1 Introduction

Bisphenols (BPs) are a known group of endocrine disrupting compounds with adverse effects on human health and the environment [1]. Bisphenol A (BPA), 2, 2-bis (4-hydroxyphenyl) propane, is the most used among BPs with a wide range of applications, such as polycarbonate plastics manufacturing, with potent endocrine-disrupting activity. Bisphenol E (BPE), 1,1-bis(4-hydroxyphenyl) ethane, which has a structure very similar to BPA (Fig. 1) is another important industrial chemical used as monomer for the production of polycarbonate plastics and epoxy resins [2]. BPE has been reported to have similar acute toxicity and estrogenic activity to that of BPA [, , , , 2,3,4,5,6].

Fig. 1
figure 1

Bisphenols A and E structures

Full size image

Conventional chromatography, such as liquid chromatography (LC) or gas chromatography (GC), sometimes combined with mass-spectroscopy (MS); LC–MS and GC–MS; and/or UV–Vis spectroscopy are sensitive and selective analytical procedures for the routine determination of BPs. Disposable or simple manufactured tools have been widely used for environmental analysis in the last couple of years, in which electrochemical sensors or biosensors stand as good options due to their wide applicability and feasibility [, , 1, 7, 8]. Advances in the area of nanostructured materials, have made possible the design of a series of different electrochemical sensors, where nanocomposites are used to decorate conventional glassy carbon electrodes (GCE) or screen printed electrodes (SPE) with reported increased sensitivity and/or selectivity for the determination of BPs [, , , , , , , , , , 9,10,11,12,13,14,15,16,17,18,19]. Graphene oxide (GO) nanocomposites or hybrids with other materials [, , , , , , 20,21,22,23,2.6 Electrochemical tests

The electrochemical characterization of the sensor was made by use of cyclic voltammetry (CV) and differential pulse voltammetry (DPV). The CV voltammograms were recorded in a potential range from − 0.3 to 1.0 V using a scan rate of 50 mVs−1. The supporting electrolyte initially used was a 0.1 M sodium phosphate buffer solution, pH 5. A typical experiment was performed as follows: 20 mL of 50 µM BPE solution in phosphate buffer was poured into the voltammetric cell. Before measurements, nitrogen gas was purged into the solution for 5 min, and the atmosphere was kept under nitrogen flow throughout the measurements to avoid oxygen gas dissolution. The pH dependence was studied using Britton–Robinson buffer to reduce the influence of supporting electrolyte on current values, as it can cover a broad range of pH values. Prior to DPV measurements, the freshly made electrode was submitted to 20 CV’s cycles in phosphate buffer, pH 5, to verify its stabilization (reproducible baselines). Following, 15 mL of various concentrations of analyte solutions were added into the cell. A pre-conditioning step before DPV measurement was performed, in which the solution was purged with nitrogen gas while under vigorous stirring at an applied potential of 0.5 V for 90 s. The DPV recording proceeded in static state with no further purging. The increment potential, pulse period, pulse amplitude, and pulse width were adjusted at 0.075 mV, 0.4 s, 0.05 V, and 0.075 s, respectively. All measurements were performed in triplicate for statistical mean.

2.6.1 Recovery tests

Tap water from Liverpool John Moores University was collect in a 200 mL glass container, filtered using 45 µm filters and stored at 4 °C for further analysis. Measurements were performed using DPV optimized parameters. Recovery tests carried out using the standard addition method with concentrations within the linear range obtained for the analytical calibration.

3 Results and discussion

3.1 BPE cyclic voltammetry at bare and modified GCE

The search for a more sensitive electrode consisted in comparative experiments using CV to study the oxidation of BPE at bare GCE and at a series of modified electrodes: BEA and BEA derivatives only; GO only; GO derivatives; and finally, the BEA and BEA derivatives mixtures with GO. Figure 2 shows the oxidative part of CVs curves arising from 50 µM BPE in 0.1 M phosphate buffer solution (pH 5) for some of the modified electrodes in comparison with bare GCE. To illustrate the analyte oxidation current despite background, the baselines (only buffer) were subtracted from the analyte’s measurements and only the oxidative scan presented (see Supplementary Info, Fig. S3, to check the raw data full voltammograms). Oxidation peaks are noted in all curves, and the absence of reduction peaks (not shown) indicated that the process was irreversible (see Fig. S3). The results are similar to what is known for the analogous bisphenols BPA [, , 10, 53, 54], BPF [, 12, 13], and TBrBPA [, 55, 56].

