Optimization and characterization of electrochemically reduced graphene oxide formation

Figure 1A shows the UV spectra of GO and its change following the electrochemical reduction of GO. It is observed that the absorption peak of GO at 223 nm due to the π–π* transition of the C=C bond disappeared in ERGO. The amount of residual oxygenated functional groups in ERGO films is likely to vary depending on the experimental conditions, such as applied potential, reduction times, and the electrolyte used [25]. The process parameters for electrochemical reduction were optimized to develop better functioning electrodes. Raman spectroscopy has been frequently used as a reliable technique to optimize the electrochemical parameters for the synthesis of ERGO in terms of the intensity ratio of D- (disordered band) to G-band (graphitic band) (ID/IG). It measures the change in size of the sp2 ring clusters in a network of sp3- and sp2-bonded carbon [33]. Previous reports have indicated the possibility of converting GO to ERGO at different electrochemical parameters, but its effect on the ID/IG value have not been reported [25,29,33]. In this report, the pH value and buffer composition of the electrolyte were optimized to increase the deoxygenation of the GO sheet during ERGO formation. Figure 1B depicts three significant Raman peaks of GO at 1350 cm−1 for the D band (associated with defects in the sp2 lattice), 1596 cm−1 for the G band (due to vibrations of the hexagonal lattice), and 2700 cm−1 for the 2D band (related to numbers of layers in the graphene sheet). Table 1 shows the values of ID/IG at different electrolytic buffers during one-step electroreduction of GO at a constant potential of −0.9 V. The intensity of ID/IG predominantly increased for ERGO compared to the that of the as-prepared GO, which suggests a decrease in size of the sp2 domain due to extensive deoxygenation of the graphene sheets after electrochemical reduction. The comparative values of ID/IG (Table 1) also indicate that a higher defect in the sp2 domain was observed at acidic pH values of the electrolytic buffer during electrochemical reduction of GO. The highest value of ID/IG was found to be 1.454 for the conversion of ERGO using PBS (pH 4.5), which suggests the formation of higher defects between the graphene layers during electrochemical reduction [26,34]. Thus, 50 mM PBS, pH 4.5, has been chosen for an efficient conversion of GO to ERGO.

Figure 1: (A) UV–vis spectra of GO and ERGO. (B) Raman spectra for GO to ERGO conversion using different buffers as electrolytes. (C) FTIR spectra and (D) XRD patterns of GO and ERGO (PBS pH 4.5).

Table 1: Experimental sample table showing variation of Raman peak intensity ratio of ERGO using different electrolytes.

Sample in different pH ID IG ID/IG I2D I2D/IG GO 0.975524 1.01748 0.958765 0.43182 0.442654 acetate buffer, pH 4.5 1.00602 0.819277 1.227936 0.21328 0.212004 BR buffer, pH 4 1.01807 0.792169 1.285168 0.225525 0.221522 PBS buffer pH 4.5 1.00301 0.68976 1.454143 0.33042 0.329428 PBS buffer, pH 7 1.03313 0.978915 1.055383 0.29371 0.284291 PBS buffer, pH 9 1.03012 0.942771 1.092651 0.25699 0.249476

Figure 1C shows the characteristic FTIR spectra of GO and ERGO (in PBS, pH 4.5) to identify the change of functional groups due to electrolytic reduction of GO. The predominant characteristic absorption peaks of GO include a broad peak at 3426 cm−1 corresponding to the O–H stretching vibration originating from carboxyl groups. Besides, an intense peak at 1641 cm−1 was assigned to the C=O stretching of carboxyl and/or carbonyl groups, a sharp peak at 1387 cm−1 corresponding to a –OH bend, and a strong peak at 1068 cm−1 ascribed to an alkoxy and/or epoxy C–O stretching vibration. The significant reduction of the FTIR signal intensity of ERGO for –OH, –C=O, and –C–O suggests the successful formation of ERGO due to the electrochemical deoxygenation of GO, which corroborates the Raman analysis.

Figure 1D depicts a characteristic XRD peak of GO at 2θ = 9.98 (interplanar spacing = 0.843 nm) corresponding to 001 reflections. Two characteristic peaks of ERGO at 2θ = 21.15 (interplanar spacing = 0.413 nm) and 2θ = 29.65 (d-spacing = 0.343 nm) for the reflection of (020) and (200), respectively, confirm the successful formation of ERGO from GO.

