Introduction

Quantitative determination of glucose and hydrogen peroxide is in great demand in clinical applications. Enzymatic electrochemical sensor such as test stripes have received extensive attention for its high sensitivity, good selectivity and fast response. But the inherent limitations originated from the intrinsic nature of enzymes such as insufficient stability and reproducibility hinder their further development [1].

To resolve these issues, researchers pay more attention to non-enzymatic electrochemical sensor, which can avoid tedious immobilization procedures of enzyme and offer numerous advantages, such as excellent stability, favorable reproducibility and cost effectiveness [2]. However, compared with enzymatic electrochemical sensor, the selectivity of some non-enzymatic sensors are not suitable for practical application [3]. Therefore, non-enzymatic sensor with excellent electrochemical activity and selectivity are urgently needed. Precious metal nanomaterial, especially Au-based nanoparticles, is of particular interest to fabricate non-enzymatic electrochemical sensor for their excellent conductivity and superior electrocatalytic properties [4]. The gold nanoparticles (AuNPs) [5,6,7], lamellar-ridge architectured gold (lamellar ridge-Au) [8], gold nanorods (AuNRs) [9, 10], and nanoporous gold [11] have been developed to construct non-enzymatic glucose sensor in alkaline solution. The oxidation of glucose is dependent on the crystalline structures of nanoparticles [12]. Cubic gold nanoparticles with dominant {100} facets exhibited superior electrocatalytic activity toward glucose oxidation [13], but the synthesis process is somewhat tedious and difficult [14, 15]. The AuNBPs with two sharp tips is another type of elongated Au nanocrystal and possesses {100} facets. Besides, there are a large fraction of the surface atoms presented on edges and corners of AuNBPs, which may provide more incipient gold oxide. Thus, AuNBPs expect to improve the sensitivity of the sensors.

The aggregation of nanoparticles during the electrocatalytic process lead to decrease the electrocatalytic activity. To overcome this problem, a variety of materials such as polypyrrole (PPy), polyaniline (PANI), reduced graphene oxide (rGO) and MWCNTs have been investigated as the supporting material. Among these materials, MWCNTs have received great interest primarily for their excellent electrical conductivity and large specific surface area [16]. Due to the synergistic effect of AuNBPs and MWCNTs, the electrocatalytic activity of AuNBPs can be further improve.

A electrochemical sensing platform based on AuNBP/MWCNTs is developed to detect glucose and hydrogen peroxide (Scheme 1). The electrochemical sensor shows excellent performance such as outstanding selectivity, high sensitivity, wide linear range, good long-term stability and reproducibility due to the merits of AuNBPs and MWCNTs. Furthermore, it has the possibility of application in real sample analysis.

Scheme 1
scheme 1

Schematic illustration of the AuNBP/MWCNT nanocomposites that were used to modify a glassy carbon electrode (GCE) for the detection of glucose and hydrogen peroxide

Full size image

Material and methods

Chemicals and reagents

Gold(III) chloride trihydrate (HAuCl4·3H2O, ≥49.0% Au basis) was purchased from Sigma-Aldrich (https://www.sigmaaldrich.com/). Sodium citrate dihydrate, sodium borohydride (NaBH4), hexadecyl trimethyl ammonium bromide (CTAB), hydrochloric acid (HCl), silver nitrate (AgNO3), ascorbic acid (AA), hexadecyltrimethylammonium chloride (CTAC), ammonia solution (NH3·H2O, 25–28%), hydrogen peroxide solution (30 wt% in H2O), potassium chloride (KCl), nitric acid (HNO3), potassium ferricyanide (K3[Fe(CN)6]), potassium ferrocyanide trihydrate (K4[Fe(CN)6]), potassium nitrate (KNO3), D-fructose, saccharose, D-(+)-maltose monohydrate, sodium hydroxide (NaOH), D-(+)-glucose, uric acid (UA), dopamine hydrochloride (DA), 4-Acetamidophenol (AAP), sodium phosphate dibasic dodecahydrate (Na2HPO4·12H2O), potassium phosphate monobasic (KH2PO4), chitosan, and multi-walled carbon nanotubes (MWCNTs, 2~5 nm diameter, 10~30 nm length, >95%) were purchased from Aladdin Biological Technology Co. Ltd. (http://www.aladdin-e.com/). For the detection of hydrogen peroxide, 0.1 M phosphate buffer (PB, pH 7.0) was prepared using Na2HPO4·12H2O and KH2PO4 without deoxygenation.

