Introduction

Acetaminophen, also commonly known as paracetamol, is an analgesic and antipyretic drug that is frequently used to reduce fever and provide pain relief [1], with good cardiovascular, renal and gastrointestinal safety when taken at the recommended doses. As it is unable to inhibit the function of the cyclooxygenase (COX) enzyme in the peripheral tissues, it has poor anti-inflammatory properties, but this can be advantageous over the nonsteroidal anti-inflammatory drugs (NSAIDs), such as aspirin [2]. Once acetaminophen is ingested, deacetylation begins and occurs to give the primary amine, 4-aminophenol. This molecule, while implicated in the formation of an endogenous cannabinoid, which activates the cannabinoid CB1 receptors to give pain relief [2], is also linked to the toxic effects of acetaminophen [3]. With the excessive accumulation of acetaminophen and 4-aminophenol, acute liver failure can occur. Therefore, the concentration of acetaminophen in pharmaceutical formulations is highly controlled and regulated. Various analytical methods are employed in the analysis of acetaminophen-containing formulations [4], including chromatography [5] and various spectroscopy techniques [6, 7]. While all these techniques can be employed successfully, they are not always suitable for routine analysis, with high equipment costs, time-consuming analyses, and in some cases poor selectivity and low sensitivity is observed. Acetaminophen is readily oxidised through a two electron two proton transfer reaction at potentials considerably lower than the oxygen evolution reaction at most electrodes to form N-acetyl benzoquinone imine [8]. Accordingly, there is considerable interest in the development of electrochemical sensors that can be employed in the detection and analysis of acetaminophen.

Various materials have been employed to give sensitive and selective electrochemical sensors for the determination of the concentration of acetaminophen, with carbon-based materials, including graphene [9,10,11,12,13], graphite [14, 15], carbon nanotubes [8, 16] and mesoporous carbon [17] featuring in several studies. For example, Kachoosangi et al. [15] utilised plane pyrolytic graphite electrodes modified with carbon nanotubes, Li et al. [18] combined Pd with graphene oxide to give a nanocomposite, while Li et al. [8] employed multi-walled carbon nanotubes for the simultaneous electrochemical detection of acetaminophen, tyrosine and levodopa. Enhanced detection of acetaminophen has also been observed with the covalent modification of carbon-based electrodes, using for example diazonium cations [19] and 4-amino benzoic acid [20]. In addition, various metal nanoparticles and metal oxides/hydroxides have been utilised, including palladium [18, 21, 22]; nickel and copper [23, 24]; titanium oxide [11, 13]; iron oxide [25]; zinc oxide [26]; tungsten oxide [27]; cobalt [28]; gold [29, 30]; perovskite-type oxides [31]; and doped nanoparticles, such as nickel-doped cobalt ferrite nanoparticles [32]. While these sensors have been effectively used in the electrochemical detection of acetaminophen, there is always the risk of carbon nanotubes, graphene flakes and metal nanoparticles leaching from the sensor surface. This not only results in a loss in the performance of the sensor, but can result in the release of nanomaterials into the environment.

In recent times, cyclodextrins, which are well known as supramolecular systems [33], have been employed in the fabrication of sensors. These supramolecular systems can be incorporated as anionic dopants within conducting polymers [34], adsorbed onto surfaces [35] or more commonly combined with other materials to generate a slurry or suspension that can be used to modify the electrode surface [36, 37]. Furthermore, there has been considerable interest in the electropolymerisation of β-CD [38,39,40]. However, this electropolymerisation approach involves cycling the GCE to relatively high potentials in the presence of acidic solutions, where the GCE becomes oxidised and activated [41]. This oxidation leads to the formation of oxygen-containing functional surface groups, and these have been reported at glassy carbon [42], carbon nanotubes [43] and activated carbon [44]. Their presence can lead to a considerable increase in the rate of the electron transfer step [42, 45].

In this paper, results are presented on the attempted electropolymerisation of β-cyclodextrin (β-CD) in a phosphate buffer solution at GCE and, for comparative purposes, similar experiments were carried out in the absence of the cyclodextrin. In both cases, conducting surfaces were generated, providing very simple electrochemical sensors for acetaminophen. Using acetaminophen as a well-studied analyte, the roles of both the β-cyclodextrin and the activation of the GCE substrate were explored.

Experimental

All chemicals, including acetaminophen, β-CD, sulfonated β-CD, caffeine, ascorbic acid, aspirin, uric acid, sorbate and phosphate salts were obtained as Analar grade reagents from Sigma-Aldrich and used without any further purification. A glassy carbon rod electrode (GCE) (3 mm in diameter) was used in all experiments. This rod was embedded in epoxy resin and then inserted into a Teflon holder with electrical contact being achieved by means of a copper plate at the base of the carbon sample. Prior to each experiment, the GCE surface was polished to a mirror finish, using successively smaller sizes of diamond paste down to a 1-μm particle-sized paste. The electrode was then rinsed with distilled water and finally cleaned in an ultrasonic bath. A platinum wire was used as an auxiliary electrode and a saturated calomel electrode (SCE) was used as the reference electrode in the voltammetry experiments, while a silver-silver chloride reference electrode was employed in the differential pulse voltammetry (DPV) experiments.

A Hitachi S-3200-N scanning electron microscope with a tungsten filament was used to study the morphology of the β-CD-modified GCE and activated GCE, while energy dispersive X-ray (EDX) analyses were carried out using an Oxford Instrument INCAx-act EDX system. ATR-FTIR data were collected using a Varian FTS-7000 FTIR spectrometer equipped with a Pike Miracle ATR diamond crystal. All samples were thoroughly rinsed and dried before analysis.

The cyclic voltammetry data were recorded using a Solartron 1287 potentiostat, while the DPV experiments were performed using a CHI potentiostat. All measurements were recorded at room temperature. Unless otherwise stated, the β-CD-modified GCE was formed by cycling the GCE from –2.0 to 2.2 V vs. SCE in a solution of 5 mM β-CD with a phosphate buffer (pH of 5.5) at 100 mV s–1 for 20 cycles. The concentration of β-CD (5 mM) is below the solubility limit at room temperature. Therefore, during electropolymerisation, the solutions were not agitated to disperse insoluble β-CD particles, a process that has been employed in previous studies [38]. A similar approach was used for the activation of the GCE, with the electrode being polarised in a slightly acidic solution between the same potential limits of –2.0 to 2.2 V vs. SCE for 20 cycles. The modified electrodes were then thoroughly rinsed with distilled water and transferred to the buffer and cycled for 2 cycles before cycling in the acetaminophen-containing solution. Unless otherwise stated, a phosphate buffer solution, at a pH of 7.0, was used in the electrochemical detection of acetaminophen. The DPV experiments were recorded at the optimised conditions which were found to be a pulse amplitude of 50 mV, a pulse width of 0.055 s, a sampling width of 0.0167 s, a pulse period of 0.5 s and an increment of 4 mV. The experiments were repeated at least 3 times and the standard error (SE= ρ/n1/2) is shown as error bars on the relevant plots.

Results and discussion

Formation and characterisation of the β-CD-modified GCE and the activated GCE

The attempted electropolymerisation of β-CD was performed by cycling the electrode from –2.0 to 2.2 V vs. SCE in a slightly acidified phosphate buffer (pH = 5.5) solution with a 5 mM β-CD solution. This resulted in the formation of a dark coloured film, in agreement with previous studies [38]. For the phosphate buffer system containing β-CD (Fig. 1(a)), two anodic peaks are observed at about 0.6 V and 1.6 V vs. SCE and these increase in intensity with increasing cycle number. A clear reduction wave is evident at approximately − 0.6 V vs. SCE and again this peak increases with increasing cycle number. The electropolymerisation of β-CD has been described in terms of the initial adsorption of the β-CD at the carbon electrode followed by the formation of radical cations on the application of relatively high potentials, which gives rise to the coupling of the adsorbed molecules and the formation of the deposited polymeric structure [38]. Indeed, the cathodic peak at –0.6 V vs. SCE has been directly related to the extent of polymerisation [38].

Fig. 1
figure 1

Cyclic voltammograms recorded at 100 mV s-1 in a 5 mM β-CD and 0.1 M phosphate, cycles 1 to 20, b 20th cycle in 5 mM β-CD with 0.1 M phosphate and in 0.1 M phosphate and c in the presence and absence of 5 mM β-CD in 0.1 M NaCl and 0.05 M Na2SO4

However, as shown in Fig. 1(b), nearly identical voltammograms were obtained on cycling the GCE in the phosphate solution without the presence of the β-CD. This cycling activates the GCE, with alterations to its microstructure, the formation of new active edge planes, the formation of a graphite oxide layer and the formation of oxygen-containing functional groups. Instead, it appears that the reduction peak at about –0.6 V is connected with the reduction of oxides formed during the cycling of the GCE to these relatively high potentials. The peak currents associated with this reduction event increase with cycling while the peak potentials become slightly more negative with continued cycling. It has also been shown that the supporting electrolyte plays a role in the activation process [41] and this can be seen in Fig. 1(c), where the GCE was cycled in the presence of β-CD with chloride and sulfate as the supporting electrolytes. Again, identical voltammograms were recorded in the absence of the β-CD, suggesting that the redox peaks observed are connected with the oxidised and activated GCE. The voltammogram recorded in the chloride-containing solution is very different to that observed in the phosphate or sulfate solutions. The characteristic oxidation waves seen prior to the oxygen evolution reaction with the phosphate and sulfate systems are no longer evident, while more pronounced reduction waves appear during the reverse cycle. The onset of the oxygen evolution reaction occurs at a lower overpotential in the chloride-containing solution. This is a complex reaction, involving the formation of adsorbed species, OH radicals and the evolution of gaseous molecules. Moreover, the C1- ions can combine with the OH species to give various reactive intermediates, such as ClOH•- [46]. Therefore, the earlier onset of this reaction in the presence of the chloride solution is likely to alter both the activation of the GCE electrode and the adsorption and possible electropolymerisation of the β-CD molecules. In addition, the adsorption of the chloride, phosphate and sulfate anions at the GCE may also play a role in this activation process.

In an attempt to further understand the role of the β-CD and the activated GCE surface, ATR-FTIR and SEM/EDX were employed. Representative FTIR data are shown in Fig. 2, where the spectrum of β-CD is compared with the β-CD-modified and activated GCE. In terms of the β-CD, the symmetric and antisymmetric stretching of O − H is evident as a broad peak centred at 3230 cm−1, while bands for C − H stretching at 2930 cm−1, C − O stretching at 1038 cm−1, C − O − C stretching at 1153 cm−1, and bending vibrations of O − H at 1028 cm−1, are also seen. The spectra recorded for the activated GCE and β-CD-modified GCE are very similar. In this case, the broad bands in the region from 3000 to 3400 cm−1 can be attributed to O − H stretching vibrations, and these may indicate the presence of edge and basal hydroxyls. The bands in the vicinity of 1400 to 1800 cm−1 can be assigned to oxidised groups, such as carbonyl, ketones and carboxyl groups [47]. In particular, the band at 1056 cm−1 can be ascribed to C − OH stretching vibrations, the band at 1650 cm−1 can be related to C = C stretching and the vibration at 1214 cm−1 is characteristic of C − O − C. The band at 1200 cm−1 can be attributed to phenolic species, while bands at lower wavenumbers are consistent with C = C bending modes. While there are some similarities between the FTIR data recorded with the β-CD and the β-CD-modified electrode, it is difficult to prove the presence of the electropolymerised β-CD. In both cases, similar functional groups, for example, C − O and O − H contribute to the observed vibrations, while the spectra recorded for the activated and β-CD-modified electrode are nearly identical.

Fig. 2
figure 2

FTIR data recorded for β-CD (green), β-CD-modified (grey) and activated GCE (orange)

Using EDX, an elemental analysis was performed on the activated GCE and β-CD-modified GCE and these data are summarised in Table 1. Pristine GCE was used as a control, giving an elemental analysis of 100% carbon; however, all the other modified GCE surfaces exhibited significant oxygen content. An oxygen content approaching 30% was observed for the activated and β-CD-modified GCE, while a somewhat lower value of 12% was achieved using a sulfonated β-CD. This sulfonated β-CD was employed as it easily detected using EDX. Interestingly, no sulfur was detected when this β-CD (with 12–15 OH groups replaced with SO3 groups) was added to the phosphate solution, clearly showing that it is not electropolymerised at the GCE. The electropolymerisation of this sulfonated β-CD may be slower that the unsubstituted β-CD, but it should nevertheless polymerise, given its good solubility with some free OH groups. The lack of a sulfur signal suggests that oxidation of the GCE surface is the main reaction that occurs during the cycling of the GCE in the phosphate solution. Possible reaction pathways have been previously proposed in the literature leading to the formation of ketones, carbonyl, aldehydes and carboxylic groups. For example, Yi et al. [47] have suggested that the electrophilic addition of water occurs at the carbon conjugated Π-system in acidic solutions, and this is then followed by oxidation and condensation steps. The slightly acidic phosphate buffer at a pH of 5.5 used in Fig. 1, appears to be sufficiently acidic to facilitate these events.

Table 1 Elemental analysis obtained for GCE cycled in 0.1 M phosphate buffer for 20 cycles at 50 mV s–1 from –2.0 to 2.2 V vs. SCE and in the presence of 5 mM β-CD and 5 mM sulfonated β-CD. Data averaged over 5 sites on the surface

Typical SEM micrographs recorded after cycling the GCE at 100 mV s–1 in the phosphate buffer in the presence and absence of the β-CD between the potential limits of –2.0 and 2.2 V vs SCE are shown in Fig. 3(a) and (b), respectively. On comparing these to the pristine GCE, Fig. 3(e), it is evident that the surface is rougher with cracked-like features more evident on cycling in the phosphate solution without the presence of the β-CD. There is no clear evidence to suggest that the β-CD has electropolymerised at the GCE; however, the globules seen in Fig. 3(a) may be connected with β-CD deposits. On decreasing the scan rate between the potential intervals of –2.0 and 2.2 V to 50 mV s–1, the surface morphology becomes very different, as illustrated in Fig. 3(c). In this case, the GCE adopts an etched-like surface, consistent with dissolution of the GCE. Although the surface morphologies in the phosphate solution are similar to that obtained in the β-CD-containing phosphate electrolyte, more extensive etching is seen in the absence of the β-CD. This may indicate that the β-CD protects the surface to some extent when cycled to these high potentials. These data are consistent with a study by Yi et al. [47] who concluded that degradation of the GCE occurred by ring opening in the graphitic structure with the formation of oxides.

Fig. 3
figure 3

SEM micrographs recorded for GCE formed in 0.1 M phosphate between –2.0 and 2.2 V vs. SCE in the a presence of 5 mM β-CD at 100 mV s–1 and b absence of β-CD at 100 mV s–1, c presence of 5 mM β-CD at 50 mV s–1, d absence of 5 mM β-CD at 50 mV s–1 and e pristine GCE before cycling

Redox behaviour of acetaminophen at β-CD-modified and activated GCE

The performance of the modified GCE surfaces in the oxidation of acetaminophen was investigated using the electrodes modified in the chloride, sulfate and phosphate solutions. The best detection was observed using the phosphate system, while as expected the chloride-based system gave the lowest peak currents (data not shown). In Fig. 4(a), the voltammograms recorded for the modified electrodes, generated by cycling in the phosphate solution in the presence and absence of the β-CD, and the unmodified GCE are compared. The typical irreversible diffusion-controlled behaviour of acetaminophen is evident for the unmodified GCE, with a broad oxidation wave centred at about 0.55 V vs. SCE. Furthermore, the corresponding reduction peak is absent, indicating a slow irreversible redox reaction. This is consistent with the formation of N-acetyl benzoquinone imine, which then undergoes a hydrolysis reaction, as illustrated in Scheme 1.

Fig. 4
figure 4

Cyclic voltammograms recorded at 100 mV s–1 scanning from –0.2 to 1.0 V in 1.0 mM acetaminophen in a neutral phosphate buffer at a GCE, activated GCE and at β-CD/GCE and b extended activation period, with β-CD/GCE formed by cycling between –2.0 and 2.2 V vs. SCE at 50 mV s–1 for 20 cycles

Scheme 1
scheme 1

Oxidation of acetaminophen

In contrast, the peak oxidation waves appear at considerably lower overpotentials, centred at approximately 0.35 vs. SCE, for the modified GCE surfaces. Furthermore, the anodic peak currents are significantly higher, reaching values of approximately 1.95 ± 0.08 mA cm–2 and 1.78 ± 0.09 mA cm−2 for the β-CD/GCE and activated GCE, respectively, compared to a much lower current density of about 0.2 mA cm–2 for the unmodified GCE. These narrow symmetric redox waves, evident at the β-CD/GCE and the activated GCE, are consistent with an adsorption-controlled reaction, where the acetaminophen is confined at the surface. However, on extended activation and oxidation of the GCE, achieved by cycling the β-CD at a slower scan rate of 50 mV s–1 between the potential limits of –2.0 to 2.2 V vs. SCE for 20 cycles, lower peak currents, higher peak potentials and wider peaks were seen, as illustrated in Fig. 4(b). This indicates a slower rate of electron transfer. Moreover, higher background currents are evident, with a broad wave emerging in the vicinity of 0.0 to 0.2 V vs. SCE. This broad wave was also evident on cycling in the phosphate buffer, suggesting that it is associated with surface bound redox groups generated during the extended activation and oxidation of the GCE surface. Indeed, the increased capacitance is consistent with the formation of the highly porous surface layers, seen in Fig. 3(c).

The data presented in Fig. 4(a) are consistent with a surface-confined analyte. The experimental ∆Ep values, were in the vicinity of 45 mV at slow scan rates, and these peak separations are close to the theoretical value of 0 mV for the reversible electron transfer of an ideally behaved adsorbed species. Moreover, the ratio of jp(ox)/jp(red) is relatively close to unity at the lower scan rates. This indicates that the electrogenerated N-acetyl benzoquinone imine is reasonably stable and not further hydrolysed to any great extent. Therefore, it can be reduced back to acetaminophen. While this conversion of N-acetyl benzoquinone imine to acetaminophen can be observed at high scan rates at a variety of electrodes, it is seen in this case at slow scan rates in the vicinity of 15 mV s–1, suggesting that it is indeed protected from extensive hydrolysis. This protection may be related to its adsorption at both the activated GCE and β-CD/GCE surfaces. It appears that the electrogenerated edge planes and oxygenated functional groups formed during the activation of GCE facilitate the adsorption of both acetaminophen and N-acetyl benzoquinone imine, while any deposited or electropolymerised β-CD can also facilitate the formation of an inclusion complex, trapping the acetaminophen and the oxidised product, N-acetyl benzoquinone imine, at the surface.

The relationship between the oxidation peak current and the scan rate for both the β-CD-modified and activated GCE is shown in Fig. 5. In both cases, linear relationships were obtained. The linear regression equation obtained for the β-CD-modified GCE was Ep/mV = (0.015 ± 0.001) υ + (0.225 ± 0.091) (R2 = 0.98), compared to Ep/mV = (0.016 ± 0.001) υ + (0.151 ± 0.052) (R2 = 0.99) for the activated GCE, both consistent with the oxidation of acetaminophen being under adsorption control. On plotting the logarithm of the peak current as a function of the logarithm of the scan rate the slopes of the linear plots were obtained as 0.85 and 0.78 for the activated and β-CD-modified GCE electrodes, deviating somewhat from the ideal value of 1.0 for adsorption-controlled redox reactions. In particular, the value of 0.78 obtained for the β-CD-modified electrode suggests mixed control (adsorption and diffusional processes).

The influence of pH on the oxidation of acetaminophen was studied using DPV and representative plots recorded for the β-CD-modified electrode at a number of different pH values are presented in Fig. 6(a). The peak currents become lower in the more alkaline solutions, with a peak current of 1.8 mA cm−2 at a pH of 9.2, compared to 3.8 mA cm-2 at a pH of 6.5. This may be related to the ionisation of the acetaminophen molecule, which has a pKa of 9.5, in the more alkaline solution (pH of 9.2). This anionic acetaminophen should have a higher solubility in water and this may reduce its extent of adsorption at the electrode-solution interface. At pH values lower than 9.2, the acetaminophen will become increasingly more protonated as the solution becomes more acidic. This is consistent with the somewhat higher peak currents observed at a pH of 6.5, where the molecule will be neutral (Fig. 6(a)). The variation of the peak oxidation potential with pH is illustrated in Fig. 6(b). There is a clear linear variation between the pH and peak potential and the slope of the linear plot is 0.055 V which is in very good agreement with the Nernst equation. The oxidation of acetaminophen is a 2e/2H+ redox reaction, as illustrated in Scheme 1, and therefore the Nernst equation predicts that the electrode potential is related to pH with a slope of 0.059 V at 298 K.

Fig. 5
figure 5

a Plot of peak current as a function of scan rate and b logarithm of peak current as a function of the logarithm of scan rate, recorded in 1.0 mM acetaminophen, for the β-CD GCE (grey) and activated GCE (orange)

Fig. 6
figure 6

a Differential pulse voltammograms recorded in 1.0 mM acetaminophen at pH values of 9.0 (black), 7.5 (orange), 6.5 (purple), 5.1 (green), 3.5 (blue) and 2.1 (grey) and b corresponding plot of peak potential as a function of pH

The standard heterogeneous rate constant (ks) was estimated using Laviron’s model, where it is assumed that acetaminophen is irreversibly adsorbed [48]. This relationship is given in Eq. 1, where Ep corresponds to the peak potential, is the formal potential, α represents the charge-transfer coefficient, n corresponds to the number of electrons transferred, T is the thermodynamic temperature and F is Faraday’s constant.

$$E_p=E^0+\left(\frac{RT}{\left(1-\alpha\right)nF}\right)ln\left(\frac{RTk_s}{\left(1-\alpha\right)nF}\right)-\left(\frac{RT}{\left(1-\alpha\right)nF}\right)lnv$$
(1)

A typical plot for both the β-CD-modified and activated GCE is shown in Fig. 7, where good linearity is obtained, indicating good agreement with Eq. 1. The linear regression equation for the β-CD-modified GCE was obtained as Ep/mV = (0.0144 ± 0.0011) lnv + (0.325 ± 0.005), while the corresponding relationship for the activated GCE was Ep/mV = (0.0149 ± 0.0004) lnv + (0.282 ± 0.0020). An α value of 0.5, and E0 values obtained by plotting Ep as a function of v and extrapolating to v = 0, to give E0 values of 370 mV and 325 mV for the β-CD-modified GCE and activated GCE, respectively, were used in the calculations. Estimated ks values of (4.3 ± 0.3) × 10–3 cm2 s–1 and (3.4 ± 0.2) × 10–3 cm2 s–1 were obtained for the β-CD-modified and activated GCE, respectively. These values correspond to efficient electron transfer kinetics for both the β-CD and activated GCE electrodes, with a slightly lower rate constant for the activated GCE. While the presence of the β-CD does not influence the electrochemical data (Fig. 1), it does have some effect on the surface morphology (Fig. 3), protecting the GCE surface from extensive etching. Indeed, as shown in Fig. 4, extensive etching leads to wider peaks, consistent with slower kinetics (Fig. 4(b)). Therefore, this somewhat higher rate constant for the β-CD system may be related to the extent of oxidation and the nature of the oxygenated species, or indeed some adsorbed β-CD. Using the slope values, and taking n = 2, (1 − α) was computed at 0.84, giving an α value within the range of 0.0 to 1.0, but outside the normally expected value between 0.4 to 0.7.