Fig. 2
figure 2

The CV curves of 50 µM BPE solutions in 0.1 M phosphate buffer solution (pH 5) at bare or modified GCE electrodes (only oxidative scan shown). Scan rate: 50 mVs−1, under N2 atmosphere. Background was first recorded in the absence of analyte and subtracted from the respective voltammograms. (see Fig. S3 for the respective unsubtracted full voltammograms)

Full size image

Figure 2 highlights the results for the bare GCE in comparison with GCE covered with BEA/APTMS/GA/NF, GO/NF, GO/BEA/APTMS/GA/NF, and GO/APTMS/GA/NF. Lower anodic peak current intensity (Ip) was noted for modified BEA based electrode in absence of GO. However, GCE/GO/NF presented an increase of Ip, and the electrode GCE/BEA/APTMS/GA/GO/NF had a remarkable increase of around seven folds higher than the bare GCE. All components of the GCE/BEA/APTMS/GA/GO/NF electrode were tested separately, and none by itself gave a similar Ip to the BEA/APTMS/GA/GO/NF modified electrode. Intermediate electrodes of the constituent parts gave analogous or slightly higher Ip in comparison to the bare GCE electrode (Fig. S3). To evaluate whether the functional groups arising from APTMS and GA would lead to higher Ip in the absence of zeolite, the functionalization of GO with both reagents was carried out. Despite significant peak shift, the maximum current was not much different than that found for the bare electrode (Fig. 2), and significantly lower than unmodified GO, which indicated that the addition of APTMS and GA to GO had a negative influence on the electrode sensitivity.It is reasonable to attribute the increased current output to a coactive interaction between the coating components, which led to a more favorable BPE interaction with the electrode surface. Overall, Fig. 2 shows that the modified zeolite combined with GO was required for increased BPE electrooxidation.

Some reasonable understanding as to the causes of these observations can be made based on the chemistry of modified zeolites. It is clear that the nanozeolite functional surface groups have a crucial role on the sensor electrochemical performance, consistent with previous reports of very complex processes occurring where functional groups were included to improve the analyte interaction with the electrode and concomitantly the quantification [, , , , 19, 20, 22, 23, 57]. Based on our previous experience of zeolites functionalization [32], where alkoxysilane functionalization was performed, followed by glutaraldehyde cross-linking, zeolitic surface electrostatic changes were observed by zeta potential measurements. The as-synthetized zeolites presented negative zeta potentials, while APTMS functionalized zeolites showed slight positive potentials, and importantly, the cross-linked materials showed potentials that were substantially more positive. The GCE/BEA/APTMS/GO/NF did not show much improvement on BPE oxidation; however, the GCE/BEA/APTMS/GA/GO/NF electrode exceeded expectations. Thus, it suggests that the zeolite electrostatic surface features play a determinate role on analyte determination, and this could not be achieved by solely functionalizing GO. It appears that a more positively charged zeolite surface results in more favorable electrode interaction with analyte. A balance between GO and nanozeolite specific electrostatic features is necessary, and only achieved after using the GA based material, with surfaces more positively charged.

Additionally, the modified electrode did not show any fouling caused by the deposition of BPE oxidation products; phenomena that is commonly reported in bare or modified electrodes for the analysis of BPs [, , , , , , 15, 19, 22, 27, 28, 30, 31], resulting from the deposition of oxidation products hindering further analyte oxidation. This is usually an effect observed after multiple voltammetry cycles, and it was not observed for the number of cycles performed at the same electrode in this study.

Electrochemical behaviors of BPE at GCE/BEA/APTMS/GA/GO/NF with different scan rates (ν) were further investigated. Figure 3A shows the CV curves obtained. Figure 3B shows that the oxidation peak current increased linearly with scan rate in the range from 10 to 1000 mVs−1, indicating that the oxidation of BPE at GCE/BEA/APTMS/GA/GO/NF electrode was an adsorption-controlled process [58]. The regression equation were expressed as Ip = (0.0202 ± 0.0003)ν + (1.57 ± 0.06) (µA, mVs−1, R = 0.997).

