1. Introduction
In recent years, the increasing scientific interest in the subject of environmental protection and safety has led to tremendous scientific developments in the detection of hazardous substances such as explosives [
1,
2]. Energetic nitroaromatic compounds, which include 2,4,6-trinitrophenol (PA), have attracted particular attention [
3,
4,
5]. One common application of PA is the use of the compound as a standard material for analytical methods such as HPLC [
6]. Due to its toxic and carcinogenic properties, it is extremely important that PA can be detected even in trace amounts [
7,
8]. Picric acid may cause damage to the eyes and skin, anemia, liver injury, and respiratory system damage [
7,
9]. For male and female F344 rats, the LD
50 doses for oral administration of PA were determined to be 290 and 200 mg/kg respectively [
10]. It has been reported that ingestion of 1 to 2 g of picric acid causes severe poisoning in humans [
11].
Due to the relatively high solubility of picric acid in water, even the smallest concentration of PA in water is intolerable. Maximum permissible concentrations have been established for this compound, e.g., by the National Institute for Occupational Safety and Health (NIOSH) and the Occupational Safety and Health Administration (OSHA)—according to the TWA method, the contamination of maximum PA in the air should not exceed 0.1 mg/m
3 [
12].
The most commonly used methods for detecting PA include mass spectrometry [
13], the use of field-effect transistors [
14], and fluorescence spectroscopy [
15]. Unfortunately, the problem with the above methods is the structure of the compound. Due to the similarity of the chemical structure of PA to the structures of other nitroaromatic compounds, it is difficult to differentiate between them and PA when using, for example, photo-induced electron transfer, resonance fluorescence energy transfer, or strong electrostatic interactions [
16]. These methods are, therefore, limited by their poor selectivity or complicated procedures. Therefore, it is very important to develop a highly sensitive method for the detection of PA.
Many examples of selective sensors for the detection of PA and other nitroaromatic compounds have already been described in literature. These sensors are often based on metal–organic frameworks [
17,
18] or carbon dots [
19], even if other materials are also applied for this purpose. The disadvantages of these sensors are the complex manufacturing process and their high unit costs [
20].
Molecular imprinting is a group of processing methods aimed at producing layers containing pores, whose shape and size match that of a selected template molecule. Typically, molecularly imprinted polymers (MIPs) are produced via the polymerisation of an adduct between the template molecule and a monomer. The resultant polymers, after the removal of the template molecule from the polymer matrix, allow for the specific adsorption of that template, which is of great significance for producing highly selective sensors. Particular research attention has been given to molecularly imprinted conjugated polymers, such as polycarbazole or polypyrrole [
21,
22,
23].
Efforts have also been undertaken to utilise MIPs for the detection of PA (
Table 1). Despite the existence of a few reports, no data about the effect of imprinting or comparisons with bare electrodes have been provided, making it impossible to identify the effect of molecular imprinting on the detection parameters of these sensors.
To obtain selective (with a high response only to the intended analyte and preferably no response to other analytes) [
27] and sensitive (low LOD) [
28] MIPs, it is important to examine if the template is compatible with the monomer (i.e., if there are interactions between them) [
29]. The most common technique used to produce MIPs is the self-assembly approach, followed by the polymerisation of the monomer, which relies on non-covalent interactions, e.g., hydrogen bonds [
30], ionic/hydrophobic interactions, etc. [
31]. The advantage of this type of interaction is the easy removal of the template from the template–monomer complex, e.g., extraction with a solvent [
32] or immersion in a solvent [
33]. Due to the fact that non-covalent interactions are easily disrupted, it is important to choose a monomer–template pair that will create complex with strong interactions between them [
34]. It has been confirmed that higher-energy bonding leads to the formation of an adduct with stronger interactions, resulting in a more selective MIP [
35].
In this work, we have provided theoretical background for the interactions between picric acid (PA) and a conjugated monomer, i.e., carbazole, based on quantum-mechanical calculations. We investigated the process of producing a MIP polycarbazole layer on platinum and glassy carbon electrodes and investigated their performance in detecting PA.
2. Materials and Methods
The following reagents were used in this work: acetylsalicylic acid (>99%, Sigma-Aldrich, St. Louis, MO, USA), sulfuric acid (>95%, Chempur, Karlsruhe, Germany), potassium nitrate (>95%, POCH S.A, Gliwice, Poland), carbazole (>97%, TCI, Tokyo, Japan), and tetrabutylammonium tetrafluoroborate (Bu4NBF4) (>98%, TCI).
