Transfer-free CVD graphene for highly sensitive glucose sensors

1. Introduction

Graphene is a promising biosensor material for targeting glucose, thrombin, dopamine, cancer tumor markers and so on [[1], [2], [3], [4]] due to its extraordinary electrical conductivity [5], large specific surface area [6] and high capability of functionalization [7]. Rapid and sensitive glucose sensors are crucial for detecting the concentration of glucose in blood, which can be used to diagnose and prevent diabetes. Glucose oxidase (GOx) has been recognized as an effective enzyme for selectively monitoring glucose. Graphene is typically used to promote electron transfer between the electrode and GOx because it not only shows high electrical conductivity but also its large surface area would facilitate the attachment of glucose. In addition, the surface functional groups (epoxy, hydroxyl, carbonyl, etc.) of modified graphene together with the numerous structure defects can immobilize GOx and other active species [1]. Previous studies have demonstrated that graphene or its derivatives are effective for improving the performances of glucose sensors [8].

Reduced graphene oxide (RGO) is widely used to immobilize GOx for glucose detection with a high sensitivity and low detection limit due to its highly defective structure and functional groups [[9], [10], [11], [12], [13]]. However, the complex composition of electrode might lower the detection accuracy due to the possible interference of other components in the realistic blood test. Moreover, residual Mn²⁺ from the RGO would lower the activity of GOx [[14],[15]]. High-quality chemical vapor deposition (CVD) graphene films grown on metals [[16],[17]] have been proposed as an alternative to RGO due to its high electrical conductivity and uniform structure over large area. Since it is difficult to immobilize GOx on the inert surface of intrinsic CVD graphene, oxygen-plasma treatment is generally used to functionalize CVD graphene, which shows improved hydrophilicity [[18],[19]]. Thus-formed oxygen-containing groups are effective for immobilizing GOx on the surface of CVD graphene [[20],[21]]. However, one major issue with metal-based CVD graphene is that it must be transferred from metal substrate to the electrode [[21],[22]]. As a result, the transferred graphene suffers from the residue contamination of polymer support such as poly(methyl methacrylate) (PMMA) and heavy metal ions such as Cu²⁺, Ni²⁺ and Fe³⁺. PMMA residue is responsible for buffering the rate constant by altering the charge balance at the graphene electrode [23] and the metal impurities show enzymatic inhibition effects [[24],[25]]. Therefore, it is highly desirable to fabricate the graphene electrode by developing a simple and clean method.

Here, we directly grew few-layer graphene on the SiO₂/Si substrate (G/SiO₂/Si) without transfer and then fabricated the graphene-based glucose sensor by sequentially immobilizing glucose oxidase and depositing Nafion layer on the surface functionalized by oxygen-plasma treatment. Our transfer- and metal-free process shows distinct advantage over the common metal-CVD method in improving the electrochemical performance by eliminating the contamination of transfer-residue [[23],[26],[27]]. The defective few-layer graphene is also easy to be functionalized to generate electrochemically active sites for glucose sensor [28]. The fabrication process is simple and thus-obtained glucose sensor shows significantly improved electrochemical performances.

2. Experimental

2.1. Metal-free CVD growth of few-layer graphene film

The growth was carried out in a quartz tube reactor (22 mm inner diameter) inside a horizontal CVD furnace at atmospheric pressure. First, the SiO₂/Si substrates (SiO₂ = 290 nm) were cleaned sequentially by deionized water, acetone and isopropanol prior to loading into the quartz tube. The tube was then flushed by the argon flow and heated up to growth temperature. After that, a mixture of H₂ and CH₄ flow was injected and the Ar flow was turned off. An optimized parameter was used for growing few-layer graphene film with H₂/CH₄ = 75/17.5 sccm at 1100 °C for 2 h. After growth, the CH₄/H₂ flow was turned off and the furnace was cooled rapidly to room temperature under the protection of Ar flow.

2.2. Plasma treatment of G/SiO₂/Si

G/SiO₂/Si was first bonded to copper wire with conductive silver paint (test area: 5 mm × 5 mm). Then, G/SiO₂/Si was exposed to the oxygen plasma with an optimized parameter of 40 W under 0.3 mbar chamber pressure with a 5 sccm oxygen flow. After the treatment, oxygen-plasma treated graphene (OPG) was obtained.

2.3. Fabrication of graphene-based glucose sensor

OPG/SiO₂/Si was used as the working electrode (WE). 10 μL GOx solution (1900 U ml^-1, Sigma) was dropped on the WE and dried for 1 h at room temperature to form the GOx layer. A thin Nafion (5 wt%, Sigma) layer was then formed on top of GOx layer as the protective film with the same process. Thus-prepared graphene electrode is denoted as Nafion/GOx/OPG/SiO₂/Si.

2.4. Transfer of graphene and OPG

For measuring the optical transmittances of graphene before and after oxygen-plasma treatment, the graphene or OPG was transferred using a PMMA-supported etching method. Typically, the pristine sample was first spin-coating with PMMA and then immersed in the solution of NaOH (2.0 M) to detach from the SiO₂/Si substrate. Subsequently, the detached film was thoroughly rinsed in deionized water and fished onto a quartz, followed by baking on a heating plate. Finally, PMMA was removed by rinsing with acetone.

2.5. Characterizations

Raman spectrometer (Jobin Yvon Lab RAM HR800), optical microscope (Nikon Eclipse LV100) and atomic force microscope (AFM, Multimode 8, Brucker, Peakforce mode) were used to characterize the microstructure and morphology of graphene films. UV-vis-NIR spectrometer (Varian Cary 5000) and transmission electron microscopy (TEM, FEI Tecnai F20, 200 kV) were used to determine the layer-number of graphene transferred onto quartz or TEM grid. The wetting property of as-grown graphene and OPG films were measured by using contact angle meter (Powereach JC2000D1). X-ray photoelectron spectroscopy (XPS) measurements were performed on an ESCALAB 250 instrument with Al K_α radiation source. Electrochemical characterization by cyclic voltammetry and chronoamperometry were performed in a phosphate buffer solution (PBS, Sinopharm) using a three-electrode configuration with an electrochemical workstation (Multi Autolab M204), in which a platinum plate and a saturated calomel electrode (SCE) served as the counter and reference electrode, respectively.

3. Results and discussion

Fig. 1 illustrates the fabrication process of graphene glucose sensor involving graphene growth, oxidation, enzyme immobilization and Nafion deposition. In contrast to the typical metal-based CVD, graphene was directly grown on the SiO₂/Si substrate to form G/SiO₂/Si without transfer. This process is not only simple but also free of the contamination of transfer residue. After bonding with the copper wire, the G/SiO₂/Si was then functionalized by using oxygen plasma, followed by sequentially coating with GOx and Nafion layers to form the Nafion/GOx/OPG/SiO₂/Si electrode.

View original graphic| Download| PPT slide

Fig. 1.   Schematic fabrication process of the graphene sensor. First, few-layer graphene is directly grown on a SiO₂/Si substrate (a) by metal-free thermal CVD to form the G/SiO₂/Si sample (b). The Nafion/GOx/OPG/SiO₂/Si electrode is then formed by bonding the sample with copper wire (c), followed by sequential treatment with oxygen plasma (d), coating with GOx (e) and Nafion layers (f).

3.1. Characterization of graphene

We first evaluated the quality of graphene films by using Raman spectra before and after oxygen-plasma treatment. The top Raman spectrum in Fig. 2(a) shows a large ratio of D band to G band (intensity ratio I_D/I_G = 0.94) and a low I_2D/I_G of 0.52, which is typical for the defective few-layer graphene grown by metal-free CVD method [[29],[30]]. We then used the transmittance spectra and TEM image to determine the layer-number of graphene film. As shown in Fig. 2(b), the 92.6% transmittance at 550 nm suggests a film of 3-4 layers in average. This structure is further confirmed by high resolution TEM result (Fig. S1 in supplementary information). The optical image indicates that the film is uniform over a large area. It is free of structural damage that is frequently observed for transferred samples. Moreover, the AFM image reveals that the graphene film is formed by the stitching of nano-sized domains with abundant edges (Fig. 2(c)), which is also responsible for the large D peak in Raman spectrum. It can be expected that such numerous chemically active edges would facilitate the subsequent functionalization by oxygen plasma [31].

View original graphic| Download| PPT slide

Fig. 2.   Characterization of pristine G/SiO₂/Si and OPG (after the oxygen-plasma treatment for 20 s): (a) Raman spectra; (b) optical transmittance spectra (transferred on quartz); AFM images (on SiO₂/Si substrates) of the pristine graphene (c) and OPG (d). The insets of (b) show the corresponding optical micrographs.

