Impedimetric cardiac biomarker determination in serum mediated by epoxy and hydroxyl of reduced graphene oxide on gold array microelectrodes.

A label-free chemical bonding strategy mediated by reduced graphene oxide (rGO) basal plane functional groups has been developed for cardiac Troponin I (cTnI) detection. Four different chemical strategies on respective electrode sensing surface were precedingly examined using electrochemical impedance spectroscopy. The impedimetric assessment was carried out by sweeping frequency at the range 0.1–500 kHz perturbated at a small amplitude of AC voltage (25 mV). The chemical strategy-4 denoted as S-4 shows a significant analytical performance on cTnI detection in spiked buffer and human serum, whereby the pre-mixture of rGO and (3-Aminopropyl)triethoxysilane (APTES) creates a large number of amine sites (−NH₂), which significantly enhanced the antibody immobilization without excessive functionalization. The as-fabricated immunosensor exhibited an ultra-low limit of detection of 6.3 ag mL⁻¹ and the lowest antigen concentration measured was at 10 ag mL⁻¹. The immunosensor showed a linear and wide range of cTnI detection (10 ag mL⁻¹–100 ng mL⁻¹) in human serum with a regression coefficient of 0.9716, rapid detection (5 min of binding time), and stable and highly reproducible bioelectrode response with RSD < 5%. Hence, the demonstrated S-4 strategy is highly recommended for other downstream biosensors applications.

Keywords: Graphene; Immunosensor; Impedance spectroscopy; Interdigitated electrode; Acute myocardial infarction; Cardiac troponin

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00604-021-04922-x.

Graph

Cardiovascular diseases (CVDs) are heart-related diseases that can cause sudden death without any major symptoms. According to the latest report by the World Health Organization in 2018, CVDs were ranked as the number one cause of death, accounting for 44% (17.9 million) of total 41 million death globally. It was reported 85% of the 17.9 million of CVDs deaths recorded were due to a heart attack or known as acute myocardial infarction (AMI). Cardiac troponin I (cTnI) is considered a gold standard biomarker for diagnosing AMI because of its high specificity and high sensitivity towards myocardial cell damages [[1]–[5]]. Hence, a biosensor of simple, label-free, sensitive, and fast detection in human serum is indeed essential for immediate treatments to tackle the fatal rate of AMI death.

Graphene is a 2-dimensional nanomaterial that exhibits several remarkable properties such as high surface area (2630 m2 g−1), high electron mobility (200,000 cm_SP_2_sp_ V_SP_−1_sp_ s_SP_−1_sp_), high thermal conductivity (5000 Wm−1 K−1), and Young modulus of 1.1 TPa [[6]]. Graphene derivatives such as reduced graphene oxide (rGO) have attracted much attention for biosensors development due to their versatile oxygen functional groups [[7]–[10]]. Some of the functional groups are removed during a reduction process of graphene oxide, resulting in rGO. The oxygen functional groups such as carboxyl and carbonyl are located at the edges, whereas hydroxyl and epoxy are located at the basal plane of the rGO nanosheet [[6], [11]]. The presence of functional groups in the rGO is an advantage for biological recognition elements attachment on transducer or electrode surfaces [[12]]. Furthermore, rGO is highly conductive compared to graphene oxide, compatible in micro-environment, and exhibits a very high surface area, which facilitates high-density probe immobilization [[14]–[17]].

In previous work, rGO-based cTnI immunosensors were developed by using various nanomaterials with several additional costs. The chemical bonding during the functionalization was primarily by using carboxyl or carbonyl groups of rGO. This is because chemical bonding is well established over the decades [[18]–[20]]. However, the utilization of epoxy and hydroxyl of rGO is less attracted in biosensor applications.

In this work, the use of epoxy and hydroxyl groups of rGO to fabricate cTnI immunosensors was carried out without any labeling nor nanocomposites. Four different chemical bonding strategies on respective bioelectrodes were assessed. The four bioelectrodes were fabricated using gold interdigitated microelectrodes (Au-IDE) mediated by rGO. From the four chemical bonding strategies, strategy-4 denoted as S-4 shows an excellent analytical characteristics performance for the cTnI detection in spiked buffer and human serum. The S-4 strategy was bonded based on epoxy and hydroxyl groups of rGO with facile functionalization step. The chemical bonding by using epoxy and hydroxyl of rGO paves a potential route to fabricate a highly performed sensor for the early detection of AMI.

Mouse monoclonal [4C2] cardiac troponin I (cTnI) antibody and recombinant human cardiac troponin I antigen (Abcam) were purchased from BiTA Lifescience, Malaysia (http://m.bitalifesc.com.my). Reduced graphene oxide powder was procured from Graphitene Ltd., UK (https://www.graphitene.com). Commercial gold interdigitated microelectrode (Au-IDE) with 500 fingers, having a dimension of 7.6 × 22.8 mm (width × length) with fingers' gap size of 5 μm was purchased from Metrohm Dropsens, Spain (https://www.dropsens.com). Phosphate buffer saline (PBS, pH 7.4) as a buffer agent; potassium hexacyanoferrate (III) and potassium hexacyanoferrate (II) as a redox solution to facilitate electron transfer; ethanolamine as a blocking agent to block non-specific active sites; and (3-Aminopropyl)triethoxysilane (APTES), 1-ethyl-3-dimethylaminopropylcarbodiimide hydrochloride (EDC), and N-hydroxysuccinimide (NHS) and other analytical solutions were purchased from Sigma Aldrich, USA (https://www.sigmaaldrich.com), and used as cross-linking agents.

The morphological characterizations were performed by using a high-power microscope (HPM Eclipse L300ND; Nikon Corporation, Japan), an atomic force microscope (AFM SPA400; Seiko Instruments Inc., Japan), and a scanning electron microscope (SEM JSM-6010LV; Jeol Ltd., Japan). The AFM characterization was done using a tapping mode system. Structural features of the rGO were characterized by a Raman alpha300 R spectrometer (WITec GmbH, Germany), at 532-nm laser excitation wavelength. To validate the reduction degree of the purchased rGO powder, ultraviolet-visible (UV-Vis) absorbance spectral data was collected using a Lambda 35 spectrophotometer (PerkinElmer, USA). Fourier transform infrared (FTIR) spectroscopy measurement was performed on the rGO and rGO + APTES nanohybrid to study their functional group configuration. The samples were scanned from 4000 to 650 cm−1 by a Spectrum 65 (Perkin Elmer, USA). Further analysis on the elemental composition and chemical bonding of the samples were analyzed by an X-ray photoelectron spectroscopy (XPS, Thermo Scientific, K-Alpha, UK). The XPS examination of the rGO and rGO + APTES suspensions were carried out on a silicon wafer. Electrochemical impedance spectroscopy (EIS) measurements were performed using a Novocontrol alpha high-frequency analyzer (Novocontrol Technologies GmbH, Germany). All electrochemical detection and measurements were conducted at room temperature.

The rGO suspension was prepared by referring to a recently published work by our group [[21]]. Briefly, the rGO powder was diluted in deionized water (DIW) with a concentration of 0.5 mg mL−1, followed by ultrasonication for 1 h. Then, the rGO suspension was centrifuged at 4000 rpm for 20 min. Next, the centrifuged suspension was carefully transferred to a vial without any intake of large particles using a micropipette. The collected rGO suspension was further post-sonicated for 20 min. Finally, the as-prepared rGO suspension was used to prepare rGO + APTES nanohybrid and to layer on electrode sensing surface.

A 2% of APTES was prepared in 30% of ethanol. Then, an equivalent volume ratio (1:1) of rGO and APTES was mixed, where 1 mL from the as-prepared rGO suspension and 1 mL from 2% APTES solution were mixed together. The rGO + APTES solution was incubated for 2 h at room temperature for the chemical reactions.

Commercial Au-IDE was used to develop the bioelectrodes. The fabrication of the cTnI immunosensor mediated by rGO is illustrated in Fig. 1. The four different chemical strategies on respective electrode sensing surface occur at step (ii) as shown in Fig. 1. The bare Au-IDE was cleaned prior to any modifications. The bare device was soaked in 30% of isopropyl alcohol solution and ultrasonicated for 120 s in an ultrasonic bath cleaner. Then, the device was washed with deionized water (DIW) and then dried blow. The detailed fabrication steps of the four bioelectrodes are described below.

Graph: Fig. 1 Schematic illustration of the cTnI immunosensor fabrication steps. Steps (i) to (v) were performed sequentially as elaborated in the "Fabrication of the cTnI bioelectrodes" section

The bioelectrode was fabricated by depositing the as-prepared rGO suspension onto a bare Au-IDE. Thirty microliters of rGO suspension was drop-casted on the electrode surface and placed in a convection oven at a temperature of 50 °C until the droplet completely dried. Then, 40 μL of cTnI antibody with a concentration of 10 μg mL−1 was drop-casted on the electrode surface and incubated for 2 h at 4 °C. Next, the device was washed with 10 mM of PBS solution to remove any unbound antibodies. Finally, 40 μL of 100 mM ethanolamine was drop-casted on the electrode surface to block any non-specific binding area and incubated for 1 h at 4 °C. After incubation, the device was washed with 10 mM of PBS solution to remove any unbound blocking agent. Then, the as-prepared bioelectrode was stored at 4 °C with a droplet of 10 mM of PBS solution on the electrode surface until further used.