2.1. Synthesis of 2,4,6-Trinitrophenol
Sulphuric acid (60 mL, 1.12 mol) was introduced into a three-necked flask equipped with a mechanical stirrer. Next, acetylsalicylic acid (6 g, 0.03 mol) was added in small portions over the course of approximately 60 min. After the addition of acetylsalicylic acid, the mixture was heated for 60 min at 115–120
C. Next, the reaction mixture was cooled to approximately 70
C, and potassium nitrate (13.5 g, 0.134 mol) was introduced in small portions, resulting in the temperature rising to 80–95
C and being kept in that range. After all of the potassium nitrate had been added, the reaction mixture was heated up to 120
C and stirred for 20 min. Following this, the heating was disengaged. After the mixture had cooled to room temperature, the contents of the flask were transferred to a tall beaker of deionised ice water. The precipitate was filtered under a vacuum and rinsed twice with small amounts of deionised water. Next, the raw product was recrystallized from deionised water. After the mixture cooled, the precipitate was filtered off and dried, resulting in 2,4,6-trinitrophenol (4.71 g, 0.021 mol). A summary of the reaction is presented in
Figure 1. The yield of the reaction was 70%. PA melting point: 122.5
C (capillary method),
1H NMR (300 MHz, DMSO-d
6)
(ppm): 8.59 (s, 2H, Ar-H). IR-ATR (diamond) (
Figure A2): 3108 cm
−1 (O-H) 2870 cm
−1 s (C-H), 1630 cm
−1 as (NO
2), 1606 cm
−1 (C=CAr), 1341 cm
−1 s (C-N), 1275 cm
−1 (C-O), 1154 cm
−1 (C-H) in-plane bending, 779 cm
−1 (C-NO
2), 703 cm
−1 (C-H out-of-plane bending, 663 cm
−1 (C-NO
2 wagging. Raman spectroscopy (laser 840 nm) (
Figure A3): 1636 cm
−1 C-C ring str., 1348 cm
−1 NO
2asym, 1280 cm
−1 NO
2sym,
C-N str, 831 cm
−1 NO
2 in plane (scissoring).
2.2. Electrochemical Investigations
Molecularly imprinted polymer (MIP) and non-imprinted polymer (NIP) layers were produced via electrochemical polymerisation. Electrochemical polymerisation was conducted using cyclic voltammetry in acetonitrile solutions containing 0.1 M tetrabutylammonium tetrafluoroborate (Bu4NBF4/MeCN) as a supporting electrolyte and 20 mM carbazole as the monomer. NIP films were produced directly from this solution, whereas the MIP layers were produced from solutions supplemented with 80 mM PA.
For electrochemical polymerisation, constant-surface-area electrodes made out of either platinum or glassy carbon were utilised as working electrodes. A platinum coil was used as the counter-electrode, and silver wire was used as the pseudoreference electrode. In the cases of both NIP and MIP layers, the parameters of the cyclic voltammetry experiments were identical and were as follows: the applied working electrode potential range was −0.5 V to +1.85 V, the potential scan rate was 0.1 V/s, and 10 potential cycles were conducted.
The synthesized MIP and NIP layers were investigated in PA solutions of varied concentrations via differential pulse voltammetry (DPV). The initial potential in DPV was 0.2 V, and the final potential was −2 V. The step potential was −0.005 V, the modulation amplitude was −0.035 V, the modulation time was 0.05 s, and the interval time was 0.5 s. The electrode setup that was utilised was identical to that described above for the electrochemical polymerisation experiments.
The imprinting factor (IF) was calculated as the ratio of the peak current observed for the MIP layer to the peak current observed for the NIP layer. The IF values were calculated for layers deposited on Pt that were used to detect PA, as well as the two selected interfering agents. The IF values were also determined for PCz layers deposited on the GC electrodes used for the detection of PA.
For the purpose of conducting cross-selectivity investigations, nitrobenzene (9 mM) and nitromethane (18 mM) were used as interfering agents. The cross-selectivity was investigated via DPV by utilising the same experimental parameters as in the case of the measurements conducted for the detection of PA.
