Nitroaromatic explosives detection using electrochemically exfoliated graphene

Detection of nitroaromatic explosives is of paramount importance from security point of view. Graphene sheets obtained from the electrochemical anodic exfoliation of graphite foil in different electrolytes (LiClO4 and Na2SO4) were compared and tested as electrode material for the electrochemical detection of 2,4-dinitrotoluene (DNT) and 2,4,6-trinitrotoluene (TNT) in seawater. Voltammetry analysis demonstrated the superior electrochemical performance of graphene produced in LiClO4, resulting in higher sensitivity and linearity for the explosives detection and lower limit of detection (LOD) compared to the graphene obtained in Na2SO4. We attribute this to the presence of oxygen functionalities onto the graphene material obtained in LiClO4 which enable charge electrostatic interactions with the –NO2 groups of the analyte, in addition to π-π stacking interactions with the aromatic moiety. Research findings obtained from this study would assist in the development of portable devices for the on-site detection of nitroaromatic explosives.

materials, such as graphene nanoribbons and nanosheets 5,41 , hydrogenated graphene 42 as well as graphenes prepared from Hummers, Staudenmaier and Hofmann syntheses methods 11,41,42 in order to isolate the most suitable probe material for the sensing of NAC explosives in seawater. Various works performed by other research groups have also supported the high selectivity and sensitivity of functionalised graphene, including graphene modified with Ag nanoparticles 43 , PtPd concave nanocubes 44 and ionic liquid 45 , as well as other carbon-based electrochemical sensors such as carbon quantum dots for highly efficient NACs detection 25 .
Recently, a novel and highly efficient method has emerged as a possible avenue for the mass production of graphene: the electrochemical exfoliation of graphite under anodic or cathodic conditions, which enables the intercalation of electrolyte ions between the graphene layers, resulting in their exfoliation 46 . It has been previously established that the physicochemical properties of graphene is strongly affected by the exfoliation conditions such as the applied potential and electrolyte used for the procedure 47 . As such, we aim here to study the application of electrochemically exfoliated graphene prepared from different electrolytes for the detection of DNT and TNT in seawater. Graphene sheets are prepared from the electrochemical exfoliation of commercially available graphite foil using two different electrolytes under anodic conditions: LiClO 4 and Na 2 SO 4 .

Results
Inherent Electrochemistry of Electrochemically Exfoliated Graphene Materials. Cyclic voltammograms were recorded in a blank buffer (borate buffer solution, BBS, pH 9.2) to ascertain the presence of inherent reduction peaks for the graphene materials that may interfere with the electrochemical reduction of DNT and TNT (Fig. 1a). Cyclic voltammetric measurements were also performed in seawater to determine the influence of species present in the matrix (Fig. 1b). The pH of all electrolytes were maintained within the pH range of 8 to 10, which had been previously established by our group to produce the largest and most stable current response for a variety of graphene-based materials 41 . Graphene prepared from LiClO 4 (G-LiClO 4 ) showed an intense, broad reduction peak at − 819 mV and − 724 mV in BBS and seawater respectively, whereas graphene synthesised in Figure 1. Cyclic voltammograms of (a) blank buffer and (b) seawater for GC electrodes modified with graphene obtained in LiClO 4 (G-LiClO 4 ) and Na 2 SO 4 (G-Na 2 SO 4 ). Cyclic voltammograms of bare GC electrode is also shown for comparison. Conditions: (a) BBS (20 mM, pH 9.2) and (b) mixture of 9:1 volume ratio of seawater (pH 8.0) to BBS (200 mM, pH 9.2). Scan rate 100 mV s −1 .
