Electrospun CuO-Nanoparticles-Modified Polycaprolactone @Polypyrrole Fibers: An Application to Sensing Glucose in Saliva

A non-invasive method for detecting glucose is pursued by millions of diabetic patients to improve their personal management of blood glucose. In this work, a novel CuO nanoparticles (NPs) decorated polycaprolactone@polypyrrole fibers modified indium-tin oxide (denoted as CuO/PCL@PPy/ITO) electrode has been fabricated by electrospinning combined with the electrodeposition method for non-enzymatic detection of glucose in saliva fluid. The electrospun composite fibers exhibit high sensitivity for the glucose detection. The synergistic effect between CuO and PPy together with the unique three-dimensional net structure contributes the reliable selectivity, good test repeatability, large-scale production reproducibility in massive way, the reasonable stability and a high catalytic surface area to the sensor. Quantitative detection of glucose is determined in the linear range from 2 μM to 6 mM and the lowest detection limit is 0.8 μM. The CuO/PCL@PPy/ITO electrode shows potential for the non-invasive detection of salivary glucose.


Introduction
Diabetes is a chronic and serious disease caused by the metabolic disturbances. It is well known that blood glucose detection is a routine method to monitor the illness's condition and evaluate the therapeutic effects. However, it brings so much psychological stress and additional pain to the patient [1,2]. It is obvious that the development of responsive and non-invasive methods for detecting glucose is a need for diabetic patients, which could create the benefit of improving personal management of their blood glucose [3].
Saliva is a biological fluid with great important biomarkers for the indication of a person's health level [4][5][6]. For instance, glucose in saliva, which has good correlation with its concentrations in blood. It is believed that saliva is a promising alternative for non-invasive determination of glucose and might become a better indicator of diseases than blood, due to its easy collection by individuals and with barely any discomfort [7][8][9][10] as compared with other body fluids such as tears and sweat that serve the same purpose.
To date, there have been few articles on the detection of glucose in saliva. This is because the glucose content in saliva is much lower than the detection range of blood glucose in normal sensors.

Instrument and Equipment
The electrochemical characteristics of ITO, PCL@PPy/ITO, Cu/PCL@PPy/ITO, and CuO/PCL@PPy electrodes were examined by using cyclic voltammetry (CV), amperometric response (i-t), and electrochemical impedance spectra (EIS). Electrochemical experiments were performed with a CHI660C electrochemical workstation (Chen Hua Instruments Co., Shanghai, China). A platinum electrode and a saturated calomel electrode (SCE) were used as auxiliary and reference electrodes, respectively. The morphologies of the modified electrode surface were analyzed by field-emission scanning electron microscopy (FE-SEM, using a JSM-6700F system from JEOL, Tokyo, Japan) and transmission electron microscopy (TEM, JEOL JEM-2000 FX, Tokyo, Japan). The X-ray diffraction (XRD) patterns of CuO/PCL@PPy and PCL@PPy were recorded on a D/Max 2500V/PC X-ray diffractometer (Cu, 40 kV, 200 mA, Rigaku, Tokyo, Japan), at a scan rate of 0.05 • s −1 with 2 ranging from 10 • to 80 • .

Preparation of Electrospinning PCL Nanofibers
The PCL nanofibers were prepared by a method similar to previous report [33]. Briefly, the precursor solution consisted of 0.7 g PCL granules, 3 mL HFIP, 1 mL acetic acid (20%) and 1 mL formic acid (60%). The precursor solution was then stirred for 12 h at the room temperature until entirely dissolved. The voltage difference between the stainless steel nozzle (positive) (0.5 mm inner diameter) and the sample collector (negative) (ITO) was about 18 kV. The distance between the stainless steel nozzle and the ITO was about 15 cm. The flow rate of 1 mL/h was applied during the electrospinning.

