Characterisation of the prepared nano-GO-SiO2 nanoparticles
The nano-GO-SiO2 nanoparticle–modified electrode was characterised using SEM (scanning electron microscopy). The SEM uses a focused beam with high-energy electrons that generate different signals on the surface of electrode. These signals reveal information about the external morphology, the chemical composition and the visuals of the crystalline structure of the SiO2 nanoparticles. This shows the spherical morphology and homogenous size distribution (Fig. 1a) of the synthesised SiO2 nanoparticles. The spherical particles had sizes ranging from 100 to about 200 nm, as can be seen in Fig. 1 a, whilst Fig. 1 b shows the morphology of the nano-GO-SiO2 nanoparticle mixture on the surface of the GCE.
As reported earlier by Tashkhourian and Nami-Ana  and Wang et al. , the nanosized SiO2 nanoparticles as seen on the SEM images creates a large active surface area, better transmission performance and high accumulation efficiency.
Fourier transform infrared spectroscopy (FTIR) was also used for the characterisation of the SiO2 nanoparticles, nano-GO and nano-GO-SiO2 nanoparticle combination, to show the decoration of the SiO2 nanoparticles on the nano-GO. The spectra were recorded as can be seen in Fig. 2, where the nano-GO spectrum and SiO2 nanoparticles agreed with other works. The bands around 3250 and 1408 cm−1 were attributed to the deformed –OH bonds of graphene oxide and the CO–H groups, respectively. Meanwhile, the bands centred around 1039 cm−1 were associated with the C–O bond stretching and the 1724 cm−1 was then associated to the stretching vibration of the carbonyl or carboxyl groups [23, 24]. With the mixing of the SiO2 nanoparticles, the characteristic peaks of silica nanospheres were formed on the surface of the nano-GO, where the peaks at 533 and 795 cm−1 are attributed to Si–O–Si bending and symmetric stretching vibration, respectively. The main strong and broad band at 1066 cm−1 with a bit of shoulder is usually assigned to the transverse (TO) and longitudinal optical (LO) modes of the Si–O–Si asymmetric stretching vibrations [25, 26]. On the other hand, the band observed at 953.59 cm−1 on the SiO2 IR image attributed to the silanol group (Si–OH) is seen to have diminished at the combination of nano-GO and SiO2 nanoparticles. This diminished band can be attributed to the hydrogen bonding between the silanol groups (Si–OH) to the GO nanocolloids and the GA. Furthermore, there was no significant change in the area between 1575 and 1727 cm−1 of the nano-GO, which would have demonstrated a reaction of the silica with the carbonyl groups forming Si–O–C bonds; hence, a confirmation of the hydrogen bonding proposed mechanism.
The effect and influence of scan rate were studied towards the electrochemical determination of GA, using CV to record the voltammograms of 0.5 × 10−3 mol L−1 of GA in 0.1 mol L−1 phosphate buffer of pH 2.0. At increasing scan rates of 10, 25, 50, 75, 100, 150 and 200 mVs−1, the Ip of GA oxidation increased as the scan rate increased as seen in Fig. S1. This shows that the oxidation peak current (Ip) was linear with the scan rate (ν), indicating that the process at the electrode was an adsorption-controlled step. The linear regression equation was Ip (μA) = 0.25 ± 0.10, ν (mVs−1) + 14 ± 1 and R2 = 0.9926.
It can also be observed that, with the increase in the scan rate, there was a positive shift of the peak potentials, which could suggest a kinetic limitation in the reaction [20, 27].
Electrochemical behaviour of gallic acid at the nano-GO-SiO2 nanoparticle–modified glassy carbon electrode
The voltammetric behaviour of gallic acid was studied using cyclic voltammetry (CV). Modification of the GCE electrode by the different reagents was carried out by drop casting with an optimised volume of 5 μL of nano-GO, SiO2 nanoparticles and nano-GO-SiO2 nanoparticles. The nano-GO-SiO2 nanoparticles were dropped on the GCE (see Graphical abstract) and left to dry at room temperature.