Fig. 7
figure 7

Plot of Ep (oxidation wave) as a function of the logarithm of the scan rate for β-CD modified GCE (grey symbols) and activated GCE (orange symbols), recorded in 0.5 mM acetaminophen

Sensing performance of the modified GCE electrodes

The performance of the β-CD-modified GCE and the activated GCE electrode as a sensor for the detection of acetaminophen was studied using DPV measurements. The resulting calibration curves are presented in Fig. 8. The relationship between the peak oxidation currents recorded using DPV for both the β-CD-modified GCE and activated GCE are shown in Fig. 8(a), where it is evident that a linear relationship, extending from 0.1 to 80 μM, is observed at the lower concentrations, while the currents reach a limiting or saturation point with the higher concentrations, to give a curve over the entire concentration range. The linear regression of the calibration curves, shown in Fig. 8(b), were obtained as Ip/mA cm−2 = (0.0327 ± 0.0006) c/μM + (0.0462 ± 0.0240) for the activated GCE and as Ip/mA cm−2 = (0.0344 ± 0.0003) c/μM + (0.0004 ± 0.0040) for the β-CD-modified GCE, with correlation coefficients > 0.99. Using the slopes of these equations, the sensitivity of the measurement, was calculated as 32.7 ± 0.6 μA cm–2 μM–1 for the activated GCE and as 34.4 ± 0.3 μA cm–2 μM–1 for the β-CD-modified GCE. The corresponding LOD values were estimated as (9.8 ± 0.4) × 10–8 M and (9.7 ± 0.4) × 10–8 M, respectively, using the standard deviation of the blank and the sensitivity. Clearly, there is no significant difference between the β-CD-modified and activated GCE in the electrochemical sensing of acetaminophen.

Fig. 8
figure 8

Calibration curves recorded over a a wide concentration range and the b linear range, using DPV data recorded for the β-CD-modified GCE (grey) and activated GCE (orange)

These sensing parameters compare reasonably well, with a number of previous investigations, where multiwalled carbon nanotubes (MWCNTs) [49], graphene [9] and various nanoparticles (NPs) [50], and other electrode modifications, such as bismuth [51], were used. While the linear region is somewhat lower than some of the reported values (Table S1 (supplementary data)) that have linear ranges extending from 5 to 200 μM [49], the sensitivities are high at 32.7 and 34.4 μA cm–2 μM–1. These compare very favourably with reported values of 37.28 μA mM−1 [30], 0.618 μA cm–2 μM–1 [23] and 0.15 μA μM–1 [26].

The selectivity of the detection of acetaminophen at the β-CD/GCE was studied using caffeine, ascorbic acid, aspirin, uric acid and sorbate as interference compounds. A representative differential pulse voltammogram is shown in Fig. 9, which was recorded in a neutral solution with 0.025 mM acetaminophen and 0.05 mM caffeine, 0.05 mM aspirin and 0.05 mM ascorbic. Here, it is clearly seen that at these concentrations the oxidation of acetaminophen is not influenced by the added interferences, with the oxidation of caffeine observed at 1.28 V vs. Ag/AgCl and the oxidation of ascorbic acid appearing as a broad wave centred at 0.04 V vs. Ag/AgCl, with no overlap with the oxidation wave of acetaminophen. Indeed, the peak currents obtained for acetaminophen in the absence and presence of the interferences were very similar, indicating good selectivity. This is illustrated in Fig. 9(b), where the peak current obtained for acetaminophen is shown in the absence and presence of 0.05 mM caffeine, 0.05 mM aspirin, 0.05 mM ascorbic, 0.05 M uric acid and 0.05 mM sorbate. Furthermore, it is evident from Fig. 9(a) that the sensor has applications in the simultaneous sensing and detection of acetaminophen and caffeine. This good selectivity is partially related to the fact that many of the interferents oxidise at potentials that are different to the Ep value of acetaminophen, but it may also be connected with adsorption of the acetaminophen, partially blocking the interferents from the surface. However, on increasing the concentration of the interferents, interference was observed with uric acid and when a tenfold excess of UA was employed, the acetaminophen peak current was reduced to 0.65 mA cm−2, with the emergence of the UA oxidation wave, corresponding to a 25% reduction in the peak current.

Fig. 9
figure 9

a DPV recorded in 0.025 mM acetaminophen with 0.05 mM ascorbic acid, 0.05 mM caffeine and 0.05 mM aspirin in a phosphate buffer, at a pH of 7.0, b acetaminophen (Acet.) peak currents recorded for 0.025 mM acetaminophen in the absence and presence of various interference compounds (0.05 mM).

The ability of the sensor to detect acetaminophen in a commercial pharmaceutical tablet containing caffeine was tested using a standard spiking and recovery experiment. The tablet was dissolved in the phosphate buffer and made up to different concentrations. No other treatments were carried out, with the samples containing various excipients. The sensor was then employed in these real samples and the currents recorded were compared to the predicted values obtained using the acetaminophen standards. The % recovery was very good with the lower added concentrations (97 to 100%, with additions varying between 0.24 and 1.90 μM), but with the much higher concentrated spikes the recovery values become lower (92.5% with a spike of 19.9 μM) (Table S2). Nevertheless, these data do illustrate the satisfactory selectively of the sensor in a complex solution with caffeine and other excipients.

Comparison of the β-CD and activated glassy carbon

It is clear that activation of the GCE and its modification in the presence of β-CD alters the oxidation of acetaminophen from the normally diffusion-controlled reaction to an adsorption-controlled event. Furthermore, the electrogenerated N-acetyl benzoquinone imine is protected from hydrolysis through adsorption at the modified electrodes. Similar data were obtained using both the activated GCE and the β-CD-modified GCE, clearly showing that while the β-CD may play a minor role, it is the activated GCE that contributes to the enhanced detection of acetaminophen. It is well known that the electrooxidation of carbon-based electrodes polarised beyond the oxygen evolution reaction, results in the formation of oxygen-containing groups [52]. At these high anodic potentials, OH radicals are produced which then attack the surface carbon atoms at defect sites leading to the breakage of C–C bonds, with the formation of various oxygenated groups. Microcrystalline graphite with oxide-enriched edge planes are formed, enhancing the electrochemical properties of the activated electrode [52]. It appears that these events are responsible for the enhanced electrochemical oxidation and sensing of acetaminophen, providing a very simple, low cost and environmentally acceptable electrode modification approach. Indeed, this activation is likely to play a significant role in the sensing events of several studies where the GCE was polarised to high potentials during its modification [53,54,55,56].

Conclusion

In attempting to electropolymerise β-CD at a GCE surface, it was found that the electrochemical window normally employed in the electropolymerisation of β-CD results in the activation and oxidation of the GCE surface. This activation results in the enhanced detection of acetaminophen with evidence that acetaminophen is adsorbed at the activated surface, resulting in the protection of the acetaminophen oxidation product from hydrolysis. The presence of the β-CD in the phosphate solution during the activation of the GCE has little influence on the enhancement of the detection of acetaminophen. Indeed, similar rate constants, LOD values and sensitivities were observed with the simple activation of the GCE in the absence of the β-CD. These results clearly show that the GCE is altered and activated during electrode modifications that require the application of high potentials in the vicinity of the oxygen evolution reaction. This activation is likely to contribute to the increased detection reported in a number of studies.