Preparation of CuO/PCL@PPy/ITO
The ITO with PCL nanofibers was firstly dipped into a 0.025 mM (15 mL) pyrrole solution containing the solution of 0.04 M FeCl 3 for 1.5 h until the color of the nanofibers changed from a white color to a light black color. The as-formed PCL@PPy nanofibers were then washed with DI water and dried in the air. The electrochemical deposition of copper onto the substrate of ITO with PCL@PPy nanofibers was then performed via multi-potential steps scan at ambient temperature. Different applied potentials (−0.8, −1.0, 1.2, and −1.4 V versus SCE), different deposition laps (1000, 1200, 1400, and 1600) and various concentrations of copper sulfate (5,10,15, and 20 mM) have been investigated for obtaining the optimal electrodeposition condition. The Cyclic Voltammogram (CV) scanning was applied to oxidize the Cu nanoparticles in the solution of 0.1 M NaOH and the potential was scanned between −1 V and 1 V for 10 cycles at the scan rate of 50 mV/s. Scheme 1 describes the above fabrication process.

Instrument and Equipment
The electrochemical characteristics of ITO, PCL@PPy/ITO, Cu/PCL@PPy/ITO, and CuO/PCL@PPy electrodes were examined by using cyclic voltammetry (CV), amperometric response (i-t), and electrochemical impedance spectra (EIS). Electrochemical experiments were performed with a CHI660C electrochemical workstation (Chen Hua Instruments Co., Shanghai, China). A platinum electrode and a saturated calomel electrode (SCE) were used as auxiliary and reference electrodes, respectively. The morphologies of the modified electrode surface were analyzed by field-emission scanning electron microscopy (FE-SEM, using a JSM-6700F system from JEOL, Tokyo, Japan) and transmission electron microscopy (TEM, JEOL JEM-2000 FX, Tokyo, Japan). The X-ray diffraction (XRD) patterns of CuO/PCL@PPy and PCL@PPy were recorded on a D/Max 2500V/PC X-ray diffractometer (Cu, 40 kV, 200 mA, Rigaku, Tokyo, Japan), at a scan rate of 0.05° s −1 with 2 ranging from 10° to 80°.

Preparation of Electrospinning PCL Nanofibers
The PCL nanofibers were prepared by a method similar to previous report [33]. Briefly, the precursor solution consisted of 0.7 g PCL granules, 3 mL HFIP, 1 mL acetic acid (20%) and 1 mL formic acid (60%). The precursor solution was then stirred for 12 h at the room temperature until entirely dissolved. The voltage difference between the stainless steel nozzle (positive) (0.5 mm inner diameter) and the sample collector (negative) (ITO) was about 18 kV. The distance between the stainless steel nozzle and the ITO was about 15 cm. The flow rate of 1 mL/h was applied during the electrospinning.

Preparation of CuO/PCL@PPy/ITO
The ITO with PCL nanofibers was firstly dipped into a 0.025 mM (15 mL) pyrrole solution containing the solution of 0.04 M FeCl3 for 1.5 h until the color of the nanofibers changed from a white color to a light black color. The as-formed PCL@PPy nanofibers were then washed with DI water and dried in the air. The electrochemical deposition of copper onto the substrate of ITO with PCL@PPy nanofibers was then performed via multi-potential steps scan at ambient temperature. Different applied potentials (−0.8, −1.0, 1.2, and −1.4 V versus SCE), different deposition laps (1000, 1200, 1400, and 1600) and various concentrations of copper sulfate (5, 10, 15, and 20 mM) have been investigated for obtaining the optimal electrodeposition condition. The Cyclic Voltammogram (CV) scanning was applied to oxidize the Cu nanoparticles in the solution of 0.1 M NaOH and the potential was scanned between −1 V and 1 V for 10 cycles at the scan rate of 50 mV/s. Scheme 1 describes the above fabrication process.