The cyclic voltammograms (Fig. 3) show the determination of 1.0 × 10−2 mol L−1 gallic acid using a bare GCE, SiO2 nanoparticle–modified GCE and nano-GO-SiO2 nanoparticle–modified GCE (nano-GO-SiO2-GCE) in phosphate buffer of pH 2 at a scan rate of 100 mVs−1 in a potential range of 0.0 to 1.8 V, at room temperature.
The modified electrodes were then used to test GA (1 × 10−2 mol L−1) and produced peak currents of 261 ± 1 μA at 0.63 V for the SiO2 nanoparticle GCE, 300 ± 3 μA at 0.63 V for the nano-GO-GCE and 411.7 ± 0.9 μA at 0.64 V for nano-GO-SiO2 nanoparticles. In contrast, the bare (uncoated) GC electrode produced a peak current (Ip) of 241 ± 4 μA at 0.63 V, which was lower than Ip for the modified electrodes, as can be seen in Fig. 5. Furthermore, no oxidation peak was observed within the potential range of 0.0–1.8 V and specifically, no peaks over the range where GA would normally occur with just the phosphate buffer (0.1 mol L−1) blank.
From the voltammograms shown in Fig. 5, it was deduced that, though the GCE was modified with SiO2 nanoparticles and nano-GO, the combination of both nano-compounds produced better peak currents compared with the individual modifiers.
Gallic acid in this determination showed two oxidation peaks during the anodic sweep between the applied potential ranges of 0.0 to 1.8 V. The two oxidation peaks are found at peak potentials of 0.64 V for the first and 0.91 V for the second peak. These results are similar to other reports in the literature [12, 14, 20].
These two peaks have been characterised in the literature to be the formation of a semiquinone radical (first peak) which is oxidised to a quinone in the second peak . The oxidation that occurs at the first peak is because of the oxidation of the galloyl group, as can be seen in Fig. 4, which then leads to the second peak that is assumed to have been developed from the third –OH group in the galloyl moiety of the compound [14, 28, 29]. This assumption is reinforced by the fact that the electro-oxidation of the carboxylic group (–COOH) occurs at 2.0 V and produces CO2 .
Suggested mechanism of action
The chemical effect of nano-GO-SiO2 nanoparticles towards the oxidation of GA and the enhancement of the voltammograms can be attributed to the structural, chemical and physical properties of GO and SiO2. The SiO2 nanoparticles with silanol groups bind with the GO nanocolloids via hydrogen bonding as depicted in Fig. 5. The presence of hydrogen bonds on the surface of SiO2 under aqueous conditions, contributes to the electrical transport properties of graphene oxide on SiO2 surface structures. Whilst adsorbed on the surface of the graphene oxide nanocolloids, SiO2 nanoparticles also adsorbs with GA.
The mechanism proposed for this work is based on the work of S. Milonjić [30, 31] and colleagues, where they investigated the interaction of SiO2 and GA. They found out that SiO2 and GA underwent an adsorption interaction which was linked to the pKa (4.41) of GA, where the amount of adsorbed GA increases with increasing equilibrium concentration of GA in the solution or decreasing pH of the solution. Meanwhile, there was a decrease in the amount adsorbed when there was an increase in pH of the solution.
The chemical behaviour of SiO2 and GA was then explained using the pKa of GA molecules and the surface charge of the silica powder. It should be noted that the adsorption of weak organic electrolytes like GA on the surface of amphoteric solids like silica depends amongst other things on the degree of dissociation of the electrolyte and the surface charge of the solid, which all depends on the pH of the aqueous solution . The point of zero charge values (pHpzc), for various forms of SiO2, lie between the pH range of 2.5–3.6 [30, 31]; hence, when the pH is below the pHpzc, the surface of SiO2 becomes positively charged, whilst in a pH higher than the pHpzc, the surface becomes negatively charged. With a pKa of a weak acid, GA at pH values less than 3.5 is mostly un-dissociated or in its stable form and dissociates when the pH increases.
It can therefore be interpreted that the maximum adsorption of GA to SiO2 occurs at a pHpzc where the silica surface is uncharged, and the GA molecule is un-dissociated. Hence, in a phosphate buffer solution of pH 2.0 < 3.5, where the electrochemical determination takes place, GA is adsorbed on the SiO2. Meanwhile, at pH > 3.5, GA’s dissociation increases, leading to an increase in the negative ion C6H2 (OH)3 COO− concentration and a negatively charged SiO2 surface. With the increase in the negative molecules and negative surface, the adsorption between GA and SiO2 nanoparticles decreases due to the repulsive negative-negative forces in the solution.