Figure 2 represents the deconvoluted C 1s and O 1s XPS spectra of GO (Figure 2A and Figure 2B) and modified ERGO (Figure 2C and Figure 2D) electrodes. An asymmetric peak centered on ≈284.8 eV appeared due to the graphitic nature of GO and ERGO (Figure 2A and Figure 2B). Four different carbon types are observed from the deconvolution of the peaks shown in Figure 2A. They show an increase in binding energies evidencing the presence of C–OH, C–C, C–O–C, and C=O bonds in GO. The O 1s spectra of synthesized GO can be deconvoluted into three peaks, corresponding to contributions from carbonyl and carboxyl-type oxygen (531.4 eV), C–OH type (532.5 eV), and hydroxyl (533.6 eV). The intensity of the peaks is significantly reduced in RGO samples (Figure 2C and Figure 2D) compared to pristine GO, indicating considerable deoxygenation. The C 1s spectra of RGO (Figure 2C) can also be deconvoluted into four peaks at 284.7, 285.96, 292.8, and 295.7 eV. However, the relatively intense doublet appeared at 292.8 ± 0.1 eV and beyond 295 eV every time we performed the scan. Peaks in the range of 290 eV in these types of materials are mainly due to aromatic π–π* transitions. However, considering the intensity of the peak and our repeated measurements, we believe that the presence of a well-defined deconvoluted doublet peak beyond 290 eV corresponds to K 2p3/2 and K 2p1/2, which may have resulted from the contribution coming from the potassium salt present in the buffer during electrochemical conversion. The deconvoluted analysis of the peaks and the relative atomic percentages of GO and RGO are summarized in Supporting Information File 1, Table S2.

Figure 2: Deconvoluted XPS core-level spectrum of (A) C 1s, (B) O 1s for GO, (C) C 1s, and (D) O 1s for ERGO samples, respectively.

Figure 3A–D depicts TEM micrographs of as-prepared GO and synthesized ERGO at different pH values, indicating that the intensity of the electrons is attenuated by the platelets of graphene sheets with varying transparencies due to thickness variation [26,31]. Dark areas of the micrograph suggest thick stacking layers of GO and/or RGO with intercalated oxygen-containing functional groups. A few layers of graphene sheet in ERGO (in PBS, pH 4.5) have areas with higher transparency due to the exfoliation of stacking layers of GO. This suggests an increased surface area due to delamination of graphene layers (thickness of about one to a few layers) by electrochemical reduction. The high-resolution TEM of ERGO shows a d-spacing of 0.413 nm (Figure 3E), indicating a reduced graphite nature of GO. This confirms that the oxygen functional groups were removed from the graphene layers by electrochemical reduction of GO, decreasing the interspacing distance between graphene layers which facilitates electron transport. Thus, the conductivity of ERGO was enhanced compared to that of GO. The SEM micrograph of ERGO (Figure 3G) also shows graphene sheet exfoliated layers compared to GO (Figure 3F). The FESEM image also depicts the flaked nanostructure of RGO (Figure 3H).

Figure 3: (A) TEM images of as-synthesized GO, ERGO synthesized in different electrolytes: (B) PBS pH 4.5, (C) pH 7, and (D) pH 9.6. (E) HRTEM image of ERGO in PBS pH 4.5. (F) SEM micrographs of as-synthesized GO, (G) ERGO in PBS pH 4.5 and (H) FESEM of ERGO in PBS pH 4.5 at different magnifications.

Electrochemical characterization of the modified electrode

The electronic properties of graphene materials depend on the number of layers and the distance between the layers, which can be changed by a variation of the synthesis protocol to achieve a higher electroactive surface area and electrical conductivity. Figure 4A displays a higher oxidation/reduction peak current of Fe2+/3+ redox couple for the synthesized ERGO in PBS pH 4.5. It forms the highest electroactive surface area compared to other electrolytic buffers and pH values to prepare ERGO/GCE. To confirm the increase in the electroactive surface area of ERGO/GCE in comparison to bare GCE, CV was performed at different scan rates (10–300 mV/s) in 1.0 mM K3Fe(CN)6 as a redox probe (Supporting Information File 1, Figure S1). The electroactive surface areas were calculated according to Randles–Sevcik equation (Equation 1) [28,32]:

(1)

where Ip is the peak current (A), ν is the scan rate (V s−1), n is the number of electrons transferred (n = 1), Ac is the electrode active area (cm2), Dr is the diffusion coefficient (7.6 × 10−6 cm2·s−1), and C0 is the concentration of K3Fe(CN)6 (mol·cm−3). From the slope of the plot of Ip vs ν1/2, the effective surface area for bare GCE and ERGO/GCE was calculated to be 0.0707 and 0.121 cm2, respectively, which indicates that the effective electroactive surface area of ERGO has been improved by ≈71.14% due to exfoliation of graphene sheets.

Figure 4: (A) Cyclic voltammograms of GO and ERGO using different electrolytic buffers: Acetate buffer pH 4.7, PBS pH 4.6, PBS pH 6, PBS pH 7.4, PBS pH 9. (B) Nyquist plot of bare GCE, GO/GCE, and ERGO/GCE in the presence of 1 mM [Fe(CN)6]4−/3− containing 0.1 M KCl.

Electrochemical impedance spectroscopy was performed to investigate the electron transfer capability of ERGO (Figure 4B). Supporting Information File 1, Table S3 depicts the values of charge-transfer resistance (Rct), capacitance (Cdl), and Warburg impedance (W) of bare GCE, GO/GCE, and ERGO/GCE. The Nyquist plot of the bare GCE electrode depicts a semicircle with Rct of 4.692 Ω. A nearly straight line for ERGO with a negligible Rct (1.618 Ω) value suggests opened porous microstructures of ERGO, which makes the graphene sheets more accessible to the electrolyte. It also facilitates electron transfer and diffusion of ions during the electrochemical process [28,34].

Electrochemical behavior of parathion at modified nanosensors

Figure 5A depicts the CVs (first cycle) of bare GCE, GO/GCE, and ERGO/GCE in PBS (0.05 M, pH 7) in 10 μM PT. The CV of PT on bare GCE (inset of Figure 5A), shows a reduction peak at −0.65V and a little anodic peak due to autocatalysis of PT. A robust cathodic peak at −0.56 V and an anodic peak at +0.015 V were mainly observed on GO/GCE due to the absorption of PT through π stacking interaction between aromatic moieties of GO and the benzene ring of PT. In comparison, the highest cathodic/anodic peak was obtained at −0.58 and −0.05 V, respectively, for the electro-reduction/oxidation of PT on ERGO/GCE. The oxidation/reduction potentials of PT on ERGO/GCE were shifted to less positive values, effectively inhibiting the surface fouling caused by the reaction products, making ERGO-modified GCE more suitable for determining PT. The electrocatalytic ability of PT (10 µM) on the modified ERGO/GCE was investigated in PBS (pH 7) (Figure 5B) in the potential range from +0.2 to −1.0 V with a scan rate of 100 mV·s−1 and compared with the control group (bare GCE and GO/GCE). It is in good agreement with the literature reports that a sharp cathodic peak (Epc1) at −0.58 V was observed in the first cycle due to the reduction of the nitro group of PT (NO2–PT) to form its hydroxylamine derivatives (NHOH–PT) involving a four electron-transfer process as shown in Figure 5C [16-18,35]. An anodic peak appeared at −0.05 V in the backward segment of the first cycle, which is related to the oxidation of NHOH–PT to a nitroso group (NO–PT). This reversible two-electron-transfer process further generated a reduction peak (Epc1) at −0.11 V during the second potential scan of CV (Figure 5C). Nitroaromatic OPs such as parathion, methyl parathion, ethyl parathion, and fenitrothion, paraoxon exhibit this kind of electrocatalytic behavior, which is consistent with previous reports [21,23,24]. In this study, we chose the irreversible reduction peak of PT (NO2–PT to NHOH–PT) of the first cycle due to its suitability for important measurements in nanosensor applications.

Figure 5: (A) Cyclic voltammograms of PT (10 mM) with bare GCE (a), GO/GCE (b), and ERGO/GCE (c). (B) Electrochemical behavior of PT at ERGO/GCE. (C) Schematic diagram of the proposed electrochemical reaction of parathion at ERGO/GCE.

The amount of exfoliated GO dispersed on bare GCE is vital in optimizing the sensing matrix. Figure 6A depicts the CVs using a variation of deposited GO on bare GCE to prepare ERGO/GCE to measure the reduction and oxidation peak current for 10 μM PT in PBS, pH 7. Figure 6B shows that the highest reduction peak for 10 μM PT was obtained using 8 μL of GO to prepare modified ERGO/GCE. As the autocatalytic response for the electrochemical oxidation/reduction process is an absorption process, the accumulation time is another vital parameter to achieve the highest response for monitoring the amount of parathion residue in samples [35,36]. It has been shown in Figure 6 that as the immersion time of the modified electrode in a PT solution increased, the accumulation of PT on the electrode surface also enhanced. It was found that the highest peak current for 10 μM PT was obtained after immersion for 240 s in the PT solution. A further increase in the accumulation time was unaffected as the active area of the electrode surface was saturated (Figure 6C).

Figure 6: (A) Cyclic voltammograms of 10 μM PT in PBS, pH 7, with different amounts of GO deposited on bare GCE for the preparation of modified ERGO/GCE. (B) Peak current of cyclic voltammetry using 10 μM PT with different amounts of GO deposited on bare GCE for preparation of modified ERGO/GCE. (C) Peak response of 10 μM PT with variation of the accumulation time before voltammetric measurements.

Effect of scan rate and pH values on the electrolyte

The effect of scan rate on the reduction of PT at ERGO/GCE was investigated by applying different scan rates from 10 to 250 mV·s−1 (Figure 7A). The linear peak current increase with the scan rate suggests a surface-confined diffusion-controlled electrocatalytic process [21]. The slope of log Ipc as a function of log ν is 0.611 (>0.5), which confirms an adsorption-based reduction of PT on the modified electrode surface (Figure 7B). The reduction peak potential was shifted towards a more negative potential by increasing the scan rate. A linear equation of Ep as a function of log ν was represented as Ep = −0.088log ν − 0.694, with a correlation coefficient of (R²) 0.992. From the Laviron’s equation (Equation 2) for an irreversible reaction, Ep could be represented as

(2)

where ∝ is the transfer coefficient; n is the number of electron transfers; and R, T, F represent constants (R = 8.314 J·K−1, T = 298 K, F = 96480 C·mol −1). The standard redox potential (E°’) was found to be −0.523 V from the linear plot of Ep as a function of ν (Ep = −0.908ν − 0.523), at a scan rate 0 Vs−1. The standard heterogeneous rate constant (k°) for electrocatalysis of PT was 38.81 s−1. The value of ∝n was calculated to be 0.672, and the n value was found to be 0.954 (i.e., one-electron transfer process [37]).

Figure 7: (A) Cyclic voltammograms of ERGO/GCE under different scanning rates (10, 25, 40, 50, 65, 80, 100, 125, 150, 200, 250 mV·s−1) in PBS, pH 7, containing 10 μM PT. (B) Plot of the logarithm value of the reduction peak current as a function of the scanning rate (log Ipc as a function of log ν). (C) Plot of the reduction peak current of PT as a function of electrolyte pH. (D) Plot for the reduction (a) and oxidation (b) peak potential of PT (40 μM PT) as a function of the electrochemical cell pH, scan rate: 100 mV·s−1.

The protonation reaction influences the electrochemical reaction. Figure 7C shows the effect of pH on the electroreduction of PT (40 μM) by varying the pH values of PBS from 4.6 to 9. The irreversible reduction potential of PT was shifted towards a more negative potential as the pH values of the electrolyte varied from 4.5 to 9 (Figure 7D). The slope of the reduction peak (Epc) and oxidation peak (Epa) potential of PT as a function of pH is near −59 mV, which suggests that the same number of e− and H+ is involved in the reaction [38,39].

Optimization of square-wave voltammetry parameters

Square-wave voltammetry analysis is more accurate compared to an electrochemical method such as cyclic voltammetry and differential pulse voltammetry. It can minimize background current to obtain an intense, sharp, and well-defined peak of the targeted analyte at a particular potential. To obtain the maximum peak current, the parameters of SWV were optimized using 10 μM PT in PBS (pH 7). The variation of reduction peak current with accumulation potential (A), starting potential of scan (B), frequency (C), and pulse amplitude (D) are shown Figure 8A–D.

Figure 8: Relationship of stripping peak current in SWV measurements of 10 μM PT as a function of accumulation potential (A), starting potential of scan (B), frequency (C), and pulse amplitude (D). The optimized parameters for PT detection in SWV are: pulse amplitude: 100 mV, starting potential: 0.3 V, frequency: 25 Hz, accumulation potential: −0.1 V, and accumulation time: 240 s.

Analytical performance and selectivity of the proposed nanosensor

Figure 9A represents SWV curves obtained from the ERGO modified electrode for sequential additions of PT into phosphate buffer (pH 7). A sharp increase in the reduction peak current was observed for each addition after dipping the electrode into a particular solution for 240 s at an applied potential of −0.1 V (i.e., deposition potential). The peak was shifted to a negative potential as the concentration of PT enhanced, indicating a diffusion-controlled process [40]. The concentration-dependent linear plot depicts good linearity (Figure 9B, Figure 9C) with a calibration equation of Ip (μA) = 3.5735 [PT] + 12.018 (R² = 0.9871) for the range of 0.1–11 μM, Ip (μA) = 0.2916 [PT] + 3.7526 (R² = 0.9936) for 3–15 nM, Ip (μA) = 19.176 [PT] + 4.2723 (R² = 0.9367) for 0.03–0.15 nM. The corresponding sensitivity was found to be 50.5 μA·μM−1·cm−2 with a wide linear range for quantification of PT. The limit of detection (LOD = [(3 * standard deviation of blank)/slope of the lowest range of linear curve (i.e., 0.03–0.15 nM)] and limit of quantification (LOQ = [(10 * SD of blank)/slope]) were calculated as 10.9 pM and 36.5 pM, respectively, from the lower calibration equation [39,41].

Figure 9: (A) SWV response of ERGO/GCE electrode in electrolytes (PBS, 0.05 M, pH 7) with different concentrations of PT ranging from 0.1 to 3 μM. (B) Calibration plot of peak current as a function of PT concentration for a wide range (i.e., 3 × 10−5 to 11 µM, inset: linear regression curve for 0.1 to 11 µM) and (C) 0.03–0.15 nM. (D) Selectivity studies of PT (10 μM) detection with probable interfering substances such as Na+, K+, Fe2+, Fe3+, Cd2+, Ni2+, Mg2+, Mn2+, NH4+, Cl−, NO3−, CO32−, SO4−, CH3COO−, 4-nitrophenol, 2,4-dinitrophenol, acephate, chlorpyrifos, dicofol, and lindane, respectively using SWV and keeping all other parameters constant. Experimental conditions: pulse amplitude: 100 mV, SWV frequency: 25 Hz, starting potential: 0.3 V, accumulation potential: −0.1 V, and accumulation time: 240 s.

The selectivity of the proposed ERGO/GCE modified nanosensor (Figure 9D) was investigated in the presence of other possible substances in water and soil samples. Square-wave voltammetry measurements were performed in PBS (50 mM, pH 7.0) containing 10 μM PT along with some inorganic ions (e.g., Na+, K+, Fe2+, Fe3+, Cd2+, Ni2+, Mg2+, Mn2+, NH4+, Cl−, NO3−, CO32−, SO4−, CH3COO-), nitroaromatic compounds (e.g., 4-nitrophenol, 2,4-dinitrophenol), and other pesticides, such as acephate, chlorpyrifos, dicofol, lindane. As shown in Figure 9D, the modified electrode showed almost the same peak current when PT coexists with other substances. This indicates that the added substances have no significant effect on PT sensing in environmental samples. Other interfering OP (acephate, chlorpyrifos) and organochloride (dicofol, lindane) pesticides also did not significantly affect the response current of PT reduction as they have different redox potential and adsorption potential on the modified electrode surface.

Reproducibility, repeatability, and stability are essential parameters for practical applications of electrochemical nanosensors. Inter-assay measurements of 10 μM PT using five independent ERGO/GCE were performed, and a 3.4% relative standard deviation (RSD) was obtained for five replicate scans, indicating good reproducibility of the proposed nanosensor. Similarly, a single modified electrode exhibits good repeatability with an RSD of 1.81% for five repeated measurements performed in PBS (50 mM, pH 7.0) containing 10 μM PT.

The analytical performance of the ERGO/GCE, such as detection limit and linear range, are compared with previously reported modified electrodes for the detection of PT (Table 2) [16-18,36,42-44]. The proposed electrode showed better stability, sensitivity, and the lowest detection limit in comparison to previous reports [16-18,36,42-44]. As ERGO showed thermal and mechanical stability, ERGO/GCE could be a suitable electrode material for rapid screening of PT in actual samples.

Table 2: Experimental sample table for a comparative analytical performance of the proposed nanosensor with the reported nonenzymatic nanosensor.

Modified electrode Method Molecule Linear range
(μM) LOD
(nM) pH Samples Ref NanoTiO2-SAM/GCE DPV PT 0.05–10 10 PBS 5 cucumber, cabbage [17] NanoAg/Naf ion/GCE DPV PT 0.103–0.62 80 BR buffer, pH 2.56 water [16] MP 0.300–1.444 0.0874 ZrO2/MAS/Au SWV PT 0.017–3.4 2.8 pH 6, 0.1 M KCl vegetables, water [43] SPAN(sulfonated Pani)/GCE DPV PT 0.01–10 1.5 BR buffer 2.5 urine sample [42] ordered mesoporous carbon/GCE DPV PT 0.015–0.5 3.4 PBS 6 – [44] NiO-SPE DPV PT 0.1–30 24 0.05 M BR buffer, pH 6.0 urine, tomato [18] Al-doped mesoporous cellular foam (Al-MCF) SWV PT 0.01–1 mg/L 17.16 0.1 M KCl, pH 6.0 cabbage [36] ERGO/GCE SWV PT 3 × 10−5–11 10.9 × 10−3 PBS, pH 7 groundwater, soil, tomato, rice present work

To determine the storage stability, the electrocatalytic response of 10 μM PT was monitored in seven-day intervals for the first two months, and it retained about 96.17 ± 0.2% of its initial response. It was shown a consistent response to PT sensing during two months of storage. After that, the response was measured in intervals of 10 days, and 90.53 ± 0.3% of the initial response was retained after six months. This indicates good stability of the modified electrode at room temperature (Supporting Information File 1, Figure S2A). The feasibility of the proposed robust sensing platform was demonstrated by quantifying environmental samples such as groundwater and a soil sample from an agricultural land. Food (e.g., boiled rice) and vegetable (e.g., tomato collected from local market) samples were also analysed. As the concentration of PT in the collected samples was negligible, a specific amount of PT was spiked from the standard PT solution (1 mM). Supporting Information File 1, Figure S2B depicts the SWV response of groundwater spiked with 1.5, 2.5, and 5 μM PT, and detailed experimental results are shown in Table 3. The amount of spiked [PT] was monitored by the SWV response, and the results were validated using standard UV results. The UV spectra with increasing PT concentration (1–35 µM) are shown in Supporting Information File 1, Figure S3A. The concentration of PT in real samples was further calculated from the standard calibration curve obtained from UV spectra at 273 nm (Supporting Information File 1, Figure S3B). The quantitative spiked recoveries of PT ranged from 97.0–102.4%, with an RSD of 0.998–1.62%. In addition, the proposed method also depicts satisfactory relative error (1.53–3.96%) with standard absorption results for the quantification of PT in environmental samples.

Table 3: Experimental sample table for recovery studies of spiked PT in actual samples.

Real samples Added
(μM) Detected
(μM) Detected by UV–vis Recovery
(%) Relative error
(%) RSD
(%) ground water 1.5 1.46 1.52 97.33 3.947 0.998 2.5 2.48 2.52 99.20 1.587 1.518 5 5.12 5.2 102.4 1.538 1.369 soil 1 0.97 1.01 97.00 3.961 1.620 3 2.95 3.07 98.33 3.909 1.114 5 4.96 5.04 99.20 1.587 1.240 tomato 1 0.96 0.99 96.00 3.030 1.120 3