Characterization of as-Prepared Electrodes
The surface morphology of PPy/PCL and CuO/PPy/PCl were characterized by FE-SEM. As shown in Figure 1a, the PCL nanofibers with diameters varying from 150 to 250 nm are observed to be relatively smooth with no residue attached, which is consistent to with previous work [34]. Figure 1b shows the PPy/PCL nanofibers becoming obviously rough. The modification of PPy led to the formation of core-shell PPy/PCL and the white color of PCL nanofibers turned into the black color of PPy/PCL nanofibers. Figure 1c shows the morphology of CuO/PCL@PPy nanofibers. The CuO NPs are homogeneously and uniformly distributed at the surface of PCL@PPy nanofibers. The average size of the CuO NPs is approximately 100 nm, which may bring the benefits of large surface area, good catalytic activity, and large number of active sites. The TEM image of the CuO/PCL@PPy nanofibers further confirms the well-dispersed state of CuO NPs on the surfaces of the fibers. It also confirms the core-shell structure of the PCL@PPy nanofibers in Figure 1d.

Characterization of as-Prepared Electrodes
The surface morphology of PPy/PCL and CuO/PPy/PCl were characterized by FE-SEM. As shown in Figure 1a, the PCL nanofibers with diameters varying from 150 to 250 nm are observed to be relatively smooth with no residue attached, which is consistent to with previous work [34]. Figure  1b shows the PPy/PCL nanofibers becoming obviously rough. The modification of PPy led to the formation of core-shell PPy/PCL and the white color of PCL nanofibers turned into the black color of PPy/PCL nanofibers. Figure 1c shows the morphology of CuO/PCL@PPy nanofibers. The CuO NPs are homogeneously and uniformly distributed at the surface of PCL@PPy nanofibers. The average size of the CuO NPs is approximately 100 nm, which may bring the benefits of large surface area, good catalytic activity, and large number of active sites. The TEM image of the CuO/PCL@PPy nanofibers further confirms the well-dispersed state of CuO NPs on the surfaces of the fibers. It also confirms the core-shell structure of the PCL@PPy nanofibers in Figure 1d.   Figure 2, which might be due to a low crystallinity the CuO NPs.   Figure 2, which might be due to a low crystallinity the CuO NPs. In order to clarify the formula of copper oxide, the CuO/PPy/PCL nanofilm was further characterized by XPS. Figure 3A shows the XPS survey spectrum of the CuO/PCL@PPy substrate, which depicts a binding energy peak at 934.4 eV corresponding to the Cu 2p region. In Figure 3B, the strong satellite peaks at 935 and 955 eV confirm the presence of Cu(II) species. The absence of the satellite peak at 939−945 eV suggests the lack of Cu(I). The format of copper within the nanofibers thus includes both Cu(II) oxide and pure Cu(0).  Figure 4 presents the electrochemical behaviors of bare ITO, PCL@PPy/ITO, Cu/PCL@PPy/ITO and CuO/PCL@PPy/ITO in a solution of 5 mM Fe(CN)6 3− by using cyclic voltammetry. The anodic peak current of ITO is about 71.51 A, and peak currents of PCL@PPy/ITO and the CuO/PCL@PPy/ITO are 455.6 A and 512.6 A, respectively. Obviously, the large surface area of CuO/PCL@PPy/ITO contributed to the larger electrochemical response. The difference of potentials (Ep = Epa − Epc, where Epa and Epc are the anodic and cathodic peak potentials, respectively) was measured to be 72 mV for ITO, 155 mV for CuO/PCL@PPy/ITO, 266 mV for Cu/PCL@PPy/ITO, and 167 mV for PCL@PPy/ITO. The modification reduced the conductivity and the reversibility of electron transfer as comparison with ITO. It is interesting to note that CuO/PCL@PPy/ITO shows relatively better reversibility than Cu/PCL@PPy/ITO. In order to clarify the formula of copper oxide, the CuO/PPy/PCL nanofilm was further characterized by XPS. Figure 3A shows the XPS survey spectrum of the CuO/PCL@PPy substrate, which depicts a binding energy peak at 934.4 eV corresponding to the Cu 2p region. In Figure 3B, the strong satellite peaks at 935 and 955 eV confirm the presence of Cu(II) species. The absence of the satellite peak at 939−945 eV suggests the lack of Cu(I). The format of copper within the nanofibers thus includes both Cu(II) oxide and pure Cu(0). In order to clarify the formula of copper oxide, the CuO/PPy/PCL nanofilm was further characterized by XPS. Figure 3A shows the XPS survey spectrum of the CuO/PCL@PPy substrate, which depicts a binding energy peak at 934.4 eV corresponding to the Cu 2p region. In Figure 3B, the strong satellite peaks at 935 and 955 eV confirm the presence of Cu(II) species. The absence of the satellite peak at 939−945 eV suggests the lack of Cu(I). The format of copper within the nanofibers thus includes both Cu(II) oxide and pure Cu(0).     6 3− by using cyclic voltammetry.  The Anson equation was quoted to estimate the effective surface area for electrochemical reaction.