Scientific RepoRts | 6:33276 | DOI: 10.1038/srep33276 Na 2 SO 4 (G-Na 2 SO 4 ) showed a low-intensity peak at − 655 mV in BBS which does not seem to appear when measuring in seawater. Overall, relatively similar voltammograms were obtained in both BBS and seawater within the 0 mV to − 1000 mV potential range, which is indicative of the absence of electroactive compounds in seawater that may complicate the analysis of NAC explosives. The intense inherent reduction peak recorded for G-LiClO 4 in both BBS and seawater can be attributed to the irreversible reduction of the oxygen functionalities present onto this material that can undergo electrochemical reduction. This phenomenon has been investigated more in detail and explained in previous reports 48,49 . Furthermore, the intensity of the reduction peak recorded for both graphene materials is in agreement with the corresponding C/O ratio which is of 4.0 for the G-LiClO 4 and 8.8 for the G-Na 2 SO 4 (see Supplementary Fig. S2).
Electrochemical Detection of 2,4-Dinitrotoluene (DNT) and 2,4,6-Trinitrotoluene (TNT). Cyclic voltammetry scans were first obtained in BBS to investigate the electrochemical behaviour of DNT and TNT on the electrochemically exfoliated graphenes and then repeated in seawater to determine the feasibility and reliability of real-time environmental analysis (Fig. 2).
The reduction of DNT revealed two major reduction peaks in both BBS and seawater systems, and they are labelled a and b by order of appearance in the CV cathodic wave (Fig. 2a,b). Notably, peak b coincides with the inherent reduction peak of oxygen functionalities on the graphene materials. As such, the first reduction peakpeak a -is conveniently chosen for the quantitative analysis of DNT. For all three electrode systems, the reduction potential of peak a was less negative in seawater than in BBS ( Table 1). The peak height of the reduction wave a was generally comparable in both BBS and seawater systems.
Cyclic voltammograms of TNT revealed, as expected, three major reduction peaks in both BBS and seawater systems, labelled as c, d and e by order of appearance in the CV cathodic wave (Fig. 2c,d). Notably, peaks d and e are almost undetectable for bare GC and G-LiClO 4 in BBS, and for bare GC and G-Na 2 SO 4 in seawater. In addition, peak e coincides with the inherent reduction peak of oxygen functionalities for G-LiClO 4 . As such, the first reduction peak -peak c -is chosen for the quantitative analysis of TNT. Similar to the electrochemical sensing of DNT, peak c consistently appeared at lower reduction potentials for all three electrodes systems when switched from the BBS system to the seawater system ( Table 1). The peak height of the reduction wave c was also generally comparable in both BBS and seawater systems, with the exception of G-LiClO 4 displaying a marked increase in peak intensity in the seawater system.
After the investigation carried out using cyclic voltammetry, a more sensitive electrochemical techniquedifferential pulse voltammetry -was then chosen for the trace analysis of DNT and TNT. As shown in Fig. 3, differential pulse voltammograms were obtained in BBS and seawater, for a range of DNT concentrations spanning across 0 ppm and 20 ppm, in increments of 4 ppm and using the optimized amount of electrocatalyst deposited on the GC electrode surface (see Supplementary Fig. S3). A larger signal response is apparent in the seawater system for both electrochemically exfoliated graphene materials, in particular with the G-LiClO 4 demonstrating a doubling of current intensities at all concentrations, as compared to the BBS system (see Supplementary Fig. S4). In addition, G-LiClO 4 exhibited the largest reduction peak intensities across all concentrations in both BBS and seawater systems, as compared to the other two electrode systems.
Differential pulse voltammetry study in BBS and seawater for TNT (Fig. 4) revealed a larger signal response in the seawater system for both electrochemically exfoliated graphene materials, with G-LiClO 4 demonstrating an increase in current intensities by 2 folds at all concentrations, as compared to the BBS system (see Supplementary  Fig. S5). In addition, G-LiClO 4 exhibited the largest reduction peak intensities across all concentrations in both BBS and seawater systems.
Calibration graphs based on the current intensities of reduction peak a obtained from DPV measurements were then plotted as a linear function of spiked DNT concentrations (Fig. 5a,b).