Synthesis of gold nanobipyramids (AuNBPs) and preparation of sensors

The AuNBPs were synthesized in aqueous solution by seed-mediated growth methods with minor modification [17, 18]. The experiment details is described in the the ESM.

Before the modification, a glassy carbon electrode (GCE) was polished with 0.05 μm alumina slurry and thoroughly rinsed by ultrasonication in ethanol and deionized water. The AuNBP/MWCNTs were prepared by mixing MWCNTs (2.3 mg) with AuNBPs suspension (500 μL) under the ultrasound condition. 20 μL of the well dispersed suspension was dropped onto the polished GCE. After drying, 4 μL of chitosan solution (0.5%) was dropped onto the modified GCE. The AuNBPs or MWCNTs modified GCE prepared under the same conditions were used as control.

Apparatus and electrochemical measurements

The preparation process of the AuNBPs were tested by UV-vis spectrum (UV-vis, PERSEE TU-1901). The morphologies of the AuNBPs were investigated by field emission scanning electron microscopy (FESEM, HITACHI SU8010). The crystal structure of the prepared AuNBPs were testified by X-ray diffraction patterns (XRD, Bruker D8 Advance) with Cu Kα radiation. X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi) with Al Kα source was employed to analyse the composition of AuNBPs.

The CHI 660E electrochemical workstation equipped with three-electrode arrangement was employed to conduct the electrochemical measurements. The working electrode was the AuNBPs, MWCNTs or AuNBP/MWCNTs modified GCE. The Pt foil and a saturated calomel electrode (SCE) served as the counter electrode and the reference electrode, respectively. All potentials are reported with respect to the SCE throughout this paper.

Results and discussion

Choice of materials

The irregular morphologies of AuNPs with more corners and edges result in more incipient gold oxide, which may act as a fast redox mediator for the oxidation of glucose [19]. It was report that glucose oxidation is strongly dependent on the shape of Au nanocrystals and {100} facets of Au nanocrystals are significantly more active for glucose oxidation. What’s more, Au nanocrystal catalysts exhibit remarkably high stability during the electrocatalytic oxidation process [13]. So, the AuNBPs are expected to be a promising electrocatalyst. To further improve the sensitivity of the sensor, MWCNTs with excellent conductivity are used as supporter to form nanocomposites with AuNBPs.

Characterization of the gold nanobipyramids (AuNBPs)

As shown in Fig. S1(A) and Fig. S1(B), the AuNBPs with uniform size about 76 nm are well dispersed, and there are a small amount of AuNPs about 50 nm coupling with the AuNBPs. This phenomenon is consistent with the result of UV-vis spectrum (Fig. S1(C)). For comparison, the AuNRs about 51 nm are also synthesized (Fig. S2). The crystallinity was investigated by XRD, as shown in Fig. 1(a). For the two typers of nanoparticles, the peaks positioned at 38.2o, 44.2o, 64.4o, 77.4o and 81.5o correspond to the (111), (200), (220), (311) and (222) plane of face-centered cubic (fcc) of Au, respectively. What’s more, the intensity ratios of (200) to (111) planes for AuNBPs and AuNRs are 0.62 and 0.51, respectively. This indicates that AuNBPs have more {100} facets than that of AuNRs.

Fig. 1
figure 1

a XRD of AuNBPs (a) and AuNRs (b); b The CVs of the MWCNTs (a), AuNBPs (b), AuNBP/MWCNTs (c), bulk Au electrode (d) and AuNRs (e) in N2-saturated 0.5 mM H2SO4 at scan rate of 50 mV/s

Full size image

Electrochemical characterization

As shown in Fig. 1(b), the cyclic voltammetry (CV) in 0.5 M H2SO4 solution was conducted to investigate the electrochemical behaviors. Both of the AuNBPs (b), AuNBP/MWCNTs (c) and AuNRs (e) exhibit significant electro-chemical responses similar to the bulk Au electrode (d), which correspond to the signatures of polycrystalline Au in acidic medium. The reduction peak current at +0.90 V of AuNBPs is 14 times as much as that of the bulk Au electrode, indicating that the electrocatalytic activity and surface area of the AuNBPs are bigger than those of bulk Au [20]. What’s more, the reduction peak current increases significantly for introducing the MWCNTs into the AuNBPs to form nanocomposites. It suggest that the AuNBP/MWCNTs can enhance the electron transfer rate due to the synergistic effect between AuNBPs and MWCNTs thus improve the sensitivity of the electrochemical sensor [5, 21].

Response to glucose

The oxidation of glucose is strongly dependent on the number of AuOH sites [13]. As shown in Fig. S3(A), the oxidation current of AuNBP/MWCNTs toward glucose in 0.1 M NaOH is bigger than that in 0.1 M PB. This indicates that there are more AuOH sites in alkaline solution. Consequently, 0.1 M NaOH is chosen as electrolyte to improve the electrocatalytic activity of AuNBP/MWCNTs. As shown in Fig. 2 (a), the CVs of MWCNTs (a, a’), AuNBPs (b, b’), and AuNBP/MWCNTs (c, c’) modified electrode in 0.1 M NaOH solution with (a’, b’, c’) or without (a, b, c) 20 mM glucose were conducted. For the MWCNTs (a, a’), no redox peaks are observed with or without glucose, indicating that the overpotential of glucose oxidation at MWCNTs is high. The electrochemical behavior of AuNBP/MWCNTs (c) in 0.1 M NaOH without glucose is similar to that of AuNBPs (b). The peak I (−0.10 V) and peak II (+0.40 V) are attributed to the oxidation of AuNBPs to form AuOH by the adsorbed OH− and the formation of Au oxides by the oxidation of AuOH, respectively. The cathodic peak around +0.10 V (peak III) is attributed to the reduction of Au oxides to AuOH, and the peak at −0.19 V (IV) resulted from the reduction of AuOH into AuNBPs [22]. It is worth noticing that there are many differences occurred with 20 mM glucose for AuNBPs (b’) and AuNBP/MWCNTs (c’). In the positive scan, the peak V and VI are attributed to the oxidation of glucose to dehydrogenated glucose, and subsequently to gluconolactone. As the potential goes to more positive, the current decreases due to the formation of Au oxides [19]. Whereas in the negative scan, a sharp peak (VII) is attributed to the oxidation of glucose resulting from the reduction of the Au oxides [23]. What’s more, the oxidation current of AuNBP/MWCNTs is bigger than that of AuNBPs and the oxidation potential (peak VI) shifts more than 50 mV toward negative potential as compared to that of AuNBPs. It reveals that the AuNBP/MWCNTs can enhance the electrocatalytic activity and decrease the overpotential of glucose oxidation. It is shown in Fig. S3(B) that the AuNBPs show better electrocatalytic activity than that of AuNRs. This because more {100} facets and incipient gold oxide are available in AuNBPs. As shown in Fig. 2(b), when the glucose concentration increase from 10 to 80 mM, the peak current of glucose oxidation (I, II, III) gradually increase. It indicates that the electrochemical sensor based on AuNBP/MWCNTs can be used to detect of glucose with good sensitivity.

Fig. 2
figure 2

a The CVs of the MWCNTs (a, a’), AuNBPs (b, b’), AuNBP/MWCNTs (c, c’) in 0.1 M NaOH with (a’, b’, c’) and without (a, b, c) 20 mM glucose; b The CVs of the AuNBP/MWCNTs in 0.1 M NaOH containing different concentration of glucose; c The CVs of the AuNBP/MWCNTs in 0.1 M NaOH with 20 mM glucose at various scan rates, Insert: The calibration curve of current vs. square root of scan rate; d Amperometric response of the MWCNTs (a), AuNBPs (b), AuNBP/MWCNTs (c) at +0.15 V; e Amperometric response of the AuNBP/MWCNTs at +0.15 V with successive additions of different concentrations of glucose; f Dependence of the current as a function of the glucose concentration

Full size image

A further investigation of glucose oxidation on AuNBP/MWCNTs was conducted by CVs in 0.1 M NaOH with 20 mM glucose at different scan rates from 20 to 300 mV/s. Obviously, the peak current increases along with the increasing applied scan rate and the current has linear relation with the square root of the scan rate (Fig. 2(c)). This indicates that the glucose electro-oxidation at the AuNBP/MWCNTs is a diffusion-controlled process.

For the amperometric method used to detect glucose, the working potential is crucial to obtain high sensitivity and good selectivity. As shown in Fig. S4(A), +0.15 V is selected as the optimal working potential for further use. After each addition of 2 mM glucose, a stepwise current response is observed. The response time reaching 95% of the steady-state current is less than 2 s for the excellent electrocatalytic activity of AuNBP/MWCNTs. It is shown in Fig. 2(d), the current of the AuNBP/MWCNTs towards 2 mM glucose is bigger than that of AuNBPs and MWCNTs at +0.15 V, which agrees with the result of CVs.

Under +0.15 V, the linear relationship of AuNBP/MWCNTs is obtained by the successive addition of glucose (Fig. 2(e)). As shown in Fig. 2(f), the current response increases linearly along with the glucose concentration from 10 μM to 36.7 mM, which covers the normal physiological glucose level of 4.4–6.6 mM [24]. The sensitivity and limit of detection (LOD) of the sensor towards glucose is 101.2 μA mM−1 cm−2 and 3.0 μM (S/N = 3), respectively. Compared with other non-enzymatic glucose sensors, the advantages of the sensor based on AuNBP/MWCNTs are the wider linear range and the sensitivity warrant further improvement (Table S1). The good performance of AuNBP/MWCNTs might result from more incipient gold oxide provided by the AuNBPs and the synergistic effect of AuNBPs and MWCNTs.

Response to hydrogen peroxide

As shown in Fig. 3(a), no obvious redox peaks is observed at all the modified electrodes without hydrogen peroxide. In presence of 15 mM hydrogen peroxide, no reduction peak is observed at MWCNTs (a’). This suggests that the electro-reduction of hydrogen peroxide on MWCNTs required a large overpotential. There is a reduction peak appeared at AuNBPs (b’) and AuNBP/MWCNTs (c’) around −0.50 V, which is ascribed to reduction of hydrogen peroxide. The peak current of AuNBP/MWCNTs is bigger than that of AuNBPs. In addition, the peak current originating from the hydrogen peroxide reduction increases along with the concentration of hydrogen peroxide (Fig. 3(b)). The results demonstrate that hydrogen peroxide is easily reduced on AuNBP/MWCNTs due to the excellent electrocatalytic activity of AuNBP/MWCNTs. Furthermore, the hydrogen peroxide reduction on the AuNBP/MWCNTs is a surface-controlled process (Fig. 3(c)).

Fig. 3
figure 3

a The CVs of the MWCNTs (a, a’), AuNBPs (b, b’), AuNBP/MWCNTs (c, c’) in 0.1 M PB with (a’, b’, c’) and without (a, b, c) 15 mM H2O2; b The CVs of the AuNBP/MWCNTs in 0.1 M PB containing different concentration of H2O2; c The CVs of the AuNBP/MWCNTs in 0.1 M PB with 15 mM H2O2 at various scan rates, Insert: The calibration curve of current vs. scan rate; d Amperometric response of the MWCNTs (a), AuNBPs (b), AuNBP/MWCNTs (c) at −0.50 V; e Amperometric response of the AuNBP/MWCNTs at −0.50 V with successive additions of different concentrations of H2O2; f Dependence of the current as a function of the H2O2 concentration

Full size image

It is shown in Fig. S4(B), −0.50 V is selected as optimal working potential. Moreover, the amperometric response of AuNBP/MWCNTs at −0.50 V is the highest. It indicates that the electrocatalytic activity of the AuNBP/MWCNTs towards hydrogen peroxide reduction is the best (Fig. 3(d)). The reduction current response increases along with the concentration of hydrogen peroxide under −0.50 V (Fig. 3(e)). It exhibits a good linear range from 5.0 μM to 47.3 mM with a LOD of 1.5 μM and the sensitivity of 170.6 μA mM−1 cm−2 (Fig. 3(f)). Compared with other non-enzymatic electrochemical sensors for detection of hydrogen peroxide, the results are comparable or even better than other reports (Table S2).