Electrochemical Response of Glucose
where n is the number of electron transfers, F is Faraday's constant, A is the effective surface area of the working electrode, c is the concentration of the reactant, Qdl is the double layer charge which can be eliminated by background subtraction, and Qads is the Faradaic charge. Figure 4C shows  The Anson equation was quoted to estimate the effective surface area for electrochemical reaction.
where n is the number of electron transfers, F is Faraday's constant, A is the effective surface area of the working electrode, c is the concentration of the reactant, Q dl is the double layer charge which can be eliminated by background subtraction, and Q ads is the Faradaic charge. Figure 4C shows  Figure 4B, the bare ITO (curve a) and the PCL@PPy-modified ITO (curve b) electrodes show no responses to glucose oxidation. As compared to the CuO/PCL@PPy/ITO electrode without glucose (curve e), the CuO/PCL@PPy/ITO electrode with glucose (curve c) demonstrates a significant current change starting at about +0.50 V and has a shoulder oxidation hump at +0.70 V in the anodic scan. Obviously, the current responded to the addition of glucose. The glucose oxidation at CuO/PCL@PPy/ITO was an irreversible process, suggesting that the CuNPs played a key role in the electrochemical catalytic oxidation.
In the electrode system of CuO/PCL@PPy/ITO, the glucose can be directly oxidized into gluconolactone under the alkaline condition. However, the mechanism of the glucose oxidation at the copper based system remains unclear and might be dependent on the transformation of the Cu redox couples from Cu(II) to Cu(III) [36]. The possible glucose oxidation process at CuO is depicted as follows [37]: An EIS experiment has been performed to study the performance of the interface property of the electrode surface. In the presence of the redox probe [Fe(CN) 6 ] 4−/3− , the Nyquist plots of the bare ITO electrode, the PCL@PPy/ITO electrode and the CuO/PCL@PPy/ITO electrode were recorded. In the inset of Figure 4D, by using ZinView software (Scribner Associates, Southern Pines, NC, USA) to fit the impedance spectra ( Figure 4D), the equivalent electrical circuits are illustrated. The diameter of a semicircle portion is equal to the electron transfer resistance (R ct ), indicating in the electron-transfer kinetics of the redox reaction at the electrode surface and indirectly reflecting conductivity. The PCL@PPy/ITO electrode exhibits minimum R ct and hence the conductivity of PCL@PPy/ITO is the best one.among the electrode types. Clearly, covering electrospun nanofibers covered with PPy improved the conductivity properties with theas a significant increase on thein surface area. Modification was observed. The modification of PCL@PPy/ITO with Cu species containing CuO onto the PCL@PPy/ITO increased R ct .