On the other hand, the GO nanocolloids used possess excellent electrochemical properties due to their favourable electron mobility and unique surface properties like high specific surface area that would accommodate active species like GA and facilitate electron transfer . The capacity of GO to be chemically or physically functionalised, showing non-covalent ϖ-ϖ (ϖ-stacking), cation-ϖ, van der Waals or hydrogen bonding , formed the bases of this proposed mechanism. The SiO2 nanoparticles physically functionalise the GO nanocolloids, thus enhancing the availability of the nanosized surface area for electron transfer, whilst the adsorption of the GA to SiO2 and nanosized area provides better mass transport of the GA to the electroactive sites on the electrode surface. Meanwhile, Zhang and Choi  proposed an electrorheological characteristic of Si–GO hybrid composite, hence easy transportation of electrons. Moreover, the conductive GO nanocolloids support facilitates the efficient collection and transfer of electrons to the GCE surface as described by Chen and colleagues .
Furthermore, an alternative and plausible explanation for the phenomenon observed in the synergistic effect of the nano-GO and SiO2 nanoparticles could be that of voltammetry at a three-phase junction, discussed by Oldham and Scholz et al. in their work [35,36,37,38]. They asserted that when insoluble insulating crystals like SiO2 nanoparticles adhere to an electrode, the three-phase junction (where the electrolyte solution, electrode and crystal meet) is the only feasible site for an electrochemical reaction to occur. Therefore, this implies that the redox process of gallic acid might have occurred at the three-phase junction between graphene oxide nanocolloids and SiO2 nanoparticles.
Effect of concentration changes of modifier
The effect of the changes in concentration of the nano-GO and the SiO2 nanoparticles in the modifying mixture was investigated. The following concentrations (0.1 mg/mL SiO2 nanoparticles + 2 mg/mL nano-GO; 1 mg/mL SiO2 nanoparticles + 0.2 mg/mL GO; 1 mg/mL SiO2 nanoparticles + 2 mg/mL GO and 5 mg/mL SiO2 nanoparticles + 2 mg/mL GO) were used for the modification of the glassy carbon electrode and used to determine 1 × 10−2 mol L−1 GA. The combinations studied had equal amounts of nano-GO and SiO2 nanoparticles of variable concentration, sonicated for 1 h to enable uniform dispersion and bonding of the two compounds. From the results (Fig. 6), the combination of 1 mg/mL SiO2 nanoparticles + 2 mg/mL GO nanocolloids showed the highest peak current (Ip) of 411.7 μA; hence, this combination was subsequently used for modification of the GCE.
Effect of changes in the volume of modifier on GCE
The quantity of nano-GO-SiO2 nanoparticle drop-cast on the glassy carbon electrode and the effect on the determination of 1 × 10−2 mol L1 GA in a 0.1 mol L−1 phosphate buffer at a pH of 2.0, using CV at a scan rate of 100 mVs−1, were investigated. Volumes of 2 to 5 μL were studied, and it was observed that 5 μL of nano-GO-SiO2 nanoparticles gave the highest peak current (Ip) of 405.18 μA, as a further increase in the volume made no difference, but instead reduced the peak current, as they seem to foul the electrode surface. The volume of the mixture used for the GCE modification for further experiments was 5 μL.
Effect of adsorption time
The effect of the adsorption of GA to the electrode with time was demonstrated by the detection of GA (1 × 10−3 mol L−1) in a 0.1 mol L−1 phosphate buffer of pH 2.0 at room temperature using DPV at a scan rate of 100 mV/s−1.
The voltammograms in Fig. 7 demonstrated the adsorption of gallic acid onto the nano-GO-SiO2 nanoparticle electrode by showing a sharp increase after the first scan to an optimum by the fourth to fifth minute, reaching saturation of the GA on the electrode after 6 min. This phenomenon confirms the proposed interaction mechanism of GA and the nano-GO-SiO2 nanoparticles as being an adsorption reaction and is in agreement with the work of Tashkhourian .