2.3. Quantum Chemical Calculations
For the calculations, DFT/TDDFT (Time-Dependent Density Functional Theory) was used with the B3LYP [
36] hybrid functional combined with the 3–21 G(d) basis set. For all optimised structures, the frequency calculations were systematically achieved (at the same level of theory) to confirm the minimum nature of the optimised geometries. All calculations in this work were performed using the ORCA 4.1.1 [
37] package programs. Input files and molecular orbital plots were prepared with the Gabedit 2.4.7 software [
38].
2.4. SEM Analyses
The morphology of MIP PCz and NIP PCz layers deposited on the Pt electrodes was investigated using a Phenom ProX (Waltham, MA, USA) scanning electron microscope (SEM). The basic SEM operation parameters were the following: The working distance was 10–11 mm, the acceleration voltages of the incident electron were 15 kV, and images were recorded at a 6000× and 15,000× magnification.
4. Conclusions
Quantum-mechanical calculations indicated that PA interacts strongly (31.43 kJ/mol) with both carbazole (monomer) and the repeat units of polycarbazole, which is typically sufficient for achieving a significant increase in the sensitivity of sensors due to molecular imprinting. Despite the existence of these interactions, the electrochemical detection results show only a marginal effect of molecular imprinting in the case of modified Pt electrodes (LODs of 0.26 and 0.62 mM, respectively, for MIP PCz/Pt and NIP PCz/Pt), translating into an imprinting factor of 1.77. Conversely, in the case of modified GC electrodes, molecular imprinting appeared to be counter-productive (IF = 0.95), as it results in an increase in the LOD values (0.57 and 0.12 mM, respectively, for MIP PCz/GC and NIP PCz/GC).
The very minor improvement of PA detection upon molecular imprinting likely stems from the fact that not only are the conjugated polymer chains highly rigid, but upon doping and de-doping, they undergo dearomatisation and rearomatisation, significantly changing their arrangement in space. This process likely leads to the gradual deformation of any pores remaining after the removal of the template, translating into a decrease in the performance of the MIP over time down to the NIP performance baseline.
The deformation of pores hypothesis is also supported by the results of cross-selectivity investigations, as the NIP layers show higher selectivity towards nitrobenzene than the MIP layers. The IF value calculated for the layers deposited on Pt and used to detect nitrobenzene is 5.70, much higher than the value of 1.77 observed for PA. Conversely, the IF observed in the case of nitromethane is 0.59. These results indicate that while molecular imprinting increased the response of the layer towards nitroaromatics in general against nitroalkanes, it is not sufficiently selective to differentiate between nitrobenzene and PA. This feature can be attributed to the change in the shape of the pores present in the MIP layers, as contrasted to the random distribution of pore sizes in the NIP layers. Where the random distribution in the NIP layers allows pores of different sizes, imprinting increases the share of pores with sizes roughly corresponding to the size of the template molecule. Consequently, even though the pore shape begins deviating due to repeated doping and de-doping, pore size will remain roughly similar, explaining the observed IF values that were >1 for PA and nitrobenzene, as well as the IF < 1 value for nitromethane.
The lower performance of electrodes modified with either NIP or MIP PCz layers in comparison to that of the unmodified electrodes may be caused by the relatively lower conductivity of the conjugated polymer layers in comparison with either Pt or GC electrodes. Moreover, polycarbazole typically produces layers that vary significantly in thickness, due to its nucleation mode, which may also hinder the adsorption of the planar and highly polar PA molecules on the surface of this polymer in comparison with the highly planar PT and GC electrode surfaces.
Taking the above into consideration, two main factors necessary for the successful use of molecularly imprinted conjugated polymers can be postulated. Firstly, during electrochemical polymerisation, the precipitating polymer film must not undergo repeated doping/de-doping, as this process appears to distort the size and shape of the existing pores, as discussed above. This is evidenced by the fact that molecularly imprinted polycarbazole derivatives were utilised as receptor layers for sensors when their electrodeposition did not involve their de-doping [
41]. This factor can also explain the very broad application of polypyrrole-based MIP sensors, as polypyrroles undergo de-doping only at very strongly negative potentials, usually exhibiting a similar doping state across the typical conditions of their electrosynthesis process. Secondly, a conjugated polymer with a nucleation mode more suited to the template molecule should be used so as to promote the adsorption of the template onto the surface of the molecularly imprinted conjugated polymer film.