Optimization of Preparation Conditions for CuO/PCL@PPy/ITO Electrodes
The fabrication process of CuO NPs deposition onto the PCL@PPy nanofiber layer has been optimized. The CuO NPs were deposited onto PCL@PPy nanofibers by the electroplating method via multi-potential steps. The concentration of CuSO 4 , the deposition potential, and the electrochemical deposition laps affect the final products of CuO/PCL@PPy, which were investigated by means of i-t measurements in 0.01 M NaOH solution containing 0.01 mM glucose. As seen in Figure 5, the optimal CuO/PCL@PPy with the highest catalysis can be reached when the deposition parameters were set at the usage of 0.01 M CuSO 4 solution, an applied potential of −1 V and the deposition lap number of 1200 cycles.
Potential-dependent i-t responses were recorded by continuously adding 1 M glucose (1 µL) into 0.01 M NaOH solution under the stirring condition. In Figure 6, the working potential set at 0.7 V shows the relatively better sensitivity than the others, which is selected to conduct the following amperometric determination of glucose.  Potential-dependent i-t responses were recorded by continuously adding 1 M glucose (1 μL) into 0.01 M NaOH solution under the stirring condition. In Figure 6, the working potential set at 0.7 V shows the relatively better sensitivity than the others, which is selected to conduct the following amperometric determination of glucose.   Potential-dependent i-t responses were recorded by continuously adding 1 M glucose (1 μL) into 0.01 M NaOH solution under the stirring condition. In Figure 6, the working potential set at 0.7 V shows the relatively better sensitivity than the others, which is selected to conduct the following amperometric determination of glucose.   Figure 7 exhibits the i-t curve for the detection of glucose by using the CuO/PCL@PPy nanofibers electrode in a stirred solution of 0.01 M NaOH with an optimal potential at +0.70 V. There are two linear relationships between the concentrations of glucose and current responses, ranging from 2 µM to 1 mM and from 1 mM to 6 mM, respectively. It meets the requirement of the determination of glucose in human saliva because of its glucose concentration varying in a range from 2 µM to 6 mM [5,7,8]. Figure 7 exhibits the i-t curve for the detection of glucose by using the CuO/PCL@PPy nanofibers electrode in a stirred solution of 0.01 M NaOH with an optimal potential at +0.70 V. There are two linear relationships between the concentrations of glucose and current responses, ranging from 2 μM to 1 mM and from 1 mM to 6 mM, respectively. It meets the requirement of the determination of glucose in human saliva because of its glucose concentration varying in a range from 2 μM to 6 mM [5,7,8].

Selectivity and Stability Studies
In order to examine the selectivity for the as-made glucose biosensor, the i-t curves have been taken at 0.7 V in 0.01 M NaOH solution containing 0.1 mM glucose by successively adding the possible interfering species involving 0.01 mM AA, DA, and UA. As seen in Figure 8A, for the CuO/PCL@PPy/ITO electrode, no obvious interference is from the above-observed species, suggesting the good selectivity of the biosensor. The long-term stability of the CuO/PCL@PPy/ITO electrode was also explored by recording the i-t response toward 0.1 mM glucose over 10 days. As displayed in Figure 8B, the amperometric response can retain approximately 88% of the level of the original value after 25-days of monitoring and storage in ambient conditions. It suggests the reasonable stability of the CuO/PCL@PPy/ITO-based sensor.

Selectivity and Stability Studies
In order to examine the selectivity for the as-made glucose biosensor, the i-t curves have been taken at 0.7 V in 0.01 M NaOH solution containing 0.1 mM glucose by successively adding the possible interfering species involving 0.01 mM AA, DA, and UA. As seen in Figure 8A, for the CuO/PCL@PPy/ITO electrode, no obvious interference is from the above-observed species, suggesting the good selectivity of the biosensor. Figure 7 exhibits the i-t curve for the detection of glucose by using the CuO/PCL@PPy nanofibers electrode in a stirred solution of 0.01 M NaOH with an optimal potential at +0.70 V. There are two linear relationships between the concentrations of glucose and current responses, ranging from 2 μM to 1 mM and from 1 mM to 6 mM, respectively. It meets the requirement of the determination of glucose in human saliva because of its glucose concentration varying in a range from 2 μM to 6 mM [5,7,8].

Selectivity and Stability Studies
In order to examine the selectivity for the as-made glucose biosensor, the i-t curves have been taken at 0.7 V in 0.01 M NaOH solution containing 0.1 mM glucose by successively adding the possible interfering species involving 0.01 mