Talanta 139 (2015) 181–188
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
Electrochemical sensor for nitroaromatic type energetic materials using gold nanoparticles/poly(o-phenylenediamine–aniline) film modified glassy carbon electrode Şener Sağlam, Ayşem Üzer, Yasemin Tekdemir, Erol Erçağ, Reşat Apak n Istanbul University, Faculty of Engineering, Department of Chemistry, 34320 Istanbul, Turkey
art ic l e i nf o
a b s t r a c t
Article history: Received 11 November 2014 Received in revised form 23 February 2015 Accepted 28 February 2015 Available online 9 March 2015
In this work, a novel electrochemical sensor was developed for the detection of nitroaromatic explosive materials, based on a gold nanoparticle-modified glassy carbon (GC) electrode coated with poly(o-phenylenediamine–aniline film) (GC/P(o-PDA-co-ANI)-Aunano electrode). Nitroaromatic compounds were detected through their π-acceptor/donor interactions with o-phenylenediamine–aniline functionalities on the modified electrode surface. The enhanced sensitivities were achieved through π–π and charge-transfer (CT) interactions between the electron-deficient nitroaromatic compounds and s-/π-donor amine/aniline groups linked to gold nanoparticles (Au-NPs), providing increased binding and preconcentration onto the modified GC-electrodes. Selective determination of nitroaromatic type explosives in the presence of nitramines was enabled by o-PDA and reusability of the electrode achieved by Au-NPs. Calibration curves of current intensity versus concentration were linear in the range of 2.5–40 mg L 1 for 2,4,6-trinitrotoluene (TNT) with a detection limit (LOD) of 2.1 mg L 1, 2–40 mg L 1 for 2,4-dinitrotoluene (DNT) (LOD¼1.28 mg L 1), 5–100 mg L 1 for tetryl (LOD¼3.8 mg L 1) with the use of the GC/P(o-PDA-co-ANI)-Aunano electrode. For sensor measurements, coefficients of variation of intra- and inter-assay measurements were 0.6% and 1.2%, respectively (N¼5), confirming the high reproducibility of the proposed assay. Deconvolution of current contributions of synthetic (TNTþDNT) mixtures at peak potentials of constituents was performed by multiple linear regression analysis to provide high sensitivity for the determination of each constituent. Determination options for all possible mixture combinations of nitroaromatic explosives are presented in this work. The proposed methods were successfully applied to the analysis of nitroaromatics in military explosives, namely comp B, octol, and tetrytol. Method validation was performed against GC–MS on real post-blast residual samples containing both explosives. & 2015 Elsevier B.V. All rights reserved.
Keywords: Cyclic voltammetry Nitro-aromatic explosives Gold nanoparticles Poly(o-phenylenediamine–aniline) Modified electrode
1. Introduction The detection of explosives and explosive-related compounds has become a heightened priority in recent years for homeland security and counter-terrorism applications. There are many studies on the detection of explosives utilizing a multitude of different analytical techniques. The highly sophisticated current techniques for detecting/quantifying trace explosives include ion mobility spectrometry (IMS) and gas–liquid chromatography (GLC) coupled to mass spectrometry (MS); however, most of these devices are bulky, expensive, and require qualified personnel and time-consuming procedures [1]. In this regard, voltammetric techniques constitute cheaper and portable alternatives [2] if sufficient selectivity is attained with the aid of rapidly growing sensor electrodes. Naturally the aimed selectivity is n
Corresponding author. Tel.: þ 90 212 473 7028; fax: þ90 212 4737180. E-mail address: [email protected] (R. Apak).
http://dx.doi.org/10.1016/j.talanta.2015.02.059 0039-9140/& 2015 Elsevier B.V. All rights reserved.
closely related with the accessible potential range, number and variety of redox-active agents in the range of concern, and signal widths. Generally, the advantages of electrochemical systems in on site/in situ measurements of explosives comprise good sensitivity and selectivity, reasonable linear concentration ranges, minimal space and power requirements, and low-cost instrumentation. Electrochemical sensors offer good prospects for meeting the growing needs of explosives field detection. The most characteristic redox property of nitro-based explosives, namely the presence of easily reducible nitro-groups, ideally lends itself to electrochemical detection [3–6]. Different electrode materials found application for electrometric estimation of nitroaromatic and nitramine explosives. Such studies include detection of explosives at Hg-film electrodes [7–9], glassy carbon electrodes (GCEs) [10,11] and carbon fiber electrodes (CFE) [12–14]. There have also been a range of studies on the voltammetric detection of explosives at screen-printed carbon electrodes [4–10,15–18]. The use of imprinted polymers as functional
182
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
sensing matrices may suffer from several limitations, e.g., thick polymer matrices may be required for increased sensitivity due to the low density of imprinted sites. But as a disadvantage, this may lead to slow binding of the analytes to the recognition sites (long analysis time intervals) and to inefficient communication between the binding sites and the transducers. In fact, several studies suggested the use of imprinted monolayers [19,20], multilayers [21,22], and thin films to overcome these difficulties. Nanoparticles possess distinct physical and chemical properties that make them perfect scaffolds for the preparation of novel chemical sensors [23]. The unique electronic and optical properties of metallic and semiconductor nanoparticles (NPs) added new dimensions to the area of sensors [24]. An example is the use of polymer particles and nanoparticles that, when bound to an explosive molecule, change one of their measurable properties resulting in specific detection [1]. In a template strategy for increasing the selectivity/sensitivity of analyte recognition by molecular imprinting, electropolymerizable p-aminothiophenol (ATP) was first assembled on the AuNPs at the glassy carbon electrode surface by the formation of Au–S bonds, and subsequently, a phosphorothioate-group pesticide template was further assembled onto the monolayer of ATP through the hydrogen-bonding interaction between the free amino group on AuNPs and the organophosphorus pesticide [25]. The aggregation of Au NPs as a result of sensing reactions, giving rise to interparticle-coupled surface plasmon resonance (SPR) absorption wavelength shifts, was used for the development of colorimetric sensors [26]. For example, our research group optimized the differential hydrolysis of nitramine explosives (RDX and HMX) into nitrite, and developed a selective sensor by reacting 4-aminothiophenol (4-ATP) modified Au NPs with the nitrite analyte and naphthylethylene diamine (NED) as coupling agent for azo-dye formation, giving rise to a bathochromic shift in SPR absorption wavelength of further aggregated Au NPs [27]. The use of nanosensors has the potential to satisfy the requirements of an effective platform for trace detection of explosives. The combination of electroanalysis with nanomaterials as electrode modifiers also allows the easy integration of nanosensors with advances in conventional detection platforms, exploiting the physical and chemical benefits of interactions at the nanomaterial electrode surface [28]. Carbon nanotube-modified glassy carbon electrode for adsorptive stripping voltammetric determination of ultratrace TNT has been claimed to reach sub-μg L 1 levels of detection with linear response between 0.1 and 1 mg L 1 [29]. By utilizing an electrochemical sensor based on polyaniline deposition as a nanofibrous composite with a polypeptide, the electron-donating amino groups of polypeptide showed charge-transfer interactions with electrondeficient TNT in close proximity to the conducting polymer, allowing TNT determination in solution by adsorptive stripping voltammetry [30]. Other researchers used a composite of polyaniline and carbon nanotubes for the detection of nitroaromatic vapors containing TNT, 2,6-DNT and picric acid, where the amine group of polyaniline forms a charge transfer complex with nitro-energetic compounds leading to signal enhancement [31]. Another conducting polymer used within a vapor sensor is poly(3,4-dioxyethylene thiophene) (PEDOT) which could be deposited to form a conductive nanowire between two gold electrodes. Coating this electrical junction with an ionic liquid allowed the system to adsorb TNT vapor from the air, where the electrochemical reduction of TNT as well as the conductance change of the polymer nanojunction arising from the reduction product could be simultaneously measured [32]. In general, many sensing methods reported in literature do not demonstrate the ability to effectively differentiate and simultaneously sense explosive compounds in natural mixtures such as contaminated soil and groundwater, because overlap regions of electroactivity in nitro-explosive compounds result in cross-contribution of current from one compound to another, requiring
deconvolution of contributions from each compound of structural similarity [33]. Thus, the aim of this study is to develop a simple, sensitive and selective nanoparticle-based electrochemical sensor for the simultaneous quantitative estimation of selected nitro-aromatic energetic materials. Spectrophotometric method development was reverted to only for the HPLC-free analysis of environmental TNT metabolites (e.g., monoamino-dinitrotoluenes) basically yielding similar voltammograms to those of TNT. Thus, synthetic and real mixture analysis at low cost was feasible for nitro-aromatic energetic materials using simple voltammetry with a specially designed electrode, supported with colorimetry where required.
2. Experimental 2.1. Chemicals, solutions and instruments The explosive materials TNT (2,4,6-trinitrotoluene), DNT (2,4dinitrotoluene), tetryl (2,4,6-trinitrophenylmethylnitramine), RDX (1,3,5-trinitro-1,3,5-triazacyclohexane), HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazosin), Comp B composite explosive (containing 60% RDX, 39% TNT, and 1% wax), octol composite explosive (containing 70% HMX and 30% TNT), and tetrytol composite explosive (containing 30% TNT, 70% tetryl), were kindly supplied by Makine Kimya Endustrisi Kurumu (MKEK: Machinery & Chemistry Industries Institution) through the supervision of Milli Savunma Bakanligi, Teknik Hizmetler Daire Baskanligi (Ministry of National Defense, Office of Technical Services) of Turkey. 2-amino-4,6-dinitrotoluene (2-ADNT) and 4-amino-2,6-dinitrotoluene (4-ADNT) were supplied from Dr. Ehrenstorfer as the reference standard samples. The solvent used in preparation and dilution of explosive solutions was HPLCgrade extra pure acetonitrile (E. Merck). The alumina slurry used for electrode cleaning was from Baikowski International Corp. (0.05 mm, Baikalox 0.05CR). The supporting electrolyte for conductivity was sodium chloride (Riedel-de Haën). For electrode cleanliness, isopropyl alcohol (Sigma-Aldrich), acetone and ethanol (both technical grade) were used. The electrode coating materials, o-phenylenediamine (o-PDA) and aniline (ANI), as well as the rest of the reagents, were supplied from E. Merck (Darmstadt, Germany). The standard stock solutions of individual explosive materials and military-purpose explosive mixtures (Comp B and octol) at 1000 mg L 1 concentrations were prepared in extra pure acetonitrile. Gold (III) chloride solution (99.99% trace metals basis, 30% by wt. in dilute HCl) was used for deposition of gold nanoparticles to GC electrode coated with o-PDA–ANI copolymer, and the 0.04% (w/v) working solution was prepared from this stock solution. Solutions used for the analysis of 2-ADNT and 4-ADNT samples: NaNO2 was prepared as 1% with water; N-(1-naphthyl)ethylenediamine (NED) weighed as 0.375 g, dissolved and diluted to 50 mL with water; sulfamic acid 3% solution prepared with water; stock solutions of the analytes, namely 990 mg L 1 4-amino-2,6-dinitrotoluene, 995 mg L 1 2-amino-4,6-dinitrotoluene, and 100 mg L 1 TNT prepared with acetonitrile. Voltammetric experiments were performed with a Gamry Instruments model Reference 600 potentiostat/galvanostat/ZRA interfaced to a PC computer and controlled Gamry Framework software. Glassy carbon (BASi stationary voltammetry electrodes; ø 1.6 mm, area 0.02 cm2) was used as the working electrode. A platinum (Pt) electrode and an Ag/AgCl electrode served as the auxiliary and reference electrodes, respectively. For validation of the proposed assay against a GC–MS system, real post-blast residual samples were analyzed with the use of a Thermo Scientific Trace gas chromatograph coupled with a DSQII mass spectrometer containing positive ion chemical ionization and quadrupole analyzer. GC was equipped with a Thermo TR-HT5 (15 m 0.25 mm, ID 0.1 mm, 5% phenyl polycarborane siloxane).
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
183
Calibration samples for GC–MS analysis were prepared from the 1000 mg L 1 stock solutions of TNT and DNT in acetonitrile with the use of the standard reference samples of ULTRA EPA-1243 TNT and Dr. Ehrenstorfer DNT. The working solutions of TNT at 3–10 mg L 1 and DNT at 1–10 mg L 1 were prepared from the indicated stock solutions using acetonitrile as diluent.
cell was filled to 5 mL with acetonitrile have a final solvent composition of 95% water 5% AcCN. Tetryl was voltammetrically analyzed in the 5.0–100 mg L 1 concentration range using a 5 mL volume of solution homogenized in an ultrasonic bath. CV was performed in a potential range from 0.2 V to 1.2 V, and the characteristic peak potentials for each nitroaromatic compound were determined.
2.2. Optimization of voltammetric method
2.7. Study of synthetic and real mixtures with deconvolution of current contributions from each constituent at peak potentials
Electrode type, scan rate, supporting electrolyte and concentration parameters have been individually examined. A glassy carbon (GC) electrode was selected as the working electrode, by which a substantial signal can be produced (as current intensity) at sufficiently separated reduction potentials. A scan rate of 50 mV s 1 was chosen, and NaCl was preferred as the supporting electrolyte. The electrolyte concentration was found to be optimal at 0.04 M. Mixtures of hexane, dichloromethane, acetonitrile, water, dioxane– methanol (1:1) and acetonitrile–methanol (1:1), acetonitrile–water (1:1) and acetonitrile–water (5–95%) were tested for solvent optimization to yield the latter acetonitrile–water mixture as optimal. 2.3. Glassy carbon electrode pre-treatment The GC electrode was polished with a suspension of alumina powder in the presence of pure water on a smooth polishing cloth [34] by circular movements for a few minutes, then washed with distilled water, and sonicated for 5 min in bidistilled water. Sonication of the electrode was repeated for another 5 min in isopropyl alcohol–acetonitrile (1:1, v/v) mixture. The efficiency of this procedure for electrode cleanliness was confirmed from the absence of any baseline (blank) peak during CV scan. 2.4. Electrochemical polymerization of o-PDA-co-ANI copolymer film The monomer mixture solution (2.5 10 2 M o-PDA–1 10 2 M ANI) (see: Supplementary material for optimizing the o-PDA–ANI mixture ratio) prepared in pure water, contained 0.1 M Na2SO4 electrolyte. The pH of this solution was adjusted to 1.0 with concentrated H2SO4, and 5 mL of the solution was taken into the working cell. Before electrochemical polymerization process, nitrogen gas was bubbled through the cell for 5 min. Polymerization was performed via cyclic voltammetry within the potential range of ( 0.4 V to 1.2 V) with 50 mV s 1 scanning speed for 20 cycles. Finally, the modified GCE electrode was rinsed with distilled water to remove any unbound material from the surface. 2.5. Electrodeposition of Au nanoparticles on the copolymer-coated electrode
Synthetic mixtures of 20 mg L 1 TNTþ30 mg L 1 RDX and 15 mg L 1 each of TNTþHMX were prepared, with reference to the composition of real mixture samples of Comp B (TNTþRDX mixture) and Octol (TNTþ HMX mixture). The results found for synthetic mixtures using the developed CV method were compared with those found for 50 mg L 1 Comp B and 50 mg L 1 Octol real samples. The developed method (using GC/P(o-PDA-co-ANI)-Aunano electrode) was applied to synthetic {TNTþ DNT} mixture solutions of nitroaromatics by fixing the concentration of one component while changing the other, i.e. while fixing the concentration of the TNT component at 5, 10, 20, 30 or 40 mg L 1 in a binary mixture, the concentration of DNT component was varied as 5, 10, 20, 30 or 40 mg L 1 making 25 synthetic mixtures altogether. The same procedure was repeated by fixing the concentration of DNT in the mixture while changing the concentration of TNT in the defined range (i.e. by fixing DNT at 5, 10, 20, 30 or 40 mg L 1 and changing TNT at 5, 10, 20, 30 or 40 mg L 1, making 25 synthetic mixtures altogether); the current intensities at the characteristic peak potentials of TNT and DNT of the described mixtures were recorded. All analyses performed for the {TNTþDNT} synthetic mixtures were repeated with an uncoated GC electrode. For deconvolution of current contributions at peak potentials, a multiple linear regression approach was used, as described earlier [35]. The peak current intensity (I) yielded by a mixture at a characteristic potential can be expressed as a linear combination (with coefficients m and n) of the concentrations (C) of TNT and DNT such that; I¼ mCTNT þnCDNT þconstant. The developed CV method was applied to synthetic mixture of 30 mg L 1 TNTþ70 mg L 1 tetryl and real tetrytol sample of 100 mg L 1 with the use of both GC/P(o-PDA-co-ANI)-Aunano electrode and uncoated GC electrode, followed by the comparison of results. Determinations of 50 mg L 1 2-ADNT and 50 mg L 1 4-ADNT with the developed method, using GC/P(o-PDA-co-ANI)-Aunano electrode, were performed both in the presence and absence of 50 mg L 1 TNT. 2.8. Recommended procedure for the determination of amino-dinitritoluenes in the presence of TNT
Glassy carbon electrode was coated with Au nanoparticles using the cyclic voltammetry method with 0.04% (w/v) HAuCl4 (2.5 ml)þ 0.1 M H2SO4 (2.5 ml) solutions via electrochemical deposition process. Coating was performed within the ( 0.4 V to 0.4 V) range at 50 mV s 1 scanning speed. The amount of deposited gold particles onto the P(o-PDA-co-ANI)/GCE surface was determined by checking cycle numbers, the optimal value of which was set at 40 cycles. The golden-colored electrode formed in this way was named as GC/P(o-PDA-co-ANI)-Aunano. (CV characterization, impedance measurements and SEM images of the modified electrode are given in Supplementary material).
The determination scheme (modified Griess procedure) for the ADNTþNED method is summarized as follows: add 0.4 mL 2-ADNT or 4-ADNT sample (in acetonitrile or acetone)þ0.6 mL waterþ 0.4 mL of 0.5 M H2SO4 þ0.2 mL of 1% NaNO2 in this order; wait for 5 min, add 0.2 mL of 3% sulfamic acid, and wait for 10 min; then add 0.5 mL of 0.75% NEDþ1.7 mL ethanol; measure the optical absorbance A565 against a reagent blank for 4-ADNT, and A504 against a reagent blank for 2-ADNT.
2.6. Electrochemical studies with nitroaromatic energetic materials
2.9. Assay in the presence of common soil ions
The working solutions of 2.5–40 mg L 1 TNT and 2–40 mg L 1 DNT were prepared from the respective stock solutions and transferred to the measurement cell; 4.75 mL of 0.5 M NaCl was added, and the
Assay in the presence of common soil ions: TNT (20 mg L 1) was determined in the presence of 10-fold concentrations of common ions. The potential interferent ions (Cl , SO42 , NO3 , CO32 , Mg2 þ ,
184
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
Ca2 þ , K þ , Cu2 þ , Pb2 þ , Fe3 þ ) were tested in 95:5 (v/v) H2O–acetonitrile mixture solution. 2.10. Validation of the CV method against GC–MS using real postblast debris samples The real post-blast residual samples were collected (by macroscopic examination of the debris for detecting any visible residues of particulate matter that can be analyzed directly) from the resulting craters after detonations of TNT and DNT individually. Prior to detonation, samples of range soil were collected and analyzed as blank. After detonation of each explosive, 10 g samples of soil from the resulting craters were appropriately collected, leached with 50 ml acetonitrile for 15 min in a stoppered flask, allowed to settle for 10 min, and filtered through a 0.45 mm filter. The samples were analyzed using the proposed voltammetric and reference GC–MS methods, directly in the leachate and in 10-fold acetonitrile-diluted solution, respectively. The literature GC–MS conditions [36] were used as starting parameters. Splitless injection was done with hold time of 2 min and injection volume of 0.5 mL. Injector temperature was 170 °C. The starting temperature of the oven was set to 70 °C (which was maintained for 3 min), followed by a temperature increase to 185 °C at a rate of 8 °C min 1 and a subsequent increase to 250 °C at a rate of 25 °C min 1. The final temperature was maintained for 5 min. GC– MS interface temperature was 260 °C, MS source temperature 200 °C, and scan m/z range was 29–400. Carrier gas was helium, with a flowrate of 1.5 mL min 1. The statistical comparisons between the findings of the proposed voltammetric and reference GC–MS methods were made with the aid of Student t- and F-tests after normalization of initial concentrations of samples.
3. Results and discussion Effectiveness of the GC electrodes may decrease in time because of contamination, giving the opportunity of longer use with coating. In this study, the electrode was initially coated with o-PDA-co-ANI copolymer film, and surface stability was further increased with Au-NP coating. It was reported in literature that an o-PDA-coated GC electrode could be used to determine H2O2 and nitrite in the presence of copper nanoparticles [37]. The successive electropolymerization from o-PDA solution on the surface of Au colloid load electrode, followed by immersion in the Au colloid overnight, enabled the formation of a bilayer of nano-Au/polymerized o-PDA on the electrode surface; in this way, the electrode was assembled with sufficient Au particles for antibody immobilization to build a biosensor from which the Prussian blue film was not easy to leak [38]. Electrode surface can also be coated with NPs via electrodeposition, as demonstrated by Lin et al. for manufacturing a gold nanoparticle-modified glassy carbon electrode (AuNPs/GCE), and then applied for the analysis of three oil-soluble antioxidants [39]. 3.1. Fabrication of o-PDA–ANI copolymer film on GCE From the cyclic voltammogram for the polymerization of o-PDA–ANI monomer mixture (Fig. 1), it can be deduced that the increase in current intensity in the first cycle may be attributed to cation radicals accompanying the oxidation of monomers. During the following cycles, the initially formed cation radicals combine with one another to form dimers and oligomers, the accumulation of which leads to the formation of the polymer film on the electrode; the typical redox behavior pertaining to the polymer is observed at lower potentials (at 0.37 V). The amount of polymer accumulated on the electrode increased with the number of cycles
Fig. 1. The cyclic voltammogram for the polymerization of 2.5 10 2 M o-PDA þ 1 10 2 M ANI monomer mixture in solution (first cycle is shown in the inset figure).
leading to a proportional increase in current intensities, accompanying a decrease in the amount of monomers left in solution. Poly(o-PDA)-coated GCE was electroactive toward energetic materials, but the coating was physically unstable and could be easily peeled off the surface after several measurements. On the other hand, when PANI was used alone on the electrode, it would exhibit stability at the expense of electrochemical non-reactivity (i.e. with no visible peaks). Only the copolymer (P[o-PDA-co-ANI]) exhibited both physical stability (i.e. cracks and peels were not visible on the electrode surface after several measurements) and electroactivity toward energetic materials. During the coating stage, irreproducible coating was observed when nitrogen gas was not purged through the working cell. If the redox behavior was tested in the supporting electrolyte medium not containing monomer, the optimal number of cycles for electropolymerization was set at 20, beyond which attenuation was noted in electrode sensitivity. The stability of the coated electrodes was tested in a supporting electrolyte medium not containing monomer (through which N2 was purged for 5 min) using a potential interval between 0.4 V and 1.2 V, scanning speed of 50 mV s 1, and 10 cycles. There was no notable difference between current intensities at the end of the first and tenth cycles. After this operation, the remaining monomeric parts on the coated electrode surface were completely polymerized, thereby making the electrode suitable for subsequent measurements. 3.2. Electrodeposition of Au-nanoparticles on o-PDA–ANI copolymercoated GCE The peak emerging at about 0.26 V in the first cycle (Fig. 2) is associated with the reduction of Au-nanoparticles on the surface of the o-PDA–ANI copolymer-coated electrode. With increasing cycles, more Au accumulated on the electrode leaving less free nanoparticles ((Au3 þ ) near by the electrode) in the medium, causing current intensities to decrease. The current value stabilized after 40 cycles, showing that the maximal quantity of Aunanoparticles that could accumulate on the electrode surface was attained. In each regeneration step of the modified electrode, the preservation of the quantity of matter accumulated on the surface should be checked by considering the values of the initial and final cycles. There was no need for either recoating or cleaning the electrode before each measurement, because the once-prepared GC/P(o-PDA-co-ANI)-Aunano electrode could be successively used for all measurements during the day. The electrode characterization using a combination of CV, impedance spectroscopy and SEM images is given in Figs. S1–S4 (Supplementary material).
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
Fig. 2. Voltammograms regarding Au-nanoparticles on o-PDA–ANI copolymercoated GCE (first cycle is shown in the inset figure).
Nitroaromatic compounds were detected through their π-acceptor/ donor interactions with o-phenylenediamine–aniline functionalities on the modified electrode surface. The enhanced sensitivities were achieved through π–π and charge-transfer (CT) interactions between the electron-deficient nitroaromatic compounds and s-/π-donor amine/aniline groups linked to gold nanoparticles (Au-NPs), possibly providing increased binding and preconcentration onto the modified GC-electrodes based on literature reports [24–40,41].
Fig. 3. Cyclic voltammograms of TNT, DNT, and tetryl recorded with GC/P(o-PDA-co-ANI)-Aunano electrode at the concentrations of 5 and 40 mg L 1 for each explosive. Table 1 LOD and LOQ values and relative standard deviations (RSD, %) of analytical results for the tested nitro-explosive analytes.
LODc LOQc RSD, % a b
3.3. Cyclic voltammetric responses of the electrode to TNT, DNT and tetryl The tested nitro-aromatics were all electro-active, and the reduction reactions (i.e. Ar-NO2-Ar-NHOH-Ar-NH2) for these compounds hypothesized to occur during CV scans in the negative potential range are listed in Table S1 of Supplementary material. The characteristic potentials of TNT appeared at 0.56 V, 0.75 V and 0.90 V, of DNT at 0.71 V and 0.88 V, and finally of tetryl at 0.4 V, 0.57 V and 0.74 V. Cyclic voltammograms of all three explosives at the concentrations of 5 and 40 mg L 1 can be seen in Fig. 3. Calibration of TNT with highest sensitivity at 0.9 V potential gave a linear dependence of current intensity versus concentration;
I−0.9 V = 0.4761CTNT + 15.41 (r = 0.999)
(1)
where I 0.9 V is the peak current intensity (mA) at 0.9 V and CTNT is the TNT concn. (mg L 1). Calibration of DNT could be made using the peak currents at 0.88 V and 0.71 V at close sensitivity levels, with linear dependence of current intensity on concentration yielding the following equations:
I−0.88
= 0.7041CDNT + 16.51 (r = 0.998)
(2 )
I−0.71 V = 0.6211CDNT + 16.21 (r = 0.998)
(3)
V
where I 0.88 V and I 0.71 V are the current intensities (mA) at 0.88 V and 0.71 V potentials, respectively, and CDNT is DNT concn. (mg L 1). Linear working curve for tetryl at 0.74 V was
I−0.74 V = 0.2075Ctetryl + 14.06 (r = 0.997)
(4 )
where I 0.74 V is the current intensity (mA) at peak potential, and Ctetryl is the concn. of tetryl (mg L 1).
185
c
TNTa
TNTb
DNTa
DNTb
Tetryla
Tetrylb
2.1 7.0 0.63–2.1
4.1 13.5 0.8–1.85
1.3 4.0 0.66–2.47
4.0 13.4 0.93–1.85
3.8 12.6 0.7–2.41
6.4 21.3 0.97–2.3
Results obtained with the use of GC/P(o-PDA-co-ANI)-Aunano electrode. Results obtained with the use of bare electrode, GCE. In mg L 1 units.
(I–C equations of secondary importance for TNT (at 0.75 V and 0.56 V) and tetryl (at 0.57 V) are given in Supplementary material). The slope–intercept values of the calibration lines (A¼mCþ n) as well as the analytical figures of merit, i.e. limit of detection (LOD)¼3sbl/m and limit of quantification (LOQ)¼10sbl/m, where sbl denotes the standard deviation of a blank and m is the slope of the calibration curve, are comparatively given in Table 1 for TNT, DNT, and tetryl, measured by the bare GC electrode and the derivatized GC/P(o-PDA-co-ANI)-Aunano electrode. As can be seen from the data in Table 1, higher sensitivities (i.e. lower LOD and LOQ values) could be obtained with the derivatized electrode compared to those with GCE. 3.4. Analytical results for synthetic and real mixtures of nitroaromatic explosives The real mixture samples of Comp B and Octol (at 50 mg L 1 each) and synthetic mixtures prepared according to the known compositions of these real samples were analyzed to yield the voltammograms in Fig. 4. From the data in Fig. 4, no reduction peak could be assigned to 30 mg L 1 RDX, and the synthetic mixture of (20 mg L 1 TNTþ 30 mg L 1 RDX) yielded an almost coincident voltammogram with 50 mg L 1 Comp B (containing 60% RDXþ39% TNT) showing that the RDX component of the synthetic mixture had no effect on the voltammogram of TNT (at 20 mg L 1). Likewise, the same observation was made for the HMX component (at 35 mg L 1) of Octol, because it had no effect on the voltammogram of the corresponding amount of TNT (at 15 mg L 1). With the use of the developed method, the recovery of the electroactive component (TNT) from Comp B and Octol samples were in the range of 99.2–102.1% and 98.7–99.4%, respectively. For the analysis of TNT and DNT mixtures with the GC/P (o-PDA-co-ANI)-Aunano modified electrode, the peak potentials of
186
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
Fig. 4. The voltammograms obtained for 30 mg L 1 RDX, synthetic mixture of (20 mg L 1 TNT þ 30 mg L 1 RDX), 50 mg L 1 Comp B (containing 60% RDXþ39% TNT), 35 mg L 1 HMX, synthetic mixture of (15 mg L 1 TNTþ 35 mg L 1 HMX), and 50 ppm Octol (70% HMX þ 30% TNT) with the use of GC/P(o-PDA-co-ANI)-Aunano electrode.
0.75 V and 0.90 V for TNT and of 0.71 V and 0.88 V for DNT were so close to each other that additive peaks were obtained for these compounds at the indicated potentials, and therefore, individual quantification could not be directly made. On the other hand, another characteristic potential for TNT at 0.56 V was significantly affected from the DNT current at 0.71 V, also making the latter alternative useless for individual determination. Therefore, for such mixtures, the use of the unmodified (bare) electrode GCE, is proposed. The characteristic potentials of TNT (10–50 mg L 1) appeared at 0.52 V, 0.73 V and 0.84 V with the bare GCE, and the TNTspecific peak potential at 0.52 V gave linear concentrationdependent peak currents with the following calibration equation:
I−0.52 V = 0.2951CTNT + 8.266 (r = 0.996)
(5)
where I 0.52 V is the peak current intensity (mA) at 0.52 V and CTNT is the TNT concn. (mg L 1). The characteristic potentials of DNT (10–50 mg L 1) appeared at 0.65 V and 0.78 V with the bare GCE, where a DNT-specific linear calibration could be performed at 0.65 V with the equation
I−0.65 V = 0.2968CDNT + 8.996 (r = 0.999)
(6)
where I 0.65 V is the peak current intensity (mA) at 0.65 V and CDNT is the DNT concn. (mg L 1). In the analysis of (TNT þDNT) mixtures with GCE, the current intensities of TNT at 0.73 V and 0.84 V and of DNT at 0.65 V and 0.78 V gave additive results. During these measurements, the current values at the 0.52 V peak of TNT were not affected from those of DNT. On the other hand, there was little interference to the 0.65 V peak of DNT from the characteristic 0.73 V peak of TNT. Thus, multiple linear regression analysis at these two peaks enabled individual quantification of (TNTþDNT) mixtures. Treatment of approximately Z90% of data yielded the two equations for calculating TNT and DNT components of synthetic mixtures measured at the reduction potentials of 0.52 and 0.65 V, respectively.
I−0.52 V−8.735 = 0.296CTNT + 0.0144CDNT
(7)
I−0.65 V−8.964 = 0.305CTNT + 0.297CDNT
(8)
When the above equations were applied to unknown solutions with peak currents measured at the indicated potentials, the
relative errors in calculation of TNT and DNT constituents varied in the ranges of 2–6% and 2–3%, respectively. Multiple regression treatment of data revealed that the DNT contribution to TNT peak current at 0.52 V produced a greater effect in calculations than that of TNT to DNT peak current at 0.65 V. In the analysis of (TNT þtetryl) mixtures with the modified electrode, the current intensities at the characteristic TNT potentials at 0.56 V and 0.75 V interfered with those of tetryl at 0.4 V, 0.57 V and 0.74 V at close values, hindering the individual determination of mixture constituents. Therefore, the use of bare GCE was preferred for analyzing such mixtures. The characteristic potentials of tetryl (at 20–100 mg L 1) with GCE appeared at 0.34 V, 0.51 V and 0.68 V, and the peak potential at 0.34 V gave linear concentration-dependent peak currents (Fig. 5). Thus, calibration of tetryl was based on peak currents at 0.34 V and an excellent linear dependence was found with the following equation:
I−0.34 V = 0.1625Ctetryl + 9.363 (r = 0.999)
(9)
where I 0.34 V is the peak current intensity (mA) at 0.34 V and Ctetryl is the tetryl concn. (mg L 1). With the use of the bare GCE, the currents of TNT at the characteristic potentials of 0.73 V and 0.84 V gave additive intensities with those of tetryl at 0.51 V and 0.68 V, and the current intensity at 0.34 V due to tetryl was not interfered by any of the TNT peaks in mixtures. Exploiting these observations, the tetryl constituent of synthetic and real mixtures (such as tetrytol) with TNT could be effectively determined, where the recovery of tetryl from such mixtures at 0.34 V varied between 99.4% and 101.9%. 3.5. Determination of aminodinitrotoluenes (ADNTs) in the presence of TNT Environmental metabolization products of TNT are formed by the bacterial reduction of nitro groups to the analogic amino derivatives. Major bacterial degradation products are 2-amino-4, 6-dinitrotoluene (2-ADNT), 4-amino-2,6-dinitrotoluene (4-ADNT), 2,4-diamino-6-nitrotoluene (2,4-ADNT) and 2,6-diamino-4-nitrotoluene (2,6-ADNT) [42]. Most common of these products in the natural environment are 2-amino-4,6-dinitrotoluene (2-ADNT) and 4-amino-2,6-dinitrotoluene (4-ADNT). These two products have more toxicity than the parent compound, TNT [43]. Therefore, there was a need to devise a method for analyzing aminodinitrotoluenes (ADNTs) along with TNT. As electroanalytical differentiation of ADNTs from TNT could not be made with the modified GC/P(o-PDA-co-ANI)-Aunano electrode using the solutions
Fig. 5. The voltammograms recorded with the bare GC electrode, for 30 mg L 1 TNT, synthetic mixture of (70 mg L 1 tetrylþ 30 mg L 1 TNT), and 100 mg L 1 tetrytol (70% tetryl þ30% TNT).
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
2-ADNT, 4-ADNT and TNT (at 50 mg L 1 each), a modified Griess method was developed for colorimetric analysis. The spectrophotometric method developed for (ADNT þTNT) mixture analysis is based on the selective diazotization of ADNT with nitrite, followed by NED coupling at pH 2 to produce a chromophore, where TNT did not yield a similar color reaction (because only aromatic amines and not nitro-groups were diazotized with nitrite). Since excessive nitrite could be decomposed for 10 min with sulfamic acid, new diazotization did not take place at the stage of NED addition, and the quantity of either the diazonium salt or its diazo-coupling product was equivalent to that of ADNT [44]. As for the individual determination of the TNT constituent of the mixture, dicyclohexyl amine (DCHA) charge-transfer complexation (DCHA being used in the form of a colorimetric sensor) selectively responded to TNT [45] without being affected by ADNTs. Thus, ADNT and TNT could be selectively determined in a mixture by means of separate color reactions. The azo-coupled final chromophores derived from acetonitrile solutions of 2-ADNT and 4-ADNT absorbed at 505 and 524 nm, respectively, the latter giving the much stronger response. Thus, the sensitivity was much lower for 2-ADNT (linear concentration range: 49.8–249 mg L 1) than for 4-ADNT (linear range: 2.48–49.8 mg L 1). TNT did not interfere, as it did not respond to diazotization in mixtures. The linear calibration equations for ADNTs were A¼1.8 10 2 C4-ADNT þ5.2 10 2 (r¼ 0.998) and A¼1.0 10 3 C2-ADNT þ4.6 10 2 (r¼ 0.996), where C4-ADNT and C2-ADNT are the mg L 1 concentrations of the corresponding ADNTs in acetonitrile. In the case of synthetic mixtures of 4-ADNT, 2-ADNT and TNT, the 4-ADNT constituent could be selectively determined in the 10–140 mg L 1 linear range with the calibration equation A¼ 6.0 10 3 C4-ADNT þ0.017 (r ¼0.997), without interference from either TNT (non-reacting) and 2-ADNT (having much lower sensitivity, with a weak absorbance at a concentration level of 103 mg L 1).
187
for Cu2 þ and Fe3 þ was 2.5-fold, yielding TNT recoveries of 93– 101%.
3.7. Analytical results for real post-blast debris samples In gas chromatograms, the standard TNT peak of the pure sample emerged at a retention time of 13.48 min, but for real postblast debris, this peak emerged at 13.46 min, and the characteristic ions produced from the analyte had m/z: 63, 89, 134, 180 and 210 in the mass spectra. The DNT peak of the pure sample emerged at a retention time of 11.21 min, but for real post-blast debris, this peak emerged at 11.22 min, and the characteristic ions produced from the analyte had m/z: 44, 63, 89, 119 and 165 in the mass spectra. TNT and DNT were analyzed from 1 to 10 mg L 1 and 1 to 10 mg L 1 working solutions, respectively, and quantitative analysis for each sample was performed using full scan mode. For each sample, the mean value of three repetitive GC injections was used for calculations. The calibration equations for TNT and DNT between Peak Area and concentration were
Peak Area = 1.79 × 105CTNT − 6.59 × 104 (CTNT: TNT concn. in mg L−1, r = 0.9992)
(10)
Peak Area = 1.0 × 106CDNT − 2.55 × 104 (CDNT: DNT concn. in mg L−1, r = 0.9999)
(11)
The leachate from post-blast debris sample was either directly analyzed (by voltammetry) or diluted (analyzed by GC–MS); measurement of TNT with the proposed voltammetric method and with GC–MS in the 1:10 (v/v) acetonitrile-diluted leachate gave 36.957 1.67 mg L 1 and 39.171.4 mg L 1 (normalized) concentrations, respectively, in the main solution (N¼5 repetitive measurements for each method). When the same measurement was repeated for DNT, the post-blast debris leachate gave 37.2471.41 and 38.771.24 mg L 1 (normalized) concentrations with voltammetry and GC–MS, respectively. Since 10 g post-blast soil sample was leached with 50 mL acetonitrile, the nitro-explosive contents of the original solid samples could be traced back to yield 18876 and 19177 mg kg 1 TNT and DNT, respectively. There was no potential shift for the voltammetric determination of TNT and DNT in the post-blast sample leachates. Statistical comparison between the results of the proposed and reference GC–MS procedures applied to post-blast residues of {TNTþDNT} mixture in acetonitrile was made on N ¼5 repetitive analyses, essentially showing no significant difference between results (Table 2).
3.6. Interference of common soil ions to the electrochemical determination of TNT The common soil ions of Cl , SO42 , NO3 , CO32 , Mg2 þ , Ca2 þ , and K þ did not interfere with the determination of 20 mg L 1 TNT at 10-fold concentration levels, with TNT recoveries ranging between 92% and 101%. When electro-active ions such as Cu2 þ , Pb2 þ , and Fe3 þ existed at 10-fold levels, their reduction peaks overlapped with the characteristic potentials of TNT, requiring the pre-masking of these heavy metal cations with EDTA. In the presence of EDTA as masking agent (added at 1:1 mol/mol), Pb2 þ could be tolerated at 5-fold levels of TNT, whereas the same ratio
Table 2 Statistical comparison of the proposed voltammetric method with GC–MS for the determination of TNT and DNT constituents of a real post-blast residue sample. Sample/analyte
Method
Mean concn., mg L 1
Std. dev. (r)
S
Real post-blast residue/TNT Real post-blast residue/TNT Real post-blast residue/DNT Real post-blast residue/DNT
Voltammetric
36.95
1.67
GC–MSc
39.1
Voltammetric c
GC–MS
ta,b
ttableb
Fb
Ftableb
–
–
–
–
–
1.4
1.54
2.207
2.306
1.43
6.39
37.24
1.41
–
–
–
–
–
38.7
1.24
1.33
1.738
2.306
1.29
6.39
a,b
a 2 S ¼ {(n1 1)s12 þ(n2 1)s22}/(n1 þ n2 2) and t ¼(ā1 ā2)/{S(1/n1 þ 1/n2)1/2}, where S is the pooled standard deviation, s1 and s2 are the standard deviations of the two populations with sample sizes of n1 and n2, and sample means of ā1 and ā2 respectively (t has (n1 þn2 2) degrees of freedom); here, n1 ¼n2 ¼ 5. b Statistical comparison made on paired data produced with proposed and reference methods; the results given only on the row of the reference method. c Leachate analysis results of post-blast residual samples was compared on a normalized concentration basis, since GC–MS determinations were made in 1:10 (v/v) acetonitrile-diluted leachates.
188
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
4. Conclusions
Appendix A. Supplementary materials
Electrochemical sensors provide a fast, inexpensive and highly sensitive option for miniaturized and automated analysis of complex samples of electroactive explosive residues in the field. Advances in the development of nanomaterials show a strong potential to create electrochemical sensors for detecting explosives. Based on these premises, a novel electrochemical sensor for nitroaromatic explosives (basically comprising TNT, 2,4-DNT and tetryl) was developed by coating a glassy carbon electrode with a copolymerized chargetransfer agent (o-PDA and ANI) and gold nanoparticles to end up with a modified electrode named as GC/P(o-PDA-co-ANI)-Aunano. This modified GC electrode did not require pre-coating before each measurement (thereby reducing the analytical costs and time) and proved to increase the sensitivity of determinations of nitro-aromatic explosives by decreasing the LOD levels several-fold. Additionally, the linear response and repeatability/reproducibility of the modified electrode were considerably better than those of the parent GCE. On the other hand, possible mixtures of nitro-explosives were analyzed by using combinations of bare GC and modified electrodes and selecting certain critical reduction potentials where necessary. As opposed to most electroanalytical techniques that cannot differentiate between the constituents of real explosive mixtures, the developed methods were successfully applied to the analysis of nitroaromatics in military explosives, namely comp B, octol, and tetrytol. The advantages of the designed electrode have been demonstrated as stability, sensitivity, reproducibility, and repeated use (i.e. having a service life of 30 days) without damage. As aminodinitrotoluenes (ADNTs) emerging as TNT metabolites in the environment gave overlapping voltammograms with TNT, a Griess-based colorimetric procedure was developed for determining ADNTs along with TNT. The developed electrochemical sensing method was statistically validated against GC–MS on real samples, including postblast debris. The developed sensor would also pose a cheap and rapid alternative for the confirmation of analytical findings of criminological police laboratories running highly sophisticated and hyphenated chromatographic analyses, because decision making procedures following criminologic events strongly require high throughput screening of great numbers of complex samples in the field.
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.talanta.2015.02.059
Acknowledgements The authors wish to express their gratitude to the Ministry of National Defense, Office of Technical Services, and to Machinery & Chemistry Industries Institution (MKEK) for the donation of nitroand composite explosive samples. The authors extend their thanks to Istanbul University Research Fund for the support given to Standard Research Project-14412. Additionally, the authors thank TUBITAK (Turkish Scientific and Technical Research Council) for their support to the Research Project 112T792. The authors wish to sincerely thank Professor Esma Sezer from Istanbul Technical University for her valuable discussions.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45]
L. Senesac, T.G. Thundat, Mater. Today 11 (2008) 28–36. H. Bai, C. Li, G. Shi, Sens. Actuators B 130 (2008) 777–782. J.S. Caygill, F. Davis, S.P.J. Higson, Talanta 88 (2012) 14–29. M. Galik, A.M. O’Mahony, J. Wang, Electroanalysis 23 (2011) 1193–1204. X. Cetó, A.M. O’Mahony, J. Wang, M. del Valle, Talanta 107 (2013) 270–276. C.K. Chua, M. Pumera, L. Rulíšek, J. Phys. Chem. C 116 (2012) 4243–4251. K. Bratin, P.T. Kissinger, R.C. Briner, C.S. Bruntlett, Anal. Chim. Acta 130 (1981) 295–311. S.-Y. Ly, D.-H. Kim, M.-H. Kim, Talanta 58 (2002) 919–926. J. Wang, M. Pumera, Talanta 69 (2006) 984–987. M. O’Mahony, G. Valdés‐Ramírez, J.R. Windmiller, A. Samek, J. Wang, Electroanalysis 24 (2012) 1811–1816. N. Pon Saravanan, S. Venugopalan, N. Senthilkumar, P. Santhosh, B. Kavita, H. G. Prabu, Talanta 69 (2006) 656–662. J. Wang, S. Thongngamdee, Anal. Chim. Acta 485 (2003) 139–144. J. Wang, S. Thongngamdee, D. Lu, Electroanalysis 18 (2006) 971–975. X. Fu, R.F. Benson, J. Wang, D. Fries, Sens. Actuators B 106 (2005) 296–301. K.C. Honeychurch, J.P. Hart, P.R.J. Pritchard, S.J. Hawkins, N.M. Ratcliffe, Biosens. Bioelectron. 19 (2003) 305–312. J.-C. Chen, J.-L. Shih, C.-H. Liu, M.-Y. Kuo, J.-M. Zen, Anal. Chem. 78 (2006) 3752–3757. J. Wang, F. Lu, D. MacDonald, J. Lu, M.E.S. Ozsoz, K.R. Rogers, Talanta 46 (1998) 1405–1412. K. Cizek, C. Prior, C. Thammakhet, M. Galik, K. Linker, R. Tsui, A. Cagan, J. Wake, J. La Belle, J. Wang, Anal. Chim. Acta 661 (2010) 117–121. M. Lahav, E. Katz, A. Doron, F. Patolsky, I. Willner, J. Am. Chem. Soc. 121 (1999) 862–863. M. Lahav, E. Katz, I. Willner, Langmuir 17 (2001) 7387–7395. T.L. Panasyuk, V.M. Mirsky, S.A. Piletsky, O.S. Wolfbeis, Anal. Chem. 71 (1999) 4609–4613. F. Shi, Z. Liu, G. Wu, M. Zhang, H. Chen, Z. Wang, X. Zhang, I. Willner, Adv. Funct. Mater. 17 (2007) 1821–1827. K. Saha, S.S. Agasti, C. Kim, X. Li, V.M. Rotello, Chem. Rev. 112 (2012) 2739–2779. M. Riskin, R. Tel-Vered, T. Bourenko, E. Granot, I. Willner, J. Am. Chem. Soc. 130 (2008) 9726–9733. C. Xie, H. Li, S. Li, J. Wu, Z. Zhang, Anal. Chem. 82 (2010) 241–249. M.-C. Daniel, D. Astruc, Chem. Rev. 104 (2004) 293–346. A. Üzer, Z. Can, İ. Akın, E. Erçağ, R. Apak, Anal. Chem. 86 (2014) 351–356. A.M. O′Mahony, J. Wang, Anal. Methods 5 (2013) 4296–4309. J. Wang, S.B. Hocevar, B. Ogorevc, Electrochem. Commun. 6 (2004) 176–179. F. Wang, W. Wang, B. Liu, Z. Wang, Z. Zhang, Talanta 79 (2009) 376–382. C.- Yuan, C.- Chang, Y.- Hong, Y. Sung, Mater. Sci. 27 (2009) 509–520. A.D. Aguilar, E.S. Forzani, M. Leright, F. Tsow, A. Cagan, R.A. Iglesias, L.A. Nagahara, I. Amlani, R. Tsui, N.J. Tao, Nano Lett. 10 (2010) 380–384. R.G. Bozic, A.C. West, R. Levicky, Sens. Actuators B: Chem. 133 (2008) 509–515. S.A. Özkan, B. Uslu, P. Zuman, Anal. Chim. Acta 457 (2002) 265–274. A. Üzer, Ş. Sağlam, Y. Tekdemir, B. Ustamehmetoğlu, E. Sezer, E. Erçağ, R. Apak, Talanta 115 (2013) 768–778. M.E. Sigman, C.-Y. Ma, J. Forensic Sci. 46 (2001) 6–11. S.A. Kumar, P.-H. Lo, S.-M. Chen, J. Electrochem. Soc. 156 (2009) E118–E123. R. Yuan, L. Zhang, Q. Li, Y. Chai, S. Cao, Anal. Chim. Acta 531 (2005) 1–5. X. Lin, Y. Ni, S. Kokot, Anal. Chim. Acta 765 (2013) 54–62. D. Gao, Z. Wang, B. Liu, L. Ni, M. Wu, Z. Zhang, Anal. Chem. 80 (2008) 8545–8553. M. Riskin, R. Tel-Vered, O. Lioubaskevski, I. Willner, J. Am. Chem. Soc. 131 (2009) 7368–7378. J.C. Pennington, J.M. Brannon, Thermochim. Acta 384 (2002) 163–172. A.T. Bowen, J.M. Conder, T.W. La Point, Chemosphere 63 (2006) 58–63. G. Norwitz, P.N. Kellher, Anal. Chem. 53 (1981) 1238–1240. E. Erçağ, A. Üzer, R. Apak, Talanta 78 (2009) 772–780.
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
Electrochemical sensor for nitroaromatic type energetic materials using gold nanoparticles/poly(o-phenylenediamine–aniline) film modified glassy carbon electrode Şener Sağlam, Ayşem Üzer, Yasemin Tekdemir, Erol Erçağ, Reşat Apak n Istanbul University, Faculty of Engineering, Department of Chemistry, 34320 Istanbul, Turkey
art ic l e i nf o
a b s t r a c t
Article history: Received 11 November 2014 Received in revised form 23 February 2015 Accepted 28 February 2015 Available online 9 March 2015
In this work, a novel electrochemical sensor was developed for the detection of nitroaromatic explosive materials, based on a gold nanoparticle-modified glassy carbon (GC) electrode coated with poly(o-phenylenediamine–aniline film) (GC/P(o-PDA-co-ANI)-Aunano electrode). Nitroaromatic compounds were detected through their π-acceptor/donor interactions with o-phenylenediamine–aniline functionalities on the modified electrode surface. The enhanced sensitivities were achieved through π–π and charge-transfer (CT) interactions between the electron-deficient nitroaromatic compounds and s-/π-donor amine/aniline groups linked to gold nanoparticles (Au-NPs), providing increased binding and preconcentration onto the modified GC-electrodes. Selective determination of nitroaromatic type explosives in the presence of nitramines was enabled by o-PDA and reusability of the electrode achieved by Au-NPs. Calibration curves of current intensity versus concentration were linear in the range of 2.5–40 mg L 1 for 2,4,6-trinitrotoluene (TNT) with a detection limit (LOD) of 2.1 mg L 1, 2–40 mg L 1 for 2,4-dinitrotoluene (DNT) (LOD¼1.28 mg L 1), 5–100 mg L 1 for tetryl (LOD¼3.8 mg L 1) with the use of the GC/P(o-PDA-co-ANI)-Aunano electrode. For sensor measurements, coefficients of variation of intra- and inter-assay measurements were 0.6% and 1.2%, respectively (N¼5), confirming the high reproducibility of the proposed assay. Deconvolution of current contributions of synthetic (TNTþDNT) mixtures at peak potentials of constituents was performed by multiple linear regression analysis to provide high sensitivity for the determination of each constituent. Determination options for all possible mixture combinations of nitroaromatic explosives are presented in this work. The proposed methods were successfully applied to the analysis of nitroaromatics in military explosives, namely comp B, octol, and tetrytol. Method validation was performed against GC–MS on real post-blast residual samples containing both explosives. & 2015 Elsevier B.V. All rights reserved.
Keywords: Cyclic voltammetry Nitro-aromatic explosives Gold nanoparticles Poly(o-phenylenediamine–aniline) Modified electrode
1. Introduction The detection of explosives and explosive-related compounds has become a heightened priority in recent years for homeland security and counter-terrorism applications. There are many studies on the detection of explosives utilizing a multitude of different analytical techniques. The highly sophisticated current techniques for detecting/quantifying trace explosives include ion mobility spectrometry (IMS) and gas–liquid chromatography (GLC) coupled to mass spectrometry (MS); however, most of these devices are bulky, expensive, and require qualified personnel and time-consuming procedures [1]. In this regard, voltammetric techniques constitute cheaper and portable alternatives [2] if sufficient selectivity is attained with the aid of rapidly growing sensor electrodes. Naturally the aimed selectivity is n
Corresponding author. Tel.: þ 90 212 473 7028; fax: þ90 212 4737180. E-mail address: [email protected] (R. Apak).
http://dx.doi.org/10.1016/j.talanta.2015.02.059 0039-9140/& 2015 Elsevier B.V. All rights reserved.
closely related with the accessible potential range, number and variety of redox-active agents in the range of concern, and signal widths. Generally, the advantages of electrochemical systems in on site/in situ measurements of explosives comprise good sensitivity and selectivity, reasonable linear concentration ranges, minimal space and power requirements, and low-cost instrumentation. Electrochemical sensors offer good prospects for meeting the growing needs of explosives field detection. The most characteristic redox property of nitro-based explosives, namely the presence of easily reducible nitro-groups, ideally lends itself to electrochemical detection [3–6]. Different electrode materials found application for electrometric estimation of nitroaromatic and nitramine explosives. Such studies include detection of explosives at Hg-film electrodes [7–9], glassy carbon electrodes (GCEs) [10,11] and carbon fiber electrodes (CFE) [12–14]. There have also been a range of studies on the voltammetric detection of explosives at screen-printed carbon electrodes [4–10,15–18]. The use of imprinted polymers as functional
182
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
sensing matrices may suffer from several limitations, e.g., thick polymer matrices may be required for increased sensitivity due to the low density of imprinted sites. But as a disadvantage, this may lead to slow binding of the analytes to the recognition sites (long analysis time intervals) and to inefficient communication between the binding sites and the transducers. In fact, several studies suggested the use of imprinted monolayers [19,20], multilayers [21,22], and thin films to overcome these difficulties. Nanoparticles possess distinct physical and chemical properties that make them perfect scaffolds for the preparation of novel chemical sensors [23]. The unique electronic and optical properties of metallic and semiconductor nanoparticles (NPs) added new dimensions to the area of sensors [24]. An example is the use of polymer particles and nanoparticles that, when bound to an explosive molecule, change one of their measurable properties resulting in specific detection [1]. In a template strategy for increasing the selectivity/sensitivity of analyte recognition by molecular imprinting, electropolymerizable p-aminothiophenol (ATP) was first assembled on the AuNPs at the glassy carbon electrode surface by the formation of Au–S bonds, and subsequently, a phosphorothioate-group pesticide template was further assembled onto the monolayer of ATP through the hydrogen-bonding interaction between the free amino group on AuNPs and the organophosphorus pesticide [25]. The aggregation of Au NPs as a result of sensing reactions, giving rise to interparticle-coupled surface plasmon resonance (SPR) absorption wavelength shifts, was used for the development of colorimetric sensors [26]. For example, our research group optimized the differential hydrolysis of nitramine explosives (RDX and HMX) into nitrite, and developed a selective sensor by reacting 4-aminothiophenol (4-ATP) modified Au NPs with the nitrite analyte and naphthylethylene diamine (NED) as coupling agent for azo-dye formation, giving rise to a bathochromic shift in SPR absorption wavelength of further aggregated Au NPs [27]. The use of nanosensors has the potential to satisfy the requirements of an effective platform for trace detection of explosives. The combination of electroanalysis with nanomaterials as electrode modifiers also allows the easy integration of nanosensors with advances in conventional detection platforms, exploiting the physical and chemical benefits of interactions at the nanomaterial electrode surface [28]. Carbon nanotube-modified glassy carbon electrode for adsorptive stripping voltammetric determination of ultratrace TNT has been claimed to reach sub-μg L 1 levels of detection with linear response between 0.1 and 1 mg L 1 [29]. By utilizing an electrochemical sensor based on polyaniline deposition as a nanofibrous composite with a polypeptide, the electron-donating amino groups of polypeptide showed charge-transfer interactions with electrondeficient TNT in close proximity to the conducting polymer, allowing TNT determination in solution by adsorptive stripping voltammetry [30]. Other researchers used a composite of polyaniline and carbon nanotubes for the detection of nitroaromatic vapors containing TNT, 2,6-DNT and picric acid, where the amine group of polyaniline forms a charge transfer complex with nitro-energetic compounds leading to signal enhancement [31]. Another conducting polymer used within a vapor sensor is poly(3,4-dioxyethylene thiophene) (PEDOT) which could be deposited to form a conductive nanowire between two gold electrodes. Coating this electrical junction with an ionic liquid allowed the system to adsorb TNT vapor from the air, where the electrochemical reduction of TNT as well as the conductance change of the polymer nanojunction arising from the reduction product could be simultaneously measured [32]. In general, many sensing methods reported in literature do not demonstrate the ability to effectively differentiate and simultaneously sense explosive compounds in natural mixtures such as contaminated soil and groundwater, because overlap regions of electroactivity in nitro-explosive compounds result in cross-contribution of current from one compound to another, requiring
deconvolution of contributions from each compound of structural similarity [33]. Thus, the aim of this study is to develop a simple, sensitive and selective nanoparticle-based electrochemical sensor for the simultaneous quantitative estimation of selected nitro-aromatic energetic materials. Spectrophotometric method development was reverted to only for the HPLC-free analysis of environmental TNT metabolites (e.g., monoamino-dinitrotoluenes) basically yielding similar voltammograms to those of TNT. Thus, synthetic and real mixture analysis at low cost was feasible for nitro-aromatic energetic materials using simple voltammetry with a specially designed electrode, supported with colorimetry where required.
2. Experimental 2.1. Chemicals, solutions and instruments The explosive materials TNT (2,4,6-trinitrotoluene), DNT (2,4dinitrotoluene), tetryl (2,4,6-trinitrophenylmethylnitramine), RDX (1,3,5-trinitro-1,3,5-triazacyclohexane), HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazosin), Comp B composite explosive (containing 60% RDX, 39% TNT, and 1% wax), octol composite explosive (containing 70% HMX and 30% TNT), and tetrytol composite explosive (containing 30% TNT, 70% tetryl), were kindly supplied by Makine Kimya Endustrisi Kurumu (MKEK: Machinery & Chemistry Industries Institution) through the supervision of Milli Savunma Bakanligi, Teknik Hizmetler Daire Baskanligi (Ministry of National Defense, Office of Technical Services) of Turkey. 2-amino-4,6-dinitrotoluene (2-ADNT) and 4-amino-2,6-dinitrotoluene (4-ADNT) were supplied from Dr. Ehrenstorfer as the reference standard samples. The solvent used in preparation and dilution of explosive solutions was HPLCgrade extra pure acetonitrile (E. Merck). The alumina slurry used for electrode cleaning was from Baikowski International Corp. (0.05 mm, Baikalox 0.05CR). The supporting electrolyte for conductivity was sodium chloride (Riedel-de Haën). For electrode cleanliness, isopropyl alcohol (Sigma-Aldrich), acetone and ethanol (both technical grade) were used. The electrode coating materials, o-phenylenediamine (o-PDA) and aniline (ANI), as well as the rest of the reagents, were supplied from E. Merck (Darmstadt, Germany). The standard stock solutions of individual explosive materials and military-purpose explosive mixtures (Comp B and octol) at 1000 mg L 1 concentrations were prepared in extra pure acetonitrile. Gold (III) chloride solution (99.99% trace metals basis, 30% by wt. in dilute HCl) was used for deposition of gold nanoparticles to GC electrode coated with o-PDA–ANI copolymer, and the 0.04% (w/v) working solution was prepared from this stock solution. Solutions used for the analysis of 2-ADNT and 4-ADNT samples: NaNO2 was prepared as 1% with water; N-(1-naphthyl)ethylenediamine (NED) weighed as 0.375 g, dissolved and diluted to 50 mL with water; sulfamic acid 3% solution prepared with water; stock solutions of the analytes, namely 990 mg L 1 4-amino-2,6-dinitrotoluene, 995 mg L 1 2-amino-4,6-dinitrotoluene, and 100 mg L 1 TNT prepared with acetonitrile. Voltammetric experiments were performed with a Gamry Instruments model Reference 600 potentiostat/galvanostat/ZRA interfaced to a PC computer and controlled Gamry Framework software. Glassy carbon (BASi stationary voltammetry electrodes; ø 1.6 mm, area 0.02 cm2) was used as the working electrode. A platinum (Pt) electrode and an Ag/AgCl electrode served as the auxiliary and reference electrodes, respectively. For validation of the proposed assay against a GC–MS system, real post-blast residual samples were analyzed with the use of a Thermo Scientific Trace gas chromatograph coupled with a DSQII mass spectrometer containing positive ion chemical ionization and quadrupole analyzer. GC was equipped with a Thermo TR-HT5 (15 m 0.25 mm, ID 0.1 mm, 5% phenyl polycarborane siloxane).
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
183
Calibration samples for GC–MS analysis were prepared from the 1000 mg L 1 stock solutions of TNT and DNT in acetonitrile with the use of the standard reference samples of ULTRA EPA-1243 TNT and Dr. Ehrenstorfer DNT. The working solutions of TNT at 3–10 mg L 1 and DNT at 1–10 mg L 1 were prepared from the indicated stock solutions using acetonitrile as diluent.
cell was filled to 5 mL with acetonitrile have a final solvent composition of 95% water 5% AcCN. Tetryl was voltammetrically analyzed in the 5.0–100 mg L 1 concentration range using a 5 mL volume of solution homogenized in an ultrasonic bath. CV was performed in a potential range from 0.2 V to 1.2 V, and the characteristic peak potentials for each nitroaromatic compound were determined.
2.2. Optimization of voltammetric method
2.7. Study of synthetic and real mixtures with deconvolution of current contributions from each constituent at peak potentials
Electrode type, scan rate, supporting electrolyte and concentration parameters have been individually examined. A glassy carbon (GC) electrode was selected as the working electrode, by which a substantial signal can be produced (as current intensity) at sufficiently separated reduction potentials. A scan rate of 50 mV s 1 was chosen, and NaCl was preferred as the supporting electrolyte. The electrolyte concentration was found to be optimal at 0.04 M. Mixtures of hexane, dichloromethane, acetonitrile, water, dioxane– methanol (1:1) and acetonitrile–methanol (1:1), acetonitrile–water (1:1) and acetonitrile–water (5–95%) were tested for solvent optimization to yield the latter acetonitrile–water mixture as optimal. 2.3. Glassy carbon electrode pre-treatment The GC electrode was polished with a suspension of alumina powder in the presence of pure water on a smooth polishing cloth [34] by circular movements for a few minutes, then washed with distilled water, and sonicated for 5 min in bidistilled water. Sonication of the electrode was repeated for another 5 min in isopropyl alcohol–acetonitrile (1:1, v/v) mixture. The efficiency of this procedure for electrode cleanliness was confirmed from the absence of any baseline (blank) peak during CV scan. 2.4. Electrochemical polymerization of o-PDA-co-ANI copolymer film The monomer mixture solution (2.5 10 2 M o-PDA–1 10 2 M ANI) (see: Supplementary material for optimizing the o-PDA–ANI mixture ratio) prepared in pure water, contained 0.1 M Na2SO4 electrolyte. The pH of this solution was adjusted to 1.0 with concentrated H2SO4, and 5 mL of the solution was taken into the working cell. Before electrochemical polymerization process, nitrogen gas was bubbled through the cell for 5 min. Polymerization was performed via cyclic voltammetry within the potential range of ( 0.4 V to 1.2 V) with 50 mV s 1 scanning speed for 20 cycles. Finally, the modified GCE electrode was rinsed with distilled water to remove any unbound material from the surface. 2.5. Electrodeposition of Au nanoparticles on the copolymer-coated electrode
Synthetic mixtures of 20 mg L 1 TNTþ30 mg L 1 RDX and 15 mg L 1 each of TNTþHMX were prepared, with reference to the composition of real mixture samples of Comp B (TNTþRDX mixture) and Octol (TNTþ HMX mixture). The results found for synthetic mixtures using the developed CV method were compared with those found for 50 mg L 1 Comp B and 50 mg L 1 Octol real samples. The developed method (using GC/P(o-PDA-co-ANI)-Aunano electrode) was applied to synthetic {TNTþ DNT} mixture solutions of nitroaromatics by fixing the concentration of one component while changing the other, i.e. while fixing the concentration of the TNT component at 5, 10, 20, 30 or 40 mg L 1 in a binary mixture, the concentration of DNT component was varied as 5, 10, 20, 30 or 40 mg L 1 making 25 synthetic mixtures altogether. The same procedure was repeated by fixing the concentration of DNT in the mixture while changing the concentration of TNT in the defined range (i.e. by fixing DNT at 5, 10, 20, 30 or 40 mg L 1 and changing TNT at 5, 10, 20, 30 or 40 mg L 1, making 25 synthetic mixtures altogether); the current intensities at the characteristic peak potentials of TNT and DNT of the described mixtures were recorded. All analyses performed for the {TNTþDNT} synthetic mixtures were repeated with an uncoated GC electrode. For deconvolution of current contributions at peak potentials, a multiple linear regression approach was used, as described earlier [35]. The peak current intensity (I) yielded by a mixture at a characteristic potential can be expressed as a linear combination (with coefficients m and n) of the concentrations (C) of TNT and DNT such that; I¼ mCTNT þnCDNT þconstant. The developed CV method was applied to synthetic mixture of 30 mg L 1 TNTþ70 mg L 1 tetryl and real tetrytol sample of 100 mg L 1 with the use of both GC/P(o-PDA-co-ANI)-Aunano electrode and uncoated GC electrode, followed by the comparison of results. Determinations of 50 mg L 1 2-ADNT and 50 mg L 1 4-ADNT with the developed method, using GC/P(o-PDA-co-ANI)-Aunano electrode, were performed both in the presence and absence of 50 mg L 1 TNT. 2.8. Recommended procedure for the determination of amino-dinitritoluenes in the presence of TNT
Glassy carbon electrode was coated with Au nanoparticles using the cyclic voltammetry method with 0.04% (w/v) HAuCl4 (2.5 ml)þ 0.1 M H2SO4 (2.5 ml) solutions via electrochemical deposition process. Coating was performed within the ( 0.4 V to 0.4 V) range at 50 mV s 1 scanning speed. The amount of deposited gold particles onto the P(o-PDA-co-ANI)/GCE surface was determined by checking cycle numbers, the optimal value of which was set at 40 cycles. The golden-colored electrode formed in this way was named as GC/P(o-PDA-co-ANI)-Aunano. (CV characterization, impedance measurements and SEM images of the modified electrode are given in Supplementary material).
The determination scheme (modified Griess procedure) for the ADNTþNED method is summarized as follows: add 0.4 mL 2-ADNT or 4-ADNT sample (in acetonitrile or acetone)þ0.6 mL waterþ 0.4 mL of 0.5 M H2SO4 þ0.2 mL of 1% NaNO2 in this order; wait for 5 min, add 0.2 mL of 3% sulfamic acid, and wait for 10 min; then add 0.5 mL of 0.75% NEDþ1.7 mL ethanol; measure the optical absorbance A565 against a reagent blank for 4-ADNT, and A504 against a reagent blank for 2-ADNT.
2.6. Electrochemical studies with nitroaromatic energetic materials
2.9. Assay in the presence of common soil ions
The working solutions of 2.5–40 mg L 1 TNT and 2–40 mg L 1 DNT were prepared from the respective stock solutions and transferred to the measurement cell; 4.75 mL of 0.5 M NaCl was added, and the
Assay in the presence of common soil ions: TNT (20 mg L 1) was determined in the presence of 10-fold concentrations of common ions. The potential interferent ions (Cl , SO42 , NO3 , CO32 , Mg2 þ ,
184
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
Ca2 þ , K þ , Cu2 þ , Pb2 þ , Fe3 þ ) were tested in 95:5 (v/v) H2O–acetonitrile mixture solution. 2.10. Validation of the CV method against GC–MS using real postblast debris samples The real post-blast residual samples were collected (by macroscopic examination of the debris for detecting any visible residues of particulate matter that can be analyzed directly) from the resulting craters after detonations of TNT and DNT individually. Prior to detonation, samples of range soil were collected and analyzed as blank. After detonation of each explosive, 10 g samples of soil from the resulting craters were appropriately collected, leached with 50 ml acetonitrile for 15 min in a stoppered flask, allowed to settle for 10 min, and filtered through a 0.45 mm filter. The samples were analyzed using the proposed voltammetric and reference GC–MS methods, directly in the leachate and in 10-fold acetonitrile-diluted solution, respectively. The literature GC–MS conditions [36] were used as starting parameters. Splitless injection was done with hold time of 2 min and injection volume of 0.5 mL. Injector temperature was 170 °C. The starting temperature of the oven was set to 70 °C (which was maintained for 3 min), followed by a temperature increase to 185 °C at a rate of 8 °C min 1 and a subsequent increase to 250 °C at a rate of 25 °C min 1. The final temperature was maintained for 5 min. GC– MS interface temperature was 260 °C, MS source temperature 200 °C, and scan m/z range was 29–400. Carrier gas was helium, with a flowrate of 1.5 mL min 1. The statistical comparisons between the findings of the proposed voltammetric and reference GC–MS methods were made with the aid of Student t- and F-tests after normalization of initial concentrations of samples.
3. Results and discussion Effectiveness of the GC electrodes may decrease in time because of contamination, giving the opportunity of longer use with coating. In this study, the electrode was initially coated with o-PDA-co-ANI copolymer film, and surface stability was further increased with Au-NP coating. It was reported in literature that an o-PDA-coated GC electrode could be used to determine H2O2 and nitrite in the presence of copper nanoparticles [37]. The successive electropolymerization from o-PDA solution on the surface of Au colloid load electrode, followed by immersion in the Au colloid overnight, enabled the formation of a bilayer of nano-Au/polymerized o-PDA on the electrode surface; in this way, the electrode was assembled with sufficient Au particles for antibody immobilization to build a biosensor from which the Prussian blue film was not easy to leak [38]. Electrode surface can also be coated with NPs via electrodeposition, as demonstrated by Lin et al. for manufacturing a gold nanoparticle-modified glassy carbon electrode (AuNPs/GCE), and then applied for the analysis of three oil-soluble antioxidants [39]. 3.1. Fabrication of o-PDA–ANI copolymer film on GCE From the cyclic voltammogram for the polymerization of o-PDA–ANI monomer mixture (Fig. 1), it can be deduced that the increase in current intensity in the first cycle may be attributed to cation radicals accompanying the oxidation of monomers. During the following cycles, the initially formed cation radicals combine with one another to form dimers and oligomers, the accumulation of which leads to the formation of the polymer film on the electrode; the typical redox behavior pertaining to the polymer is observed at lower potentials (at 0.37 V). The amount of polymer accumulated on the electrode increased with the number of cycles
Fig. 1. The cyclic voltammogram for the polymerization of 2.5 10 2 M o-PDA þ 1 10 2 M ANI monomer mixture in solution (first cycle is shown in the inset figure).
leading to a proportional increase in current intensities, accompanying a decrease in the amount of monomers left in solution. Poly(o-PDA)-coated GCE was electroactive toward energetic materials, but the coating was physically unstable and could be easily peeled off the surface after several measurements. On the other hand, when PANI was used alone on the electrode, it would exhibit stability at the expense of electrochemical non-reactivity (i.e. with no visible peaks). Only the copolymer (P[o-PDA-co-ANI]) exhibited both physical stability (i.e. cracks and peels were not visible on the electrode surface after several measurements) and electroactivity toward energetic materials. During the coating stage, irreproducible coating was observed when nitrogen gas was not purged through the working cell. If the redox behavior was tested in the supporting electrolyte medium not containing monomer, the optimal number of cycles for electropolymerization was set at 20, beyond which attenuation was noted in electrode sensitivity. The stability of the coated electrodes was tested in a supporting electrolyte medium not containing monomer (through which N2 was purged for 5 min) using a potential interval between 0.4 V and 1.2 V, scanning speed of 50 mV s 1, and 10 cycles. There was no notable difference between current intensities at the end of the first and tenth cycles. After this operation, the remaining monomeric parts on the coated electrode surface were completely polymerized, thereby making the electrode suitable for subsequent measurements. 3.2. Electrodeposition of Au-nanoparticles on o-PDA–ANI copolymercoated GCE The peak emerging at about 0.26 V in the first cycle (Fig. 2) is associated with the reduction of Au-nanoparticles on the surface of the o-PDA–ANI copolymer-coated electrode. With increasing cycles, more Au accumulated on the electrode leaving less free nanoparticles ((Au3 þ ) near by the electrode) in the medium, causing current intensities to decrease. The current value stabilized after 40 cycles, showing that the maximal quantity of Aunanoparticles that could accumulate on the electrode surface was attained. In each regeneration step of the modified electrode, the preservation of the quantity of matter accumulated on the surface should be checked by considering the values of the initial and final cycles. There was no need for either recoating or cleaning the electrode before each measurement, because the once-prepared GC/P(o-PDA-co-ANI)-Aunano electrode could be successively used for all measurements during the day. The electrode characterization using a combination of CV, impedance spectroscopy and SEM images is given in Figs. S1–S4 (Supplementary material).
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
Fig. 2. Voltammograms regarding Au-nanoparticles on o-PDA–ANI copolymercoated GCE (first cycle is shown in the inset figure).
Nitroaromatic compounds were detected through their π-acceptor/ donor interactions with o-phenylenediamine–aniline functionalities on the modified electrode surface. The enhanced sensitivities were achieved through π–π and charge-transfer (CT) interactions between the electron-deficient nitroaromatic compounds and s-/π-donor amine/aniline groups linked to gold nanoparticles (Au-NPs), possibly providing increased binding and preconcentration onto the modified GC-electrodes based on literature reports [24–40,41].
Fig. 3. Cyclic voltammograms of TNT, DNT, and tetryl recorded with GC/P(o-PDA-co-ANI)-Aunano electrode at the concentrations of 5 and 40 mg L 1 for each explosive. Table 1 LOD and LOQ values and relative standard deviations (RSD, %) of analytical results for the tested nitro-explosive analytes.
LODc LOQc RSD, % a b
3.3. Cyclic voltammetric responses of the electrode to TNT, DNT and tetryl The tested nitro-aromatics were all electro-active, and the reduction reactions (i.e. Ar-NO2-Ar-NHOH-Ar-NH2) for these compounds hypothesized to occur during CV scans in the negative potential range are listed in Table S1 of Supplementary material. The characteristic potentials of TNT appeared at 0.56 V, 0.75 V and 0.90 V, of DNT at 0.71 V and 0.88 V, and finally of tetryl at 0.4 V, 0.57 V and 0.74 V. Cyclic voltammograms of all three explosives at the concentrations of 5 and 40 mg L 1 can be seen in Fig. 3. Calibration of TNT with highest sensitivity at 0.9 V potential gave a linear dependence of current intensity versus concentration;
I−0.9 V = 0.4761CTNT + 15.41 (r = 0.999)
(1)
where I 0.9 V is the peak current intensity (mA) at 0.9 V and CTNT is the TNT concn. (mg L 1). Calibration of DNT could be made using the peak currents at 0.88 V and 0.71 V at close sensitivity levels, with linear dependence of current intensity on concentration yielding the following equations:
I−0.88
= 0.7041CDNT + 16.51 (r = 0.998)
(2 )
I−0.71 V = 0.6211CDNT + 16.21 (r = 0.998)
(3)
V
where I 0.88 V and I 0.71 V are the current intensities (mA) at 0.88 V and 0.71 V potentials, respectively, and CDNT is DNT concn. (mg L 1). Linear working curve for tetryl at 0.74 V was
I−0.74 V = 0.2075Ctetryl + 14.06 (r = 0.997)
(4 )
where I 0.74 V is the current intensity (mA) at peak potential, and Ctetryl is the concn. of tetryl (mg L 1).
185
c
TNTa
TNTb
DNTa
DNTb
Tetryla
Tetrylb
2.1 7.0 0.63–2.1
4.1 13.5 0.8–1.85
1.3 4.0 0.66–2.47
4.0 13.4 0.93–1.85
3.8 12.6 0.7–2.41
6.4 21.3 0.97–2.3
Results obtained with the use of GC/P(o-PDA-co-ANI)-Aunano electrode. Results obtained with the use of bare electrode, GCE. In mg L 1 units.
(I–C equations of secondary importance for TNT (at 0.75 V and 0.56 V) and tetryl (at 0.57 V) are given in Supplementary material). The slope–intercept values of the calibration lines (A¼mCþ n) as well as the analytical figures of merit, i.e. limit of detection (LOD)¼3sbl/m and limit of quantification (LOQ)¼10sbl/m, where sbl denotes the standard deviation of a blank and m is the slope of the calibration curve, are comparatively given in Table 1 for TNT, DNT, and tetryl, measured by the bare GC electrode and the derivatized GC/P(o-PDA-co-ANI)-Aunano electrode. As can be seen from the data in Table 1, higher sensitivities (i.e. lower LOD and LOQ values) could be obtained with the derivatized electrode compared to those with GCE. 3.4. Analytical results for synthetic and real mixtures of nitroaromatic explosives The real mixture samples of Comp B and Octol (at 50 mg L 1 each) and synthetic mixtures prepared according to the known compositions of these real samples were analyzed to yield the voltammograms in Fig. 4. From the data in Fig. 4, no reduction peak could be assigned to 30 mg L 1 RDX, and the synthetic mixture of (20 mg L 1 TNTþ 30 mg L 1 RDX) yielded an almost coincident voltammogram with 50 mg L 1 Comp B (containing 60% RDXþ39% TNT) showing that the RDX component of the synthetic mixture had no effect on the voltammogram of TNT (at 20 mg L 1). Likewise, the same observation was made for the HMX component (at 35 mg L 1) of Octol, because it had no effect on the voltammogram of the corresponding amount of TNT (at 15 mg L 1). With the use of the developed method, the recovery of the electroactive component (TNT) from Comp B and Octol samples were in the range of 99.2–102.1% and 98.7–99.4%, respectively. For the analysis of TNT and DNT mixtures with the GC/P (o-PDA-co-ANI)-Aunano modified electrode, the peak potentials of
186
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
Fig. 4. The voltammograms obtained for 30 mg L 1 RDX, synthetic mixture of (20 mg L 1 TNT þ 30 mg L 1 RDX), 50 mg L 1 Comp B (containing 60% RDXþ39% TNT), 35 mg L 1 HMX, synthetic mixture of (15 mg L 1 TNTþ 35 mg L 1 HMX), and 50 ppm Octol (70% HMX þ 30% TNT) with the use of GC/P(o-PDA-co-ANI)-Aunano electrode.
0.75 V and 0.90 V for TNT and of 0.71 V and 0.88 V for DNT were so close to each other that additive peaks were obtained for these compounds at the indicated potentials, and therefore, individual quantification could not be directly made. On the other hand, another characteristic potential for TNT at 0.56 V was significantly affected from the DNT current at 0.71 V, also making the latter alternative useless for individual determination. Therefore, for such mixtures, the use of the unmodified (bare) electrode GCE, is proposed. The characteristic potentials of TNT (10–50 mg L 1) appeared at 0.52 V, 0.73 V and 0.84 V with the bare GCE, and the TNTspecific peak potential at 0.52 V gave linear concentrationdependent peak currents with the following calibration equation:
I−0.52 V = 0.2951CTNT + 8.266 (r = 0.996)
(5)
where I 0.52 V is the peak current intensity (mA) at 0.52 V and CTNT is the TNT concn. (mg L 1). The characteristic potentials of DNT (10–50 mg L 1) appeared at 0.65 V and 0.78 V with the bare GCE, where a DNT-specific linear calibration could be performed at 0.65 V with the equation
I−0.65 V = 0.2968CDNT + 8.996 (r = 0.999)
(6)
where I 0.65 V is the peak current intensity (mA) at 0.65 V and CDNT is the DNT concn. (mg L 1). In the analysis of (TNT þDNT) mixtures with GCE, the current intensities of TNT at 0.73 V and 0.84 V and of DNT at 0.65 V and 0.78 V gave additive results. During these measurements, the current values at the 0.52 V peak of TNT were not affected from those of DNT. On the other hand, there was little interference to the 0.65 V peak of DNT from the characteristic 0.73 V peak of TNT. Thus, multiple linear regression analysis at these two peaks enabled individual quantification of (TNTþDNT) mixtures. Treatment of approximately Z90% of data yielded the two equations for calculating TNT and DNT components of synthetic mixtures measured at the reduction potentials of 0.52 and 0.65 V, respectively.
I−0.52 V−8.735 = 0.296CTNT + 0.0144CDNT
(7)
I−0.65 V−8.964 = 0.305CTNT + 0.297CDNT
(8)
When the above equations were applied to unknown solutions with peak currents measured at the indicated potentials, the
relative errors in calculation of TNT and DNT constituents varied in the ranges of 2–6% and 2–3%, respectively. Multiple regression treatment of data revealed that the DNT contribution to TNT peak current at 0.52 V produced a greater effect in calculations than that of TNT to DNT peak current at 0.65 V. In the analysis of (TNT þtetryl) mixtures with the modified electrode, the current intensities at the characteristic TNT potentials at 0.56 V and 0.75 V interfered with those of tetryl at 0.4 V, 0.57 V and 0.74 V at close values, hindering the individual determination of mixture constituents. Therefore, the use of bare GCE was preferred for analyzing such mixtures. The characteristic potentials of tetryl (at 20–100 mg L 1) with GCE appeared at 0.34 V, 0.51 V and 0.68 V, and the peak potential at 0.34 V gave linear concentration-dependent peak currents (Fig. 5). Thus, calibration of tetryl was based on peak currents at 0.34 V and an excellent linear dependence was found with the following equation:
I−0.34 V = 0.1625Ctetryl + 9.363 (r = 0.999)
(9)
where I 0.34 V is the peak current intensity (mA) at 0.34 V and Ctetryl is the tetryl concn. (mg L 1). With the use of the bare GCE, the currents of TNT at the characteristic potentials of 0.73 V and 0.84 V gave additive intensities with those of tetryl at 0.51 V and 0.68 V, and the current intensity at 0.34 V due to tetryl was not interfered by any of the TNT peaks in mixtures. Exploiting these observations, the tetryl constituent of synthetic and real mixtures (such as tetrytol) with TNT could be effectively determined, where the recovery of tetryl from such mixtures at 0.34 V varied between 99.4% and 101.9%. 3.5. Determination of aminodinitrotoluenes (ADNTs) in the presence of TNT Environmental metabolization products of TNT are formed by the bacterial reduction of nitro groups to the analogic amino derivatives. Major bacterial degradation products are 2-amino-4, 6-dinitrotoluene (2-ADNT), 4-amino-2,6-dinitrotoluene (4-ADNT), 2,4-diamino-6-nitrotoluene (2,4-ADNT) and 2,6-diamino-4-nitrotoluene (2,6-ADNT) [42]. Most common of these products in the natural environment are 2-amino-4,6-dinitrotoluene (2-ADNT) and 4-amino-2,6-dinitrotoluene (4-ADNT). These two products have more toxicity than the parent compound, TNT [43]. Therefore, there was a need to devise a method for analyzing aminodinitrotoluenes (ADNTs) along with TNT. As electroanalytical differentiation of ADNTs from TNT could not be made with the modified GC/P(o-PDA-co-ANI)-Aunano electrode using the solutions
Fig. 5. The voltammograms recorded with the bare GC electrode, for 30 mg L 1 TNT, synthetic mixture of (70 mg L 1 tetrylþ 30 mg L 1 TNT), and 100 mg L 1 tetrytol (70% tetryl þ30% TNT).
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
2-ADNT, 4-ADNT and TNT (at 50 mg L 1 each), a modified Griess method was developed for colorimetric analysis. The spectrophotometric method developed for (ADNT þTNT) mixture analysis is based on the selective diazotization of ADNT with nitrite, followed by NED coupling at pH 2 to produce a chromophore, where TNT did not yield a similar color reaction (because only aromatic amines and not nitro-groups were diazotized with nitrite). Since excessive nitrite could be decomposed for 10 min with sulfamic acid, new diazotization did not take place at the stage of NED addition, and the quantity of either the diazonium salt or its diazo-coupling product was equivalent to that of ADNT [44]. As for the individual determination of the TNT constituent of the mixture, dicyclohexyl amine (DCHA) charge-transfer complexation (DCHA being used in the form of a colorimetric sensor) selectively responded to TNT [45] without being affected by ADNTs. Thus, ADNT and TNT could be selectively determined in a mixture by means of separate color reactions. The azo-coupled final chromophores derived from acetonitrile solutions of 2-ADNT and 4-ADNT absorbed at 505 and 524 nm, respectively, the latter giving the much stronger response. Thus, the sensitivity was much lower for 2-ADNT (linear concentration range: 49.8–249 mg L 1) than for 4-ADNT (linear range: 2.48–49.8 mg L 1). TNT did not interfere, as it did not respond to diazotization in mixtures. The linear calibration equations for ADNTs were A¼1.8 10 2 C4-ADNT þ5.2 10 2 (r¼ 0.998) and A¼1.0 10 3 C2-ADNT þ4.6 10 2 (r¼ 0.996), where C4-ADNT and C2-ADNT are the mg L 1 concentrations of the corresponding ADNTs in acetonitrile. In the case of synthetic mixtures of 4-ADNT, 2-ADNT and TNT, the 4-ADNT constituent could be selectively determined in the 10–140 mg L 1 linear range with the calibration equation A¼ 6.0 10 3 C4-ADNT þ0.017 (r ¼0.997), without interference from either TNT (non-reacting) and 2-ADNT (having much lower sensitivity, with a weak absorbance at a concentration level of 103 mg L 1).
187
for Cu2 þ and Fe3 þ was 2.5-fold, yielding TNT recoveries of 93– 101%.
3.7. Analytical results for real post-blast debris samples In gas chromatograms, the standard TNT peak of the pure sample emerged at a retention time of 13.48 min, but for real postblast debris, this peak emerged at 13.46 min, and the characteristic ions produced from the analyte had m/z: 63, 89, 134, 180 and 210 in the mass spectra. The DNT peak of the pure sample emerged at a retention time of 11.21 min, but for real post-blast debris, this peak emerged at 11.22 min, and the characteristic ions produced from the analyte had m/z: 44, 63, 89, 119 and 165 in the mass spectra. TNT and DNT were analyzed from 1 to 10 mg L 1 and 1 to 10 mg L 1 working solutions, respectively, and quantitative analysis for each sample was performed using full scan mode. For each sample, the mean value of three repetitive GC injections was used for calculations. The calibration equations for TNT and DNT between Peak Area and concentration were
Peak Area = 1.79 × 105CTNT − 6.59 × 104 (CTNT: TNT concn. in mg L−1, r = 0.9992)
(10)
Peak Area = 1.0 × 106CDNT − 2.55 × 104 (CDNT: DNT concn. in mg L−1, r = 0.9999)
(11)
The leachate from post-blast debris sample was either directly analyzed (by voltammetry) or diluted (analyzed by GC–MS); measurement of TNT with the proposed voltammetric method and with GC–MS in the 1:10 (v/v) acetonitrile-diluted leachate gave 36.957 1.67 mg L 1 and 39.171.4 mg L 1 (normalized) concentrations, respectively, in the main solution (N¼5 repetitive measurements for each method). When the same measurement was repeated for DNT, the post-blast debris leachate gave 37.2471.41 and 38.771.24 mg L 1 (normalized) concentrations with voltammetry and GC–MS, respectively. Since 10 g post-blast soil sample was leached with 50 mL acetonitrile, the nitro-explosive contents of the original solid samples could be traced back to yield 18876 and 19177 mg kg 1 TNT and DNT, respectively. There was no potential shift for the voltammetric determination of TNT and DNT in the post-blast sample leachates. Statistical comparison between the results of the proposed and reference GC–MS procedures applied to post-blast residues of {TNTþDNT} mixture in acetonitrile was made on N ¼5 repetitive analyses, essentially showing no significant difference between results (Table 2).
3.6. Interference of common soil ions to the electrochemical determination of TNT The common soil ions of Cl , SO42 , NO3 , CO32 , Mg2 þ , Ca2 þ , and K þ did not interfere with the determination of 20 mg L 1 TNT at 10-fold concentration levels, with TNT recoveries ranging between 92% and 101%. When electro-active ions such as Cu2 þ , Pb2 þ , and Fe3 þ existed at 10-fold levels, their reduction peaks overlapped with the characteristic potentials of TNT, requiring the pre-masking of these heavy metal cations with EDTA. In the presence of EDTA as masking agent (added at 1:1 mol/mol), Pb2 þ could be tolerated at 5-fold levels of TNT, whereas the same ratio
Table 2 Statistical comparison of the proposed voltammetric method with GC–MS for the determination of TNT and DNT constituents of a real post-blast residue sample. Sample/analyte
Method
Mean concn., mg L 1
Std. dev. (r)
S
Real post-blast residue/TNT Real post-blast residue/TNT Real post-blast residue/DNT Real post-blast residue/DNT
Voltammetric
36.95
1.67
GC–MSc
39.1
Voltammetric c
GC–MS
ta,b
ttableb
Fb
Ftableb
–
–
–
–
–
1.4
1.54
2.207
2.306
1.43
6.39
37.24
1.41
–
–
–
–
–
38.7
1.24
1.33
1.738
2.306
1.29
6.39
a,b
a 2 S ¼ {(n1 1)s12 þ(n2 1)s22}/(n1 þ n2 2) and t ¼(ā1 ā2)/{S(1/n1 þ 1/n2)1/2}, where S is the pooled standard deviation, s1 and s2 are the standard deviations of the two populations with sample sizes of n1 and n2, and sample means of ā1 and ā2 respectively (t has (n1 þn2 2) degrees of freedom); here, n1 ¼n2 ¼ 5. b Statistical comparison made on paired data produced with proposed and reference methods; the results given only on the row of the reference method. c Leachate analysis results of post-blast residual samples was compared on a normalized concentration basis, since GC–MS determinations were made in 1:10 (v/v) acetonitrile-diluted leachates.
188
Ş. Sağlam et al. / Talanta 139 (2015) 181–188
4. Conclusions
Appendix A. Supplementary materials
Electrochemical sensors provide a fast, inexpensive and highly sensitive option for miniaturized and automated analysis of complex samples of electroactive explosive residues in the field. Advances in the development of nanomaterials show a strong potential to create electrochemical sensors for detecting explosives. Based on these premises, a novel electrochemical sensor for nitroaromatic explosives (basically comprising TNT, 2,4-DNT and tetryl) was developed by coating a glassy carbon electrode with a copolymerized chargetransfer agent (o-PDA and ANI) and gold nanoparticles to end up with a modified electrode named as GC/P(o-PDA-co-ANI)-Aunano. This modified GC electrode did not require pre-coating before each measurement (thereby reducing the analytical costs and time) and proved to increase the sensitivity of determinations of nitro-aromatic explosives by decreasing the LOD levels several-fold. Additionally, the linear response and repeatability/reproducibility of the modified electrode were considerably better than those of the parent GCE. On the other hand, possible mixtures of nitro-explosives were analyzed by using combinations of bare GC and modified electrodes and selecting certain critical reduction potentials where necessary. As opposed to most electroanalytical techniques that cannot differentiate between the constituents of real explosive mixtures, the developed methods were successfully applied to the analysis of nitroaromatics in military explosives, namely comp B, octol, and tetrytol. The advantages of the designed electrode have been demonstrated as stability, sensitivity, reproducibility, and repeated use (i.e. having a service life of 30 days) without damage. As aminodinitrotoluenes (ADNTs) emerging as TNT metabolites in the environment gave overlapping voltammograms with TNT, a Griess-based colorimetric procedure was developed for determining ADNTs along with TNT. The developed electrochemical sensing method was statistically validated against GC–MS on real samples, including postblast debris. The developed sensor would also pose a cheap and rapid alternative for the confirmation of analytical findings of criminological police laboratories running highly sophisticated and hyphenated chromatographic analyses, because decision making procedures following criminologic events strongly require high throughput screening of great numbers of complex samples in the field.
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.talanta.2015.02.059
Acknowledgements The authors wish to express their gratitude to the Ministry of National Defense, Office of Technical Services, and to Machinery & Chemistry Industries Institution (MKEK) for the donation of nitroand composite explosive samples. The authors extend their thanks to Istanbul University Research Fund for the support given to Standard Research Project-14412. Additionally, the authors thank TUBITAK (Turkish Scientific and Technical Research Council) for their support to the Research Project 112T792. The authors wish to sincerely thank Professor Esma Sezer from Istanbul Technical University for her valuable discussions.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45]
L. Senesac, T.G. Thundat, Mater. Today 11 (2008) 28–36. H. Bai, C. Li, G. Shi, Sens. Actuators B 130 (2008) 777–782. J.S. Caygill, F. Davis, S.P.J. Higson, Talanta 88 (2012) 14–29. M. Galik, A.M. O’Mahony, J. Wang, Electroanalysis 23 (2011) 1193–1204. X. Cetó, A.M. O’Mahony, J. Wang, M. del Valle, Talanta 107 (2013) 270–276. C.K. Chua, M. Pumera, L. Rulíšek, J. Phys. Chem. C 116 (2012) 4243–4251. K. Bratin, P.T. Kissinger, R.C. Briner, C.S. Bruntlett, Anal. Chim. Acta 130 (1981) 295–311. S.-Y. Ly, D.-H. Kim, M.-H. Kim, Talanta 58 (2002) 919–926. J. Wang, M. Pumera, Talanta 69 (2006) 984–987. M. O’Mahony, G. Valdés‐Ramírez, J.R. Windmiller, A. Samek, J. Wang, Electroanalysis 24 (2012) 1811–1816. N. Pon Saravanan, S. Venugopalan, N. Senthilkumar, P. Santhosh, B. Kavita, H. G. Prabu, Talanta 69 (2006) 656–662. J. Wang, S. Thongngamdee, Anal. Chim. Acta 485 (2003) 139–144. J. Wang, S. Thongngamdee, D. Lu, Electroanalysis 18 (2006) 971–975. X. Fu, R.F. Benson, J. Wang, D. Fries, Sens. Actuators B 106 (2005) 296–301. K.C. Honeychurch, J.P. Hart, P.R.J. Pritchard, S.J. Hawkins, N.M. Ratcliffe, Biosens. Bioelectron. 19 (2003) 305–312. J.-C. Chen, J.-L. Shih, C.-H. Liu, M.-Y. Kuo, J.-M. Zen, Anal. Chem. 78 (2006) 3752–3757. J. Wang, F. Lu, D. MacDonald, J. Lu, M.E.S. Ozsoz, K.R. Rogers, Talanta 46 (1998) 1405–1412. K. Cizek, C. Prior, C. Thammakhet, M. Galik, K. Linker, R. Tsui, A. Cagan, J. Wake, J. La Belle, J. Wang, Anal. Chim. Acta 661 (2010) 117–121. M. Lahav, E. Katz, A. Doron, F. Patolsky, I. Willner, J. Am. Chem. Soc. 121 (1999) 862–863. M. Lahav, E. Katz, I. Willner, Langmuir 17 (2001) 7387–7395. T.L. Panasyuk, V.M. Mirsky, S.A. Piletsky, O.S. Wolfbeis, Anal. Chem. 71 (1999) 4609–4613. F. Shi, Z. Liu, G. Wu, M. Zhang, H. Chen, Z. Wang, X. Zhang, I. Willner, Adv. Funct. Mater. 17 (2007) 1821–1827. K. Saha, S.S. Agasti, C. Kim, X. Li, V.M. Rotello, Chem. Rev. 112 (2012) 2739–2779. M. Riskin, R. Tel-Vered, T. Bourenko, E. Granot, I. Willner, J. Am. Chem. Soc. 130 (2008) 9726–9733. C. Xie, H. Li, S. Li, J. Wu, Z. Zhang, Anal. Chem. 82 (2010) 241–249. M.-C. Daniel, D. Astruc, Chem. Rev. 104 (2004) 293–346. A. Üzer, Z. Can, İ. Akın, E. Erçağ, R. Apak, Anal. Chem. 86 (2014) 351–356. A.M. O′Mahony, J. Wang, Anal. Methods 5 (2013) 4296–4309. J. Wang, S.B. Hocevar, B. Ogorevc, Electrochem. Commun. 6 (2004) 176–179. F. Wang, W. Wang, B. Liu, Z. Wang, Z. Zhang, Talanta 79 (2009) 376–382. C.- Yuan, C.- Chang, Y.- Hong, Y. Sung, Mater. Sci. 27 (2009) 509–520. A.D. Aguilar, E.S. Forzani, M. Leright, F. Tsow, A. Cagan, R.A. Iglesias, L.A. Nagahara, I. Amlani, R. Tsui, N.J. Tao, Nano Lett. 10 (2010) 380–384. R.G. Bozic, A.C. West, R. Levicky, Sens. Actuators B: Chem. 133 (2008) 509–515. S.A. Özkan, B. Uslu, P. Zuman, Anal. Chim. Acta 457 (2002) 265–274. A. Üzer, Ş. Sağlam, Y. Tekdemir, B. Ustamehmetoğlu, E. Sezer, E. Erçağ, R. Apak, Talanta 115 (2013) 768–778. M.E. Sigman, C.-Y. Ma, J. Forensic Sci. 46 (2001) 6–11. S.A. Kumar, P.-H. Lo, S.-M. Chen, J. Electrochem. Soc. 156 (2009) E118–E123. R. Yuan, L. Zhang, Q. Li, Y. Chai, S. Cao, Anal. Chim. Acta 531 (2005) 1–5. X. Lin, Y. Ni, S. Kokot, Anal. Chim. Acta 765 (2013) 54–62. D. Gao, Z. Wang, B. Liu, L. Ni, M. Wu, Z. Zhang, Anal. Chem. 80 (2008) 8545–8553. M. Riskin, R. Tel-Vered, O. Lioubaskevski, I. Willner, J. Am. Chem. Soc. 131 (2009) 7368–7378. J.C. Pennington, J.M. Brannon, Thermochim. Acta 384 (2002) 163–172. A.T. Bowen, J.M. Conder, T.W. La Point, Chemosphere 63 (2006) 58–63. G. Norwitz, P.N. Kellher, Anal. Chem. 53 (1981) 1238–1240. E. Erçağ, A. Üzer, R. Apak, Talanta 78 (2009) 772–780.