Fig. 3
figure 3

A Cyclic voltammograms of 50 µM BPE at GCE/BEA/APTMS/GA/GO/NF with different scan rates. Curves (a–h) are obtained at 10, 20, 50, 100, 200, 300, 500, and 1000 mVs−1, respectively. Inset: Curves a–c. B Dependence of the oxidation peak current (Ip) on the scan rate. C Dependence of the oxidation peak current potential (Ep) on the natural logarithm of scan rate (ν)

Full size image

Figure 3C shows the relationship between the peak potential (Ep) and the natural logarithm of scan rate ν. It can be observed that the anodic peak potential Ep changed linearly versus ln(ν) with a linear regression equation of Ep = (0.024 ± 0.001)ln(ν) + (0.598 ± 0.006) (V, mVs−1, R = 0.988) in the range of 10 to 1000 mVs−1. For a totally irreversible electrode process, the relationship between the potential (Ep) and scan rate (ν) is expressed as reported by Laviron [58]:

$$E_{p} = E_{0} + \left( {RT/\alpha nF} \right)\ln \left( {RTk_{s} /\alpha nF} \right) + (RT/\alpha nF)\ln \left( \nu \right)$$

Where α is transfer coefficient, ks is standard rate constant of the reaction, n is the electron transfer number, ν is the scan rate, E0 is the formal redox potential, R is the gas constant, T is the absolute temperature, and F is the Faraday constant (T = 294 K, R = 8.314 JK−1 mol−1, and F = 96,485 Cmol−1). According to the slope of the plot of Ep versus ln(ν), the value of αn was calculated to be 1.05. Generally, α is assumed to be 0.50 in a totally irreversible electrode process. This allowed the inference that the electron transfer number (n) for oxidation of bisphenol E was around 2.

3.2 pH dependence of BPE at GCE/GO/BEA/APTMS/GA/NF

The effect of pH on the electrochemical responses of BPE at GCE/GO/BEA/APTMS/GA/NF was studied over the pH range of 2.0–9.0 using Britton–Robinson buffer (Fig. 4A). As the process was adsorption-controlled, 90 s magnetic stirring was employed before measurements to guarantee diffusion to the electrode surface and reproducible amounts of BPE at the electrode surface for the voltammetric cycles. The oxidation currents were very similar at pH values 2.0, 3.0, 5.0, and 7.0, slightly lower for pH 6.0 and considerable lower for pH values 4.0, 8.0, and 9.0. The pKa of BPE lies in the basic range (9.81–10.42 [59]), however, our observations do not follow any trend. This behavior was different to that reported for BPA [24], where a maximum was observed at pH 8.0 using a GCE electrode decorated with magnetic nanoparticles and reduced graphene oxide. As no similar studies to the best of our knowledge have been reported for BPE, we measured the behavior of BPA at the GCE/GO/BEA/APTMS/GA/NF electrode for comparison. Interestingly, results diverge from the reported [24], with a maximum at pH 2.0, moderate oxidation current reduction at pH 3.0, and slightly lower current (compared with pH 3.0) for the pH values from 4.0 to 9.0 (Fig. S4 for BPA CV curves). This indicated the influence of the electrode composition on the pH dependence. Clearly, though, the bisphenol structure influenced the interaction with the electrode, as BPE and BPA behaved differently at the GCE/GO/BEA/APTMS/GA/NF electrode. Based on the low variation in peak current intensity at pH 5 and 2 for BPE, we reasoned to use pH 5 for the continuation of the studies, and used phosphate buffer solution as carried out through the modification screening.

Fig. 4
figure 4

A The CV curves of 50 µM BPE in different pH values (Britton–Robinson buffer) at GCE/GO/BEA/APTMS/GA/NF electrode. Background was first recorded in the absence of analyte and subtracted from the respective voltammograms. Scan rate 100 mVs−1, N2 atmosphere. B Dependence of the pH on the oxidation peak current potential (Ep)

Full size image

Further, the peak potential shifted negatively with the increase of pH value. The relationship between Ep and pH is shown in Fig. 4B for BPE, and obeys the equation Ep = (− 0.0590 ± 0.003)pH + (1.00 ± 0.02) (V, R = 0.979). A slope of about − 59.0 mV per pH unit was close to the theoretical value of − 57.6 mV/pH. This indicated that an equal number of electrons and protons were involved in this electrochemical reaction [, , , , 18, 24, 27,28,29], therefore, the electrooxidation of BPE at GCE/GO/BEA/APTMS/GA/NF was a two-electron and two-proton process, as the number of electrons were estimated to be two in the previous section. The proposed electrochemical reaction equation of BPE is shown in Fig. 5.

Fig. 5
figure 5

Proposed electrochemical reaction of BPE

Full size image

3.3 Insights of the electrode surface

Figure S5 shows the SEM images of the bare GCE electrode (Fig. S5A) in comparison with the modified electrode before (GCE/GO/BEA/APTMS/GA) (Fig. S5B) and after NF coating (Fig. S5C). A flat surface was observed for the bare GCE, while agglomerates were evident at the modified electrode. The aggregates are caused by the interaction of the modifier components, BEA/APTMS/GA and GO, and may give rise to the synergistic effect discussed before, where the disposition of the modifiers benefits the electrostatic adsorption of BPE and the electron transfer, including an increase of the electrode surface area. Thorough characterization of BEA and BEA functionalized derivatives were previously reported [32], however, as the materials used in this work consisted of new batches, XRD, SEM, and FT-IR characterization data is presented in the Supplementary Information (Fig. S1), which are consistent with the previously reported data. FT-IR data (Fig. S2) provided evidence for the effective functionalization of GO. Bands at wavelengths between 1000 and 1200 cm−1, were assigned to the C − N, Si − O − Si, and Si − O − C bonds of the aminopropyl groups, as well as the N − H broad band at aproximately 3200 cm−1 [52]. Furthermore, bands in the region 1350–1750 cm−1 were assigned to the bending mode of CH − R − CH3 and methyl group − CH3 of glutaraldehyde, as well as the C = O stretching mode [, 32, 52].

3.4 Differential pulse voltammetric determination

Differential pulse voltammetry (DPV) was chosen to check the analytical performance of the electrode toward BPE. A non-systematic optimization was performed to check the influence of parameters such as accumulation time and potential. Potentials from 0 to 0.5 V where checked at a fixed time (60 s) with magnetic stirring, and no significant peak current increase was observed. However, the current depended on accumulation time. An interval from 0 to 180 s was tested, and an increase was observed for the first 90 s and no increase was observed over longer periods. Based on these tests, an accumulation potential of 0.5 V was applied for 90 s for the DPV measurements. The DPV parameters were also tested for optimization, and the highest current was measured when increment potential, pulse period, pulse amplitude, and pulse width were adjusted to 0.075 mV, 0.4 s, 0.05 V, and 0.075 s, respectively. Under the optimized conditions, DPV of a series of BPE concentrations were collected, as shown in Fig. 6A. A linear relationship between oxidation current (Ip) and BPE concentration (CBPE) was obtained over the range of 0.07 and 4.81 μM (Fig. 6B), and the regression equation was expressed as Ip = (0.52 ± 0.01)CBPE + (0.011 ± 0.001) (µA, μM) with a correlation coefficient of 0.995. The limit of detection (LOD) was estimated to be 0.056 μM (S/N = 3). For means of comparison, BPA was also tested and a linear regression was obtained for a linear range of 0.2–4.00 μM, which obeyed the equation Ip = (1.00 ± 0.04)CBPA + (0.170 ± 0.007) (µA, μM, R = 0.991)) and LOD = 0.19 μM (S/N = 3) (See Fig. S6 for DPV curves and linear regression fitting). The optimized conditions for the two bisphenols were the same indicating that the electrode was not selective toward bisphenols of similar structure; although a lower LOD was obtained for BPE. In the work reported by Lu et al. using a MOF/tyrosinase biosensor (2016) [45], the LOD for BPE was 0.015 μM, in a linear range of 0.05–3.0 μM, which is of similar sensitivity to the electrode in this work. The observed differences regarding BPE and BPA electrode sensitivity and behavior was also in line with previous observations of the substituent group properties (electron acceptor or electron donor) on the bisphenol framework having strong influence on the electrode performance. Although the electrochemical determination of BPE is not vastly reported, the analytical performance of the presented sensor was comparable to other graphene oxide or zeolite material-based sensors reported in the literature for the determination of BPA, what is pertinent considering that both BP’s have shown similar behavior in this work. Linear ranges and LOD are summarized in Table 1. It can be seen that the presented sensor displays a comparable linear range with an acceptable detection limit.

Fig. 6
figure 6

A The DPV curves for different BPE concentration (CBPE): a − 0.07, b − 0.20, c − 0.33, d − 0.46, e − 0.71, f − 1.19, g − 2.09, h − 3.10, I − 4.00, and j − 4.81 µM. B Linear regression for the calibration CBPE versus anodic peak current (Ip). The increment potential, pulse period, pulse amplitude, and pulse width were adjusted at 0.075 mV, 0.4 s, 0.05 V, and 0.075 s, respectively

Full size image
Table 1 Comparison of the analytical performance of this work electrode with other electrochemical sensors graphene oxide/nanozeolite based for the determination of BPE and BPA
Full size table

3.5 Interference, repeatability, and stability

In order to check the robustness of the methodology, an interference assay was performed using 4-chlorophenol (CP), BPA, BPF, TBrBPA, and duroquinone (DQ). For this, the Ip of 2 μM BPE was compared with the Ip of 2 μM BPE in solutions containing the same concentration of the possible interfering chemical. Figure 7A presents the relative Ip percentage recovery of the isolated BPE (100%) within the mixtures. Among the tested compounds, only BPA showed a substantial interference of around 40%. Nevertheless, it can be pointed out that a 100% interference would be expected if BPE and BPA had no specific interactions each with the electrode surface as they oxidize at the same voltage (within error). The analytical equations obtained (DPV calibration of BPE and BPA, Fig. 6B and Figure S6B, respectively) showed different slopes, 0.52 ± 0.01 (BPE) and 1.00 ± 0.04 (BPA), which numerically showed differences in the rate current/concentration that can be attributed to the interaction between analyte and electrode.

Fig. 7
figure 7

A Interference test for BPE determination in the presence of BPA, BPF, TBrBPA, DQ, and CP. DPV measurements under the optimized conditions were employed to detect 2 µM of BPE only, or the same amount of BPE in the presence of 2 µM of the tested interfering compound. Relative anodic current (Ip) was calculated considering the assay with no interfering as 100%. B Repeatability and stability tests for BPE determination using two different electrodes. Measurements performed at the day of preparation, after 3 days, and after 7 days, using 2 µM BPE solutions. Relative anodic current (Ip) was calculated considering Electrode 1 first day as 100%

Full size image

To verify electrode repeatability and stability, two different GCE electrodes were coated on same day under the same conditions, and used to measure 2 μM BPE solutions on the day of preparation, and after 3 and 7 days, Fig. 7B. Repeatability was analyzed considering the percentage difference between consecutive measurements on the same day for one electrode, which were found to be lower than 3%. Good stability was also verified, as the electrodes did not significantly present higher or lower current output in the seven-day time interval tested, with lower and maximum relative Ip within all measurements showing less than 10% difference. It is fair to point out that after measurements the electrodes were simply rinsed with distilled water, dried with a nitrogen flush and stored at room temperature in a closed container.

3.6 Application to analysis

To validate the method proposed, the standard addition method was employed, and recoveries ranging from 91 to 109% were obtained, with residual standard deviation up to 6%, Table 2. The recovery tests indicate that the proposed method is suitable for BPE determination.

Table 2 Tap water analysis and recovery tests of BPE oxidation at GCE/GO/BEA/APTMS/GA/NF electrode
Full size table

4 Conclusion

In summary, it has been demonstrated that the modification of GCE with BEA/APTMS/GA/GO/NF nanocomposite was effective for the highly sensitive determination of BPE. The modifier composition, showed a synergistic effect toward electrode current response compared to bare GCE. This is hypothesized to be related to electrostatic changes on the zeolite surface due to the functionalization using APTMS and GA, favoring analyte adsorption, as the electrochemical process was adsorption-controlled and involves an equal number of electrons and protons. The analytical method proposed presented good detection limits and linear range, which are very close to the limit and range for the only electrochemical sensor for BPE reported (CuMOFs-Tyr-Chi/GCE) [45]. In addition, the electrode has shown good reproducibility, repeatability and storage stability, as well as low interference toward CP, DQ, BPF and TBrBPA. Despite BPE and BPA having very similar peak potential dependence of pH, the current response diverged, which was in agreement with the observation of a higher response for BPE determination compared to BPA using the optimized analytical method proposed. These differences can be hypothesized as response to BPs structural differences, what leads to specific interactions between analytes and electrode surface and may be influenced by the observed surface agglomerates. The results suggest that the functionalized nanozeolite/graphene-based electrode is a potential platform for bisphenols determination.