of ACN to the peripheral silicon centers of 2 using the Scatchard method [30–32]. ACN shows the largest effect on the 29Si NMR resonances, providing an upper limit for K a in this study. Further support of substrate/analyte binding was revealed though NMR relaxation studies. Spin-lattice relaxation times (T 1) of 5.86(3) s and 4.83(4) s were measured for ACN by 1H NMR in benzene-d 6 at 20 °C in the absence and presence of 2, respectively. The significant change in T 1 values indicates a binding event between ACN and 2 [31, 33, 34]. The measured viscosity of both solutions at 20 °C were identical (0.688(3) cP), indicating no major change occurred in the solution properties affecting T 1.

Fig. 3
figure 3

Superimposed 29Si NMR spectra of the outer Si atoms in 2 when exposed to increasing equivalents of acetonitrile at 20 °C. a No acetonitrile; b three equivalents; c nine equivalents; d 15 equivalents; e 21 equivalents; f 27 equivalents; g 33 equivalents; h 39 equivalents

Full size image

Low temperature studies were performed with the intent to observe isolated populations of bound and unbound states. However, these studies were complicated by precipitation and the limiting physical properties of the solvents and analytes used. Binding studies were carried out in dichloromethane-d 2 to determine the extent that ring current effects perturb the chemical shift at silicon in toluene solvent. The peaks shift for 33 equivalents of acetone in dichloromethane-d 2 lowered to 0.270(6) ppm from 0.307(1) ppm in toluene-d 8 at room temperature and to 0.248(1) ppm from 0.436(1) ppm in toluene-d 8 at −50 °C, indicating a small but observable solvent effect with a small temperature dependence.

Binding of an analyte to the silicon center of these fluorescent molecules might also be expected to cause a change in the absorption spectra, since the lowest excited state involves excitation to the π* orbital of the metallole ring [7, 10, 15]. UV–Vis spectra were measured in a range of basic solvents. Several solvents used in the 29Si NMR study (i.e., NM and NB) could not be investigated due to their strong UV absorption. Figure 4a shows the spectra of the silafluorene trimer in a variety of solvents. It is evident that the solvent has a distinct effect on the absorption properties of the silafluorene trimer. For example, the spectra reveal a blue shift in the absorption maximum (λ max) of 80, 56, and 12 nm for ACN, dichloromethane (DCM), and DMSO, respectively. The isosbestic point obtained on addition of UV-transparent ACN (Fig. 4b) further suggests an equilibrium binding process that is very low, since saturation and a limiting spectrum is not obtained. The UV spectra of 2 in DMSO (dielectric constant = 47) and THF (dielectric constant = 7.6) are found to be very similar. This suggests that solvatochromism is not a major contributing factor to the spectral changes observed.

Fig. 4
figure 4

UV–Vis spectra of 2. a Spectra individually taken in 100% of given solvent. DCM = dichloromethane; b spectra taken in toluene with increasing percentages of ACN

Full size image

Conclusions

Using 29Si NMR as a probe for sensor-to-analyte interactions between silafluorene-containing materials and explosive analytes, it has been shown that lone pairs on both oxygen and nitrogen atoms can bind to the silicon centers through a Lewis acid–base interaction enhanced by the ring strain in the silacycle. Further evidence for these binding events was provided by spin-lattice relaxation times and UV–Vis spectroscopy. This suggests that a fifth factor can be added to the list of metrics used to influence the ability to detect explosives with fluorescent polymers:

  1. 1.

    Exciton delocalization along the conjugated polymer chain leads to amplified fluorescence quenching and improves sensitivity [1].

  2. 2.

    Hydrophobicity and polymer porosity may enhance selective binding of organic high explosive analytes, which can be further enhanced in molecularly imprinted polymers [5, 35].

  3. 3.

    The polymer excited state energy is crucial in determining the range of explosives detected and the relative sensitivities of detection [7].

  4. 4.

    Maximizing luminescence quantum yields for emission in spectral regions easily visualized by the eye are critical to improving the sensitivity of visual imaging approaches [8].

  5. 5.

    Incorporating hard Lewis acid centers in a delocalized polymer chain, which are strongly coupled with the fluorescent chromophore, may further enhance explosive analyte binding and provide a facile pathway for inner-sphere electron transfer to improve both sensor selectivity and sensitivity to detection by the electron-transfer quenching process.