After exposing the sample to oxygen plasma for 10-30 s, the I_D/I_G of its Raman spectra increases significantly from 0.94 to 1.65-2.24 and I_2D/I_G drops from 0.52 to 0.21-0.27 together with the broadened G band, which is consistent with the formation of graphene oxide (OPG) [18]. However, Raman signal disappeared after 60 s plasma exposure, suggesting that graphene film was completely etched away by intense plasma treatment (Fig. S2). The time of plasma treatment is further optimized according to the cyclic voltammograms (CVs) of graphene electrodes. As shown in Fig. S3, the peak current of the sample treated for 20 s is much higher than that treated for 10 s. Although the 30 s treatment causes a higher peak current, the improvement is not significant. More importantly, it simultaneously increases the background current, which is undesirable for glucose detection. Therefore, the optimum treatment time is selected to be 20 s. The conversion of graphene into OPG increases the transmittance by 4.2% to 96.8%, which can be clearly seen from the lowered optical contrast (Fig. 2(b) insets). Note that the OPG film remains uniform after the plasma treatment. AFM measurement reveals that the surface roughness of OPG is further reduced, as shown in Fig. 2(d).

The formation of OPG also significantly improves the hydrophilicity of G/SiO₂/Si. As shown in Fig. 3, the pristine G/SiO₂/Si shows a contact angle of 90°, consistent with reported values for graphene [32]. After 20 s plasma treatment, the contact angle is greatly reduced from 90° to 26° for the OPG/SiO₂/Si. This markedly increased hydrophilicity can be attributed to the high affinity of oxygen-containing groups of graphene oxide to water molecules [18].

View original graphic| Download| PPT slide

Fig. 3.   Contact angle measurements of pristine graphene and OPG. Pristine graphene: θ_c = 90°; OPG: θ_c = 26°. Inset: Photographs of wetting properties of pristine graphene (left) and OPG (right).

We further used XPS to semi-quantitatively investigate the effect of plasma treatment on the chemical composition of G/SiO₂/Si. As shown in Fig. 4(a), the full XPS spectra show that the oxygen content of OPG/SiO₂/Si increases apparently as compared with the pristine G/SiO₂/Si. Further analysis reveals that the pristine graphene film shows a prominent graphitic peak of sp² C-C bond near 284.6 eV and sp³ C-C bond near 285.2 eV (Fig. 4(b)) [33], consistent with the defective structure observed in Raman spectrum. In contrast, as shown in Fig. 4(c), the C1s peak of OPG film can be assigned to four peaks, with two extra C-O (286.8 eV) and O-C=O (288.9 eV) peaks [[18],[21]]. According to the integrated area ratios of the resolved C1s peaks, the percentage of the four types of bonds in the pristine graphene and OPG films are compared in Fig. 4(d). After oxygen-plasma treatment, the percentage of C-C (sp²) and C-C (sp³) peaks decreases from 70.5% to 61.4% and 29.5% to 21.6%, respectively. The remaining 17% are C-O (8.6%) and O-C=O (8.4%). It has been proved that carboxyl plays a key role in bonding the amino group of GOx enzyme [34]. Therefore, this type of OPG could facilitate the immobilization of enzyme. Particularly, signals related to metals such as Mn, Cu, Ni and Fe (600-900 eV) are not detected [35], which further confirms the metal-free preparation of OPG/SiO₂/Si sample.

View original graphic| Download| PPT slide

Fig. 4.   (a) Full XPS spectra comparison of pristine graphene and OPG, XPS C1s spectra of (b) the pristine graphene and (c) OPG treated for 20 s and (d) percentage of integrated peak area ratios of the resolved C1s peaks in pristine graphene and OPG.

3.2. Electrochemical characterization of glucose sensors

Fig. S4 shows the photographs of pristine G/SiO₂/Si, OPG/SiO₂/Si and Nafion/GOx/OPG/SiO₂/Si electrodes. The notable changes in optical contrast or color can be clearly observed after depositing the GOx and Nafion layers. The CVs of these electrodes were measured in N₂ saturated PBS (0.1 M, pH = 7) at a scan rate of 100 mV s^-1 to compare the electrochemical performances. As shown in Fig. 5(a), a reduction current peak is only observed for the red curve and the formal potential is estimated to be nearly -0.5 V (vs. SCE), consistent with the reported results [21]. A comparison with the blue curve indicates that the redox peak originates from the presence of GOx. The effect of scan rate on the cyclic voltammetric performance of the Nafion/GOx/OPG/SiO₂/Si electrode is shown in Fig. 5(b). The redox peak current increases linearly as a function of scan rate (inset in Fig. 5(b)), indicating a surface-controlled electron transfer process.

View original graphic| Download| PPT slide

Fig. 5.   (a) Cyclic voltammograms of SiO₂/Si, G/SiO₂/Si, OPG/SiO₂/Si and Nafion/GOx/OPG/SiO₂/Si modified electrodes in 0.1 M PBS (pH = 7) saturated with N₂ at a scan rate of 100 mV s^-1; (b) Cyclic voltammograms at various scan rates from 10, 25, 50, 100, 150, 200 and 250 to 300 mV s^-1, respectively. Inset: plot of peak current versus scan rate; (c) Amperometric response at -0.5 V for Nafion/GOx/OPG/SiO₂/Si electrode with dropwise addition of glucose concentrations of 400 μM to 2000 μM in O₂ saturated 0.1 M PBS. (d) Linear regression curve of current versus glucose concentrations.

Chronoamperometric technique is a reliable and sensitive method to evaluate the electrocatalytic activity of such electrochemical biosensors. The amperometric responses of Nafion/GOx/OPG/SiO₂/Si electrode to glucose were investigated with an applied potential of -0.5 V, where glucose (1.0 M) was sequentially added into a beaker containing 25 mL of O₂ saturated PBS (0.1 M, pH = 7) and the result is shown in Fig. 5(c). A sudden increase in current responses can be observed after each addition of glucose solution. Such behavior is typical for highly sensitive electrodes, which is generally attributed to the reaction dynamics dominated by the concentration variation of glucose in the solution. The corresponding calibration curve presented in Fig. 5(d) is linear over a wide concentration range from 400 μM to 2000 μM glucose with a slope of 4.04 × 10^-3 μA μM^-1 and a correlation coefficient of 0.988. For an electrode surface area of 0.25 cm², the fabricated Nafion/GOx/OPG/SiO₂/Si biosensor shows a sensitivity of 16.16 μA mM^-1 cm^-2, which outperforms all the transfer-CVD-graphene-based sensors [[21],[22],[36], [37], [38]]. It has been demonstrated that the residual PMMA caused by common transfer process severely retards the electron transfer between graphene and redox species, thus degrading the electrochemical reactivity [23]. The superior sensitivity of our sensor demonstrates the distinct advantage of using transfer-free graphene in developing high-performance glucose sensors. The limit of detection (LOD) is estimated to be 124.19 μM (S/N = 3). Table 1 summarizes the glucose detection performances of sensors with different electrodes ever reported. The sensitivity of our electrode outperforms those RGO hybrids such as MnO₂-RGO [39] and AuNPs-RGO [40] on GCE and RGO-IL [41] on AuE. Although it is still lower than some sensors with Hummers-type RGOs [[9], [10], [11]], our fabrication process is simple and free of metal contaminations. We also note the LOD of our sensor is relatively high among glucose biosensors. We attribute it to the large standard deviation of blank signals, which mainly relates to the test conditions (i.e. oxygen bubbles injected in the solution and the disturbance caused by stirring) and needs further optimization in future work.

4. Conclusion

This study demonstrates a clean transfer-free method to fabricate graphene-based glucose sensors by using CVD few-layer graphene film directly grown on the SiO₂/Si substrate, which was functionalized by oxygen-plasma to enable GOx immobilization and Nafion deposition. Our transfer- and metal-free process shows distinct advantage over the common metal-based CVD method in improving the electrochemical performance by eliminating the contamination of transfer-residue such as PMMA and heavy metal impurities. This graphene-based glucose biosensor shows a high sensitivity of 16.16 μA mM^-1 cm^-2 with a LOD of 124.19 μM (S/N = 3).

Acknowledgements

This work was financially supported by the Ministry of Science and Technology of China (Nos. 2016YFA0200101 and 2016YFB04001104), the National Natural Science Foundation of China (Nos. 51325205, 51290273, 51521091, 51272256, 61422406, 51802317 and 61574143), the Chinese Academy of Sciences (Nos. KGZD-EW-303-1, KGZD-EW-303-3, KGZD-EW-T06 and XDPB06), the Strategic Priority Research Program of Chinese Academy of Sciences (No. XDB30000000), the Liaoning Revitalization Talents Program (No. XLYC1808013), the Liaoning Key R&D Program, and the Program for Guangdong Introducing Innovative and Enterpreneurial Teams and the Development and Reform Commission of Shenzhen Municipality for the development of the “Low-Dimensional Materials and Devices” discipline.