Effect of pH on gallic acid oxidation
The effect of the pH of the gallic acid solution towards its electrochemical activity and the activity of the nano-GO-SiO2-GCE were investigated. The electrochemical response of gallic acid (1 × 10−4 mol L−1) was studied with pH ranging from 2.0 to 8.0 using cyclic voltammetry at a scan rate of 100 mVs−1, in phosphate buffer (0.1 mol L−1). The oxidation peak and the peak potential (Ep) shown have clearly been influenced by the pH, as can be seen in Fig. 8. From the voltammograms, a pH value of 2.0 showed a well-shaped oxidation peak, with the highest peak current (Ip) of 33.72 μA.
On the other hand, as the pH increases, there was a reduction in the peak current (Fig. 8a). For example, at pH 8, the lowest peak current of 23.70 μA was observed. There was also a negative shift of the oxidation peak potential from 0.56 V downwards to 0.31 V, as can be seen in Fig. 8 a, with a gradual broadening of the oxidation peak/peak area with the increase in pH.
The relationship between the pH and the peak potential (Ep) shows a good linear relationship as can be seen in Fig. 8 b in the pH range of 2.0 to 8.0, with a linear regression equation of EpaV = 0.70 ± 0.01 − 0.05 ± 0.03 pH, with R2 = 0.9960. The slope of the regression line, 54 mV/pH, is comparable with the Nernstian value of 59 mV/pH at 25 °C, for equal number of proton and electron transfer reactions. The pH and peak potential relationship is thus consistent with a two-proton/two-electron reaction, which is consistent with other literature data [14, 20]. Based on the results, the pH condition selected as the most appropriate for the oxidation of gallic acid in this experiment was pH 2.0.
This may be attributed to the fact that GA, with a pKa of 4.4 being a weak organic acid, is most stable or is in its non-dissociated state at pH 2.0. This pH was the most appropriate for the adsorption of GA on to SiO2.
Limit of detection for the electrochemical oxidation of GA
Differential pulse voltammetry was used for the investigation of different concentrations of GA, to determine the limit of detection (LOD) at a scan rate of 100 mVs−1 and pulse amplitude of 80 mV. With the increase in concentration studied in the range of 6.25 × 10−6 to 1.0 × 10−3 mol L−1, there was a correlated increase in the peak current, whilst using the nano-GO-SiO2 nanoparticle–modified electrode. The measurements produced voltammograms (Fig. 9) in which the peak current increased proportionally with increasing gallic acid concentration. Using DPV in this measurement produced two oxidation peaks at peak potentials of 0.45 V and 0.8 V as different concentrations of GA were being determined.
Using the first main GA oxidation peak, the analytical calibration graph as seen in Fig. 9 (inset) showed a linear relationship between the peak current and the increase in gallic acid concentration, producing a linear regression equation, where IP (μA) = 221 ± 5.5, C (mmol L−1) + 9.43 ± 2 and R2 = 0.9956 within a range of 6.25 × 10−6 to 1.0 × 10−3 mol L−1. The limit of detection (LOD), defined as (3 × StdBlank)/m, where StdBlank is the standard deviation of the blank and m is the slope, was found to be 2.09 × 10−6 mol L−1.
The proposed method is comparable with other methods based on the maximum permitted antioxidant levels in food within EU and North American guidelines, which ranges from 20 to 1000 ppm (20 to 1000 mg L−1) . Hence, for GA, the permitted range would be 1.2 × 10−1 mol L−1 to 5.9 mol L−1. Thus, with an LOD of 2.09 × 10−6 mol L−1 which is far below these standard permitted levels, the electrode is comparable with others, as can be seen in Table 1.
Reproducibility and repeatability of the method
The reproducibility of the nano-GO-SiO2-modified glassy carbon electrode was investigated, by using the modified electrode to determine the oxidation peak current produced by 1.0 × 10−3 mol L−1 GA. The oxidation peak currents produced by eight replicates of 1.0 × 10−3 mol L−1 GA independently measured with eight different electrodes, with a relative standard deviation of 3.8% indicated a good reproducibility.
Meanwhile, the repeatability of the method was investigated by using a nano-GO-SiO2-modified glassy carbon electrode, for eight repetitive determination of 1.0 × 10−3 mol L−1 gallic acid. The relative standard deviation of the oxidation peak current was found to be 2.92%, which shows good repeatability of the method.
Stability of electrode
The stability of the nano-GO-SiO2 nanoparticle–modified glassy carbon electrode was also investigated. The electrode was kept for 30 days and was used for the determination of 1.0 × 10−3 mol L−1 gallic acid and the oxidation peak currents produced from the readings of the first day and that of the 30th day showed an RSD of 6.58%, depicting a relatively good stability.
Selectivity of the method
The selectivity of the modified glassy carbon electrode was tested, by the simultaneous detection of GA and uric acid (UA) in a 0.1 mol L−1 phosphate buffer solution at a pH of 2.0 using CV at a scan rate of 100 mVs−1, at room temperature. This was done by the addition of 0.48 mmol L−1 of GA into a 10-mL cell of 0.36 mmol L−1 uric acid and voltammograms recorded, as seen in Fig. 10 a. A concentration of 0.48 mmol L−1 of GA was used, because it was at this concentration that a significant distinction of the voltammograms of GA and UA was observed. GA and UA showed distinct peaks at approximately 0.51 V and 0.64 V, respectively, with the smaller second GA oxidation at a peak potential of 0.88 V. Meanwhile, the voltammograms of UA on its own without GA showed an oxidation peak at 0.64 V and that of GA on its own show oxidation peaks at 0.5 V for the first peak and at 0.88 V for the second smaller peak (Fig. 10b).
To confirm the selectivity of the electrode, the UA concentration was kept constant at 0.36 mmol L−1 in a 10-mL voltammetric cell and then aliquots of 100 μL of 10 mmol L−1 GA were intermittently added and then vigorously stirred for 2 min. Voltammograms were then recorded, as can be seen in Fig. 10 b. In this case, the nano-GO-SiO2 nanoparticle–modified GCE produced the voltammograms as can be seen in Fig. 10 b, where increasing the concentration of GA showed a proportional increase in the peak current. Hence, Fig. 10 b (inset) shows a linear relationship between the peak current and the increase in gallic acid concentration, producing a linear regression equation where IP (μA) = 13.0 ± 0.6, C (mmol L−1) + 1.41 ± 0.20 and R2 = 0.9915 within a range of 0.1 to 0.48 mmol L−1.
The interference of various species in the determination of 1 × 10−3 mol L−1 gallic acid was also investigated (Table 2). This was carried out by adding different amounts of foreign ions to a known quantity of the analyte as it was being determined. The foreign species that were used in this case for the investigation of interference were K+, Ca2+, Fe3+ and Na+, then ascorbic acid, caffeine, caffeic acid and quercetin, respectively. The selection of some of the cations was based on reports of their complexation with GA in the literature [40, 41] and the possibility of these compounds interfering with the electrochemical determination of GA. The tolerable limits of these ions for interference was defined as the highest amounts of foreign ions that would produce an error of not more that 5% in the determination of the analyte, which in this case is gallic acid. With relative standard error values of less than 5%, these ions did not interfere with the determination of gallic acid at the first GA oxidation peak, as can be seen in the results shown in Fig. 11. However, there were some minor changes in the second GA oxidation peak, which is not normally the peak used for electrochemical determination of GA.
Analytical application of the method
This analytical method with the nano-GO-SiO2 nanoparticle–modified electrode was used for the determination of GA in red wine, white wine and orange juice. The real samples (i.e. red wine, white wine and orange juice) were used based on previous work from other group showing the presence of GA in those types of samples . However, in red wine, white wine and orange juice GA were not detected. This may be due to the origin and nature of the samples. Hence, spiking methodology was applied to all samples to clearly demonstrate the potential of our proposed sensor as shown in Table 3. The experiments were done by using 10 mL of the real samples as blanks and then the standard addition of known concentrations of aliquots of gallic acid within a concentration range of 9.0 × 10−4 to 4.7 × 10−3 mol L−1 was added to the cells containing the real samples and the voltammograms recorded and the results calculated as can be seen in Table 3.