The bioelectrode was fabricated by depositing APTES onto a bare Au-IDE and followed by rGO deposition. Fifty microliters of 2% APTES was drop-casted on the bare electrode surface and placed in a convection oven at a temperature of 50 °C until it was completely dried. Then, the device was washed with DIW to remove any unattached APTES. Next, 30 μL of rGO suspension was drop-casted on the APTES modified electrode and dried at a temperature of 50 °C. Then, the device was washed with DIW to remove any unbound rGO. Forty microliters of equivolume (v/v) mixture of 400 mM EDC and 100 mM NHS prepared in 10 mM of PBS solution was drop-casted on the rGO-modified electrode and incubated for 30 min. Afterwards, the device was washed with 10 mM of PBS solution to remove any un-reacted EDC + NHS linking agent. Then, 40 μL of cTnI antibody with a concentration of 10 μg mL−1 was drop-casted on the electrode surface and incubated for 2 h at 4 °C. Next, the device was washed with 10 mM of PBS solution to remove any unbound antibodies. Finally, 40 μL of 100 mM ethanolamine was drop-casted on the electrode surface to block any non-specific binding and incubated for 1 h at 4 °C. After incubation, the device was washed with 10 mM of PBS solution to remove any unbound blocking agent. Then, the as-prepared bioelectrode was stored at 4 °C with a droplet of 10 mM of PBS solution on the electrode surface until further used.

The bioelectrode was fabricated by depositing rGO first onto a bare Au-IDE. Thirty microliters of rGO suspension was drop-casted on the electrode surface and placed in a convection oven at a temperature of 50 °C until the droplet completely dried. Next, the rGO-modified Au-IDE was washed with DIW to remove any unattached rGO. Then, a 50 μL of 2% APTES was drop-casted on the rGO-modified electrode and placed the device in a convection oven (50 °C) until it was completely dried. Afterwards, the device was washed with DIW to remove any un-reacted APTES solution. Then, 40 μL of cTnI antibody with a concentration of 10 μg mL−1 was drop-casted on the electrode surface and incubated for 2 h at 4 °C. Next, the device was washed with 10 mM of PBS solution to remove any unbound antibodies. Finally, 40 μL of 100 mM ethanolamine was drop-casted on the electrode surface to block any non-specific binding and incubated for 1 h at 4 °C. After incubation, the device was washed with 10 mM of PBS solution to remove any unbound blocking agent. Then, the as-prepared bioelectrode was stored at 4 °C with a droplet of 10 mM of PBS solution covering the sensing electrode surface until further used.

The bioelectrode was fabricated by depositing rGO + APTES solution onto a bare Au-IDE. An equivolume (v/v) of rGO suspension (0.5 mL) and 2% of APTES (0.5 mL) was mixed and incubated for 2 h. Afterwards, 40 μL of the as-prepared rGO + APTES solution was drop-casted on the electrode surface and placed the device in a convection oven at a temperature of 50 °C until the droplet completely dried. Then, the device was washed with DIW to remove any unbound solution. Then, 40 μL of cTnI antibody with a concentration of 10 μg mL−1 was drop-casted on the electrode surface and incubated for 2 h at 4 °C. Next, the device was washed with 10 mM of PBS solution to remove any unbound antibodies. Finally, 40 μL of 100 mM ethanolamine was drop-casted on the electrode surface to block any non-specific binding and incubated for 1 h at 4 °C. After incubation, the device was washed with 10 mM of PBS solution to remove any unbound blocking agent. Then, the as-prepared bioelectrode was stored at 4 °C with a droplet of 10 mM of PBS solution covering the electrode surface until further used. The proposed label-free chemical bonding strategy of S-1, S-2, S-3, and S-4 is illustrated in Fig. 2.

Graph: Fig. 2 Schematic illustration of the chemical bonding. Chemical bonding strategy (a) S-1, (b) S-2, (c) S-3, and (d) S-4

Electrochemical impedance spectroscopy (EIS) based on Faradaic mode was performed by utilizing redox solution to measure the impedance spectra of each chemical bonding strategy. The impedance spectra were observed for the real and imaginary parts of impedance denoted as Z′ at x-axis and Z″ at y-axis, respectively, as well as simultaneous measurement on AC current variation in the system. The real part of impedance (Z′) or the semicircle diameter represents the summation of applied redox solution resistance (Rs) and the charge transfers resistance (Rct) and it can be expressed as in Eq. (1) [[22]–[24]]:

Graph

Fifty microliters of redox solution containing a mixture of equivalent volume (v/v) of 2 mM potassium hexacyanoferrate (III) and 2 mM of potassium hexacyanoferrate (II), K3[Fe (CN)6]/K4[Fe (CN)6] prepared in 10 mM PBS (pH 7.4) was used to characterize the impedance spectra. The EIS was performed by sweeping frequency at a range of 0.1–500 kHz perturbated at a small amplitude of AC voltage (25 mV).

The cTnI measurements were performed by using the as-fabricated bioelectrodes. Forty microliters of cTnI antigen concentration spiked in 10 mM of PBS (pH 7.4) was drop-casted on the bioelectrode and incubated for 5 min for the binding between the probe and target. Then, the device was washed with 10 mM of PBS (pH 7.4) to remove any unbound antigens. Afterwards, 50 μL of redox solution was drop-casted on the immunosensor and performed the impedance measurements. The cTnI measurements were performed for a series of antigen concentrations ranging from low to high concentrations.

The bioelectrode was fabricated based on the selected chemical strategy from the above experiment. Human serum was spiked in 10 mM of PBS solution with a ratio of 1:1000. Then, the as-prepared serum was used as a buffer agent to dilute cTnI antigen concentrations ranging from 10 ag mL−1–100 ng mL−1. Next, 40 μL of the as-prepared cTnI antigen concentration was drop-casted on the bioelectrode and incubated for 5 min. Then, the device was washed with 10 mM of PBS (pH 7.4) to remove any unbound antigens. Afterwards, 50 μL of redox solution was drop-casted on the immunosensor and performed the impedance measurements.

The limit of blank (LOB) represents the bioelectrode without target and limit of detection (LOD) of target. The YLOB and YLOD are calculated by using Eq. (2) and Eq. (3):

2 $$Y_{\mathrm{L}\mathrm{O}\mathrm{B}}={\mu }_{\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{b}\mathrm{e}}+1.645\left({\sigma }_{\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{b}\mathrm{e}}\right)$$

Graph

3 $$Y_{\mathrm{L}\mathrm{O}\mathrm{D}}=Y_{\mathrm{L}\mathrm{O}\mathrm{B}}+1.645\left({\sigma }_{\mathrm{l}\mathrm{o}\mathrm{w}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}\right)$$

Graph

where μprobe is the mean probe of the measured parameter, σprobe is the standard deviation of probe, and σlowest concentration is the standard deviation of the lowest measured concentration. The calculated values were used and extrapolated in the linear regression plot to estimate the LOB and LOD [[25]].

Figure 3 shows the morphological examination of electrode sensing surface, validation of rGO on the Au-IDE and rGO structural features analysis. Microscopic imaging techniques were performed to examine the electrode sensing surfaces by using HPM, SEM and AFM. UV-Vis and Raman spectroscopies were carried out to analyze the reduction degree and structural characteristics of the rGO suspension, respectively. Figure 3a shows the device was compared with the Malaysian 5 cent coin, where the size was relatively small and hand-held. Figure 3b shows HPM image of the bare Au-IDE with the electrode gap size of 5 μm, at a magnification of 100×. Figure 3c and d show the as-deposited rGO on electrode surfaces were characterized by HPM and SEM, respectively. The rGO was seen clearly between the gold electrodes as a thin film with bright color and on top of the electrode with a dark color, whereby it validates the presence of rGO when compared to the bare Au-IDE (Fig. 3b). Such a formation was seen almost all over the Au-IDE region (Fig. S1). The SEM result further evidenced the presence of rGO and it clearly revealed that the rGO exhibited a wavy-like structure covering the device surface (Fig. 3d).

Graph: Fig. 3 Morphological and structural examination. (a) Sensor size comparison between Au-IDE and Malaysian 5 cent coin. (b) HPM image of the bare Au-IDE. (c) HPM image of the as-deposited rGO nanoflakes, focus on the top of Au-IDE surface. (d) SEM image of the as-deposited rGO nanoflakes. (e) UV-Vis spectra of the as-prepared rGO suspension. (f) Raman spectra of the as-prepared rGO suspension. (g) AFM image of the rGO nanosheets

Validation on the reduction degree of the as-prepared rGO suspension (Fig. 3e, inset) is investigated by UV-Vis absorption spectra as shown in Fig. 3e. The maximum absorption peak observed at 273 nm indicates the rGO was well reduced. This is because the characteristic of the maximum absorption peak of rGO is usually higher than the GO (~230 nm) corresponding to n → π* transitions of C=O bonds [[26]]. Raman spectroscopy reveals the structural information of graphene, where it exhibits three distinctive signature peak bands. The three peak bands denoted as (D), (G), and (2D) represent the defects of carbon atoms arrangement, the sp2 arrangement and the number of layers, at wavelength of ~1350 cm−1, ~1580 cm−1, and ~2700 cm−1, respectively [[27]]. Figure 3f shows the prominent peaks of the as-prepared rGO suspension with the peak bands of D, G, and 2D were observed at the wavelength of 1368, 1598, and 2900 cm−1, respectively. Moreover, the intensity ratio of ID/IG elucidates the structural defectiveness and the restoration of sp2 domain of carbon atoms. The intensity ratio (ID/IG) of the as-prepared rGO was 0.97, which explains it exhibited a less structural defects and the sp2 domain was well restored [[26], [28]]. The broad peak of 2D indicates the rGO nanosheets have various thicknesses. This was further confirmed by AFM characterization and the thicknesses observed in the range of 2.7 to 87.6 nm, representing the production of bilayer and multilayers of rGO as shown in Fig. S2. Furthermore, a 2D planar nanosheet was noticed along with stacked, waved, and crumpled-like morphology as shown in Fig. 3g. A similar morphology feature was observed previously in the SEM result (Fig. 3d). More AFM results on the surface morphology and the thicknesses are shown in Fig. S2.

Four different chemical bonding strategies denoted as S-1, S-2, S-3, and S-4 on respective electrode sensing surfaces were developed mediated by rGO for cTnI detection, spiked in 10 mM of PBS (pH 7.4) as shown in Fig. 2. The chemical bonding strategies were developed without any labeling procedure nor functionalization with other nanomaterials. The bonding assessments and cTnI antigen detection were performed through impedance spectroscopy. The real part of impedance (Z′) or the semicircle diameter was mainly used to discuss the chemical bonding and the performance of the as-developed cTnI immunosensors. This is because the Z′ provides information on the applied redox resistance (Rs) and the charge transfers resistance (Rct) as described in Eq. (1). The amount of redox solution used in this work was fixed at 50 μL; thus, the Rs value was within the range of 40–50 Ω for every impedance measurement. The Rct value varied according to the electrode sensing surface modifications strategies, which indicates the interfacial layer bonding and charge transfer resistance.

The impedimetric curves for chemical bonding S-1, S-2, S-3, and S-4 are shown in Fig. S3a, S3b, S3c, and S3d, respectively. The real part of impedance (Z′) of S-1 (see Fig. S3a (inset)) shows a very minimum change for every modification, which indicates poor functionalization. Nearly, a similar trend of change in the Z′ is observed for S-2 (Fig. S3b), which indicates poor functionalization. However, for S-3 and S-4, a significant change in Z′ for every surface modification is observed as shown in Fig. S3c and Fig. S3d, respectively. Further detailed assessment of the chemical bonding was discussed by validating the cTnI detection performance.

The cTnI detection of S-1 shows a very slight change in Z′ for every applied target concentration and the change was within the range of 5 Ω as shown in Fig. 4a (inset (A)). The immunosensor detected the antigen at a very short range of concentrations (1 fg mL−1 to 1 pg mL−1). The small variation in Z′ and a short range of detection were due to the fact of the minimum number of antibodies attached to the rGO surface. This is because the antibody was bonded to the less composed carboxylic groups found in the rGO and the chemical bonding of S-1 is shown in Fig. 4(i). Hence, only fewer active sites exist for the direct bonding between carboxylic group of rGO and antibody, explains the poor detection performance of S-1. The detection performance of S-2 is shown in Fig. 4b. The Z′ change within a 10 Ω for every applied target concentration, ranging from 1 fg mL−1 to 10 pg mL−1. A little improvement was observed in S-2 in detecting the targets compared to S-1. This is because the carboxylic groups of rGO were further enhanced by the application of EDC-NHS activation agent [[29]] as depicted in Fig. 4(ii). Meanwhile, the detection performance of S-3 showed the Z′ impedance change within a 100 Ω for every applied target concentration, ranging from 100 ag mL−1 to 10 pg mL−1 as shown in Fig. 4c. The S-3 showed a good ability in detecting a very low target concentration, where the bioelectrode can detect a target as low as 100 ag mL−1. This is because the APTES was deposited on top of rGO. The amine and ethoxysilane sites of APTES were possibly bonded to the epoxy or hydroxyl groups of rGO as depicted in Fig. 4(iii). However, the Z′ impedance of S-4 showed a very linear range detection and the semicircle diameter was linearly increased, about ~400 Ω for every applied target concentration as shown in Fig. 4d. The linear increment of Z′ indicates the chemical bonding was well established. A further addition of antigen concentrations on the bioelectrode obstructs the charge flow, thus increases the semicircle diameter as well as the impedance. This is because the biomolecules itself behave as an insulator and increases the impedance for increased antigen concentrations. The S-4 also showed a wide range of target detection, is ranging from 1 fg mL−1 to 100 ng mL−1. The S-4 contains large number of antibodies terminated with amines sites on the electrode sensing surface, which facilitates the binding of large range of antigens concentrations. The chemical bonding of S-4 is illustrated in Fig. 4(iv). Further explanation on chemical bonding and elemental composition analysis of S-4 strategy on the pre-mixture of rGO + APTES was explained in detail in below sub-section.

Graph: Fig. 4 The detection of cTnI, spiked in 10 mM of PBS (pH 7.4). Impedimetric assessments on respective bioelectrode of (a) S-1, (b) S-2, (c) S-3, and (d) S-4. (i), (ii), (iii), and (iv) chemical bonding of the S-1, S-2, S-3, and S-4, respectively

The identification of the elemental composition and the oxygen functional groups present in the rGO and rGO + APTES at atomic level were investigated by XPS and FTIR spectra analysis (Fig. 5). Figure 5a shows the high resolution of deconvoluted Gaussian curve of C1s XPS spectrum of rGO and functional groups correspond to the carbon bond types. The binding energy of sp2 double-bond carbon atoms (C=C) was assigned at 284.3 eV, carbon atoms bonded with hydroxyl or the phenol (C-OH) group configuration were assigned at 285.5 eV, the epoxide group (>C=O) was assigned at 286.1 eV, and the carbonyl (C=O) and carboxyl groups (COOH) were assigned at 288.1 and 289.2 eV, respectively [[30]–[33]]. The results of C1s core level peak spectra clearly indicate that the rGO was dominant by epoxy and followed by hydroxyl, then carbonyl, and a very small atomic composition of carboxyl groups based on the number of elemental counts/s and peaks. The XPS scan results of rGO justify that the carboxyl groups were least present, whereas the epoxy groups were dominant. Hence, this finding explains the poor detection performance of S-1, S-2, and S-3, since the three chemical bonding strategies were based on the carboxyl groups of rGO.

Graph: Fig. 5 XPS analysis of the rGO and rGO + APTES. (a) High-resolution C1s spectra of rGO that was deconvoluted into Gaussian curves. (b) Wide energy scan region of the rGO and rGO + APTES. (c) XPS C1s spectra of the rGO and rGO + APTES. (d) FTIR spectra of the rGO and rGO + APTES

Figure 5b shows the wide energy XPS scan spectrum, which gives an immediate observation on the presence of elemental composition of the rGO and rGO + APTES. The scanned results revealed that the presence of N1s elemental composition in the rGO + APTES due to the contribution of amine sites (−NH2). The atomic percentage of carbon and oxygen in rGO + APTES solution shows an increment of 18.9% and a decrement of 34.6%, respectively, when compared to the rGO. The increment of carbon atoms in rGO + APTES indicates the restoration of carbon networks; meanwhile, the decrement of oxygen atoms explains the removal of some oxygen functional groups during the reaction between rGO and APTES [[31], [33]]. Figure 5c represents the XPS scan of rGO + APTES. The C1s core level spectrum between rGO and rGO + APTES validates the chemical bonding. The chemical bonding was further affirmed by the disappearance of epoxy and hydroxyl groups (highlighted with blue circle dot line) in the rGO + APTES profile, in comparison to the rGO. There are two possible routes of chemical bonding that have occurred between the rGO and APTES. On the one hand, the APTES amine sites have covalently bonded with epoxy groups of rGO, creating secondary amine sites. On the other hand, the hydroxyl groups of rGO have covalently bonded with the ethoxysilane groups of APTES to form a large number of amine groups [[34]]. These two routes give large number of secondary amine sites, which is desirable for antibody immobilization. Thus, this finding clearly explains the linear detection performance of the S-4 strategy. The pre-mixture of rGO + APTES generates a large site of amine-terminated groups (−NH2) that enhanced antibody immobilization. The completion of two possible chemical bonds between rGO + APTES was evident by the XPS peak shift to the left.

The presence of functional groups in the rGO and rGO + APTES was further validated by FTIR analysis. Figure 5d shows the FTIR result of the rGO compared to the rGO + APTES. A broad medium peak of epoxy groups was observed in the rGO at a wavenumber of 1220 cm−1. The peak disappeared at rGO + APTES profile, which indicates the removal of epoxy groups due to chemical bonding between rGO and APTES. The results were in line with the XPS studies. Furthermore, the presence of amine sites (−NH2) was observed in rGO + APTES at a wavelength of 1043 cm−1. This indicates the S-4 has terminated with amines sites for facile antibody immobilization. Hence, the XPS and FTIR results were in good agreement with the linear detection performance of S-4 chemical strategy. It evidently confirmed that the pre-mixture of rGO + APTES generates large amine sites (−NH2) that enhanced antibody immobilization.

A bioelectrode was developed using an S-4 chemical bonding strategy to analyze the cTnI detection in human serum. The assessment of the analytical performances of the as-developed immunosensor was performed using impedimetric platform.

The modified bioelectrode contains cTnI antibodies on electrode surface spatially to select the targeted antigen. The monoclonal antibody is capable of binding specifically to a unique epitope, located at the edge of the variable region on the antibody. Human serum was used for the selectivity tests because it contains various types of biomolecules predominantly albumin and globulin, at several folds higher in abundance compared to cTnI in the serum [[36]–[38]]. Hence, cTnI antigen was spiked in human serum from lowest to highest concentrations (linearly increased) for the selectivity assessments. Figure 6a shows the impedance spectra of the cTnI detection in serum of various antigens concentrations, varied from 10 ag mL−1 to 100 ng mL−1. Randles equivalent circuit and the corresponding impedimetric responses are shown in Fig. 6a (inset (i)), where Rs, Rct, and Cdl represent the redox solution resistance, charge transfer resistance, and double-layer capacitance in the system, respectively. The Rs value was consistent for every measured concentration due to a fixed amount of redox solution used and exhibited an average value of 40 Ω. Meanwhile, the Rct was linearly increased with the variation of ~400 Ω for the tested target concentrations. The result evidently showed that cTnI antigen was selected for the linearly increased antigens concentrations, among various types of proteins found in serum. This is due to the attached cTnI antibody on the electrode surface and well supported for the selective binding with their corresponding target. The selectivity duration between the antibody and antigen of the modified bioelectrode in serum was only 5 min. This emphasized that the modified bioelectrode shows an excellent selection and fast cTnI detection in serum.

Graph: Fig. 6 Analytical characteristics of the as-developed cTnI immunosensor in human serum. (a) Nyquist plot of the cTnI detection at various concentrations. (b) Total impedance against frequency, from surface modifications until cTnI detection at various concentrations. (c) Linear regression plot of mean charge transfer resistance (∆Rct) for the cTnI detection at a range of concentrations (10 ag mL−1 to 100 ng mL−1)

Figure 6b shows the changes of total impedance |Z| against the biased frequency, ranging from 0.1 to 500 kHz. The impedance profile begins to increase upon rGO + APTES modification until the detection of the highest level of cTnI concentrations (100 ng/mL). Clearly, the sensing was initiated at 10 kHz and achieved an equilibrium state of change transfers at 1 Hz of overall frequency responses. At the frequency range of 10 kHz–1 Hz, a significant change in total impedance |Z| profiles were observed, which indicates a sensing system [[18], [39]]. Hence, a limit of detection was analyzed at 1 Hz, where an equilibrium state of charge transfer was achieved. A linear regression plot of mean charge transfer resistance (∆Rct) against cTnI detection at various antigen concentrations and the estimation of LOD are shown in Fig. 6c. The plot shows a linear and wide range of detection (10 ag mL−1 to 100 ng mL−1) with a regression coefficient (R2) of 0.9716. YLOB and YLOD are calculated using Eq. (2) and Eq. (3). The LOB and LOD were estimated through the regression plot. The estimated LOB and LOD for the developed immunosensor were 1.3 ag mL−1 and 6.3 ag mL−1, respectively. The results showed the immunosensor was highly sensitive, beyond femto-level of cTnI detection in serum and the lowest concentration measured through EIS was 10 ag mL−1.

These results summarized the as-developed immunosensor exhibited highly sensitive, linearly selective, and fast detection (5 min for binding between the target and probe). The as-fabricated immunosensor showed a desirable analytical performance in human serum because the epoxy and hydroxyl groups of rGO at basal plane were effectively utilized to fabricate the immunosensor, developed through S-4. These findings indicate the epoxy and hydroxyl groups of rGO alone are adequate to develop a high-performed biosensor without any additional nanomaterials or excessive functionalization. A comparison of the analytical performances of label-free rGO-based cTnI biosensors reported in recent literatures is summarized and highlighted in Table 1.

Table 1 Comparison of label-free and rGO based biosensor's analytical performance for cTnI on detection

DPV, differential pulse voltammetry; CV, cyclic voltammetry; EIS, electrochemical impedance spectroscopy; GCE, glassy carbon electrode; N-prGO, N-doped porous-reduced graphene oxide; PEI, polyethyleneimine; Au-SPE, gold screen–printed electrode; ITO, indium tin oxide; WO3-rGO, tungsten trioxide–reduced graphene oxide; APTES, (3-Aminopropyl)triethoxysilane; nMo3Se4-rGO, molybdenum tetraselenide–reduced graphene oxide; BA, blocking agent; Au-IDE, gold interdigitated microelectrode; rGO, reduced graphene oxide; PDDA, poly(diallyldimethylammonium chloride); PrGO, porous-reduced graphene oxide

Although the as-fabricated cTnI immunosensor has shown high analytical characteristics performance, there are few limitations with the S-4 strategy. The application of the current S-4 strategy to other electrode sensing surfaces requires an optimization study, which is a major lacking in this work. Besides that, the rGO used should be highly dominated by epoxy and hydroxyl groups for the best performed of S-4 strategy; otherwise, poor surface modification and detection performance will be encountered. Other than this, the current surface modification strategy is more suitable for glass-based substrates and it would the best interest to test the current strategy with other semiconductor-based substrates.

The reproducibility tests of the bare and bioelectrode developed from S-4 were examined by utilizing three independent devices (n = 3). Figure 7a and b show the tests for the bare devices and the electrode surface modifications steps, respectively. Clearly, the Z′ of bare devices showed high impedance at a range of ~31–35 kΩ, which describes the absence of electrical conductivity in the devices. The reproducibility of impedance curves for n = 3 was expressed in terms of relative standard deviation (RSD) and it was 4%, which indicates the devices were reproducible. Figure 7b shows the mean impedance |Z| of electrode surface modification steps, starts from rGO + APTES deposition, antibody immobilization, and blocking agent application for the three sensors. The mean impedance was linearly increased for rGO + APTES, antibody, and blocking agent deposition with RSD of 4.8%, 4.7% and 4.8%, respectively, and obtained independently under a similar fabrication procedure. These results evidently verified that the bioelectrodes developed from S-4 were highly reproducible.

Graph: Fig. 7 Reproducibility and stability assessments. Reproducibility of (a) bare Au-IDE and (b) mean impedance of surface modification steps. (c) Stability assessment tested for 10 ng mL−1 cTnI target

The stability on measurements for the as-developed immunosensor was assessed through the mean current change in the sensor for 14 days as shown in Fig. 7c. The immunosensor was tested almost every day from day 0 when it was fabricated. The sensor was stored in 10 mM of PBS solution (pH 7.4) at RT throughout the testing period. The measurements were tested for a concentration of 10 ng mL−1 because the concentration represents clinical cardiac injury. The mean current flow in the immunosensor shows a decaying trend until day 9 with a 21.7% of reduction, which indicates the measurements were stable until day 9. However, from day 10 onwards, the mean current was increased. This is due to the fact that the protein starts to denature when exposed to the RT for a long period of time. Another possible factor that can cause the increment is the removal of bio-receptors/targets due to multiple times of rinsing, thus increasing the current conductivity.

The present work paved a potential chemical route to utilize rGO basal plane functional groups for Troponin I detection that related to acute myocardial infarction disease recognition. The as-developed S-4 strategy showed the epoxy and hydroxyl groups of rGO were well employed to anchor high density of antibodies effectively for target binding, without excessive functionalization. The as-fabricated cTnI immunosensor showed desirable analytical characteristics performance with high sensitivity, fast recognition, good stability, and reproducible bioelectrodes. However, the study on the mixing ratio between rGO and APTES of S-4 and the effect on the analytical characteristic is lacking in the present work, which serves as a limitation of this work. Nevertheless, the bioelectrode developed through the S-4 strategy can be further extrapolated for further upgraded applications of various immuno-sensing.

The author would like to acknowledge the support from the Ministry of Education Malaysia under grants FRGS/1/2017/STG05/UNIMAP/03/3 and MyPAIR/1/2020/STG05/UNIMAP//1.

The authors declare no competing interests.

Graph: (DOCX 1.78 mb)

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.