Selectivity, reproducibility and stability

The good selectivity, reproducibility and stability of sensors are the key factors for practical application. For glucose detection, DA, UA, AA, AAP, fructose, saccharose, maltose, Ca2+ and Cl− may interfere the accurate detection of glucose [25, 26]. As shown in Fig. 4(a), the addition of 10 μM DA, 0.1 mM UA/AA/AAP, 2 mM fructose/saccharose/maltose/Ca2+ and 0.15 M Cl− into 0.1 M NaOH results in negligible signal changes compared with that of 2 mM glucose. This suggests that the electrochemical sensor based on AuNBP/MWCNTs for glucose detection have a good selectivity. It was reported that the oxidation of the co-existing reducing species on Au-based modified electrode usually occurs at +0.30 to +0.40 V [27]. The lower working potential (+0.15 V) contributes to the excellent selectivity. Furthermore, the alkaline electrolyte can minimize the interference from the co-existing reducing species by providing a restricted environment [28]. It is shown in Fig. 4(b), the non-enzymatic electrochemical sensor also exhibits excellent selectivity for hydrogen peroxide detection.

Fig. 4
figure 4

a Amperometric response of the AuNBP/MWCNTs at +0.15 V toward glucose (2 mM), DA (10 μM), UA/AA/AAP (0.1 mM), fructose/saccharose/maltose/Ca2+ (2 mM) and Cl− (0.15 M); b Amperometric response of the AuNBP/MWCNTs at −0.50 V toward H2O2 (0.1 mM), AA/glucose/UA (1 mM)

Full size image

The current responses of five-paralleled electrodes towards 2 mM glucose and 2.5 mM hydrogen peroxide are recorded and the relative standard deviation (RSD) is 2.3% (Fig. S5 (A)) and 3.5% (Fig. S5 (B)), respectively. This demonstrated that the sensor have a good reproducibility. In addition, the current response still retain up to 98.0% of the original value for glucose (Fig. S5(C)) and 94.7% for hydrogen peroxide (Fig. S5(D)) after 30 days of storage in air at room temperature, which demonstrates the satisfactory stability of the sensor.

Real sample analysis

The feasibility of the sensor is testified by detecting of the concentration of glucose in human serum and hydrogen peroxide in hydrogen peroxide-based antibacterial lotion (3%), respectively. The glucose concentration in human serum detected by our resultant sensor agrees with the results detected by glucometer (ONE TOUCH Ultra, LIFESCAN) and the concentration of hydrogen peroxide in hydrogen peroxide-based antibacterial lotion (3%) detected by our resultant sensor is consistent with the mark (2.49 mM). As listed in Table 1, the non-enzymatic sensor is demonstrated to be feasible for practical application with acceptable recovery and RSD.

Table 1 Non-enzymatic glucose sensor based on AuNBP/MWCNTs was applied to detect the concentration of glucose in human serum and hydrogen peroxide in hydrogen peroxide-based antibacterial lotion (3%)
Full size table

Conclusions

A non-enzymatic electrochemical sensor on the basis of using AuNBP/MWCNTs as an efficient electrode material has been developed to detect of glucose in human serum and hydrogen peroxide in hydrogen peroxide-based antibacterial lotion (3%). The sensor possess excellent sensitivity, good selectivity, high reproducibility and stability. The outstanding performance of the sensor is ascribed to the synergistic effect and more incipient gold oxide provided by the AuNBPs. The major limitation of the sensor for detection of glucose is that it need to work in strong alkaline solution.