The overall sensitivity of the electrode system for DNT detection, represented by the slopes of the linear calibration plots, was highest for the G-LiClO 4 , followed by G-Na 2 SO 4 , then the bare GC electrode, in both BBS and seawater systems. G-LiClO 4 consistently demonstrated higher current intensities than the other two electrode systems at all DNT concentrations. The LOD value was lowest for G-LiClO 4 , followed by G-Na 2 SO 4 , then bare GC in BBS buffer. In the seawater system, the LOD value was lowest for bare GC, followed by G-LiClO 4 , then G-Na 2 SO 4 ( Table 2).
It is noteworthy that both graphene materials exhibited enhancements in sensitivities when switched from the BBS system to the seawater system. In particular, doubling of system sensitivity was achieved for G-LiClO 4 .
The calibration graphs based on reduction peak c plotted as a linear function of spiked TNT concentration (Fig. 5c,d) showed that the sensitivity of the electrode system was highest for bare GC, followed by G-LiClO 4 , then G-Na 2 SO 4 in BBS buffer. When measuring in seawater, a different result was obtained. Electrode modified with G-LiClO 4 consistently demonstrated higher current intensities than the other two electrode systems at all TNT concentrations. The LOD value was lowest for bare GC, followed by G-Na 2 SO 4 , then G-LiClO 4 in BBS buffer. In the seawater system, the LOD value was lowest for G-LiClO 4 , followed by bare GC, then G-Na 2 SO 4 .
It is noteworthy that all three electrode systems exhibited enhancements in sensitivities in the seawater system as compared to the BBS system. In particular, tripling of system sensitivity was achieved for G-LiClO 4 . Improvements in electrochemical detection of TNT by G-LiClO 4 in comparison to G-Na 2 SO 4 were also observed similar to DNT sensing.
The enhanced electrochemical performance of G-LiClO 4 as compared to G-Na 2 SO 4 demonstrates that π -π stacking interactions cannot be considered as the sole factor influencing the electron transfer behaviour of the graphene materials towards the reduction of NACs. Evidently here the presence of oxygen functional groups are beneficial to the adsorption and electrochemical reduction of NACs despite a less efficient π -π stacking interaction due to the competition with water molecules 50 and also a reduced conductivity of graphene materials carrying oxygen moieties 51 . As demonstrated in other works 52,53 , concomitant factors influence the electroanalytical detection of redox active molecules with the dominant one differing in relation to each particular analyte. A careful investigation and optimisation is thus highly recommended prior the application of newly prepared carbon materials for electrochemical sensing purposes.
The improved electrochemical signal observed for G-LiClO 4 may be the result of additional electrostatic interactions between the electron-withdrawing -NO 2 groups of NACs and the oxygen functionalities on the graphene material 54 , dominating over the EDA stacking interactions. This study highlights the possibility of using graphene-based nanomaterials with high oxygen content as potential electrode material for the on-site electrochemical detection of nitro-substituted aromatic explosives. The selectivity of this sensing platform was evaluated in the seawater system, by observing changes in electrochemical response upon spiking increasing amounts of DNT into a solution containing 8 ppm of TNT. The simultaneous reduction of DNT and TNT revealed three major reduction peaks, which are denoted as f, g and h by order of appearance in the cathodic wave (Fig. 6). As it can be seen in Fig. 6, the reduction peak f can be solely assigned to TNT since it remains fairly consistent with the increase of DNT concentration and it is clearly independent. This allows a selective detection of TNT in the presence of DNT. In addition, peak g and h follow closely to the reduction signals recorded for DNT with negligible influence from the second and third reduction waves produced by TNT. This phenomenon is even more pronounced for G-LiClO 4 due to the enhanced electrochemical performance of the material. The sensing platform proposed in this work can thus selectively detect DNT and TNT in a NAC mixture sample. Conclusions This study has investigated and compared the sensing abilities of electrochemically exfoliated graphene obtained from the anodic exfoliation of graphite foil in different electrolytes (LiClO 4 and Na 2 SO 4 ), for the detection of nitro-substituted aromatic explosives (DNT and TNT) in both borate buffer and seawater. Voltammetry studies reveal that G-LiClO 4 which carries the highest amount of oxygen functional groups, is the most suitable electrode system for the detection of NAC explosives, due to a favourable electrostatic interaction with the analytes which resulted in a higher sensitivity and linearity. On the other hand, G-Na 2 SO 4 consistently demonstrates similar

of various electrode systems towards DNT and TNT detection in (A) BBS and (B) seawater.
Data is based on the first reduction peak of DNT and TNT.
electrochemical performance with the unmodified GC electrode for NAC detection. Significant improvements in electrode system sensitivity and linearity for DNT and TNT detection in seawater as compared to BBS were also observed for G-LiClO 4 , accompanied by the ability of our sensing platform to distinguish between DNT and TNT, which supports the applicability and effectiveness of using G-LiClO 4 in real-time environmental analysis of seawater. Methods Materials. Sodium tetraborate decahydrate and 2,4-dinitrotoluene (DNT) were purchased from Sigma-Aldrich. N,N-dimethyl formamide (DMF) and acetonitrile were purchased from Merck (Singapore). 2,4,6-trinitrotoulene (TNT) was purchased from AccuStandard (New Haven, CT) as an analytical standard in the diluted form (1000 ppm in acetonitrile). Graphene electrochemically exfoliated in LiClO 4 (G-LiClO 4 ) and Na 2 SO 4 (G-Na 2 SO 4 ) were prepared and characterised in detail as previously reported 47 . The graphene flakes had an average size of 3-10 μ m and an average thickness of 2.8 nm. G-LiClO 4 had a D/G ratio of 1.00 and a C/O ratio of 4.0, whereas G-Na 2 SO 4 had a D/G ratio of 0.95 and a C/O ratio of 8.8. Scanning transmission electron micrographs, XPS and Raman spectra are presented in the Supporting information file as Figs S1 and S2. Sand-filtered seawater (pH 8.0) was purchased from a local aquarium store in Singapore. Milli-Q water of resistivity 18.2 MΩ cm was used throughout the experiments.
Apparatus. Voltammetry analyses were conducted at room temperature using a μ Autolab Type III electrochemical analyser instrument (Electrochemie, Utrecht, The Netherlands) and controlled by General Purpose Electrochemical Systems Version 4.9 software. The experiments were performed in an electrochemical cell using a three-electrode system. Glassy carbon (GC, 3 mm in diameter) working electrode, platinum auxiliary electrode and Ag/AgCl (saturated) reference electrode were obtained from CH Instruments (Austin, TX).
Electrochemical measurements. Suspensions (1 mg mL −1 ) of the electrochemically exfoliated graphene materials were prepared from the dispersion of graphene flakes in DMF and sonication for 2.5 h. An aliquot (1 μ L) of the suspension was then drop-casted onto a GC electrode surface that had been polished with alumina particles (0.05 μ m) on a polishing pad, and allowed to evaporate under a heat lamp. The deposition step was performed to obtain one to three layers of graphene material on the GC electrode for the optimisation experiments. DNT was prepared as a stock solution (1000 ppm in acetonitrile). TNT (1000 ppm in acetonitrile) was used as received.
Borate buffer (BBS, pH 9.2) was prepared from dissolving Na 2 B 4 O 7 .10H 2 O in milli-Q water. Seawater (pH 8.0) was mixed with BBS (200 mM, pH 9.2) in a ratio of 9:1 prior to use. All cyclic voltammetry (CV) measurements were performed with a scan rate of 100 mV s −1 . All differential pulse voltammetry (DPV) measurements were performed with a modulation amplitude of 25 mV and a step potential of 5 mV. All voltammetry measurements were repeated three times to ensure measurement reproducibility, with the most representative voltammogram being selected for data presentation. Voltammetry measurements were also conducted on the unmodified GC electrode for reference purposes. Degassing was not carried out prior to electrochemical measurements to simulate real-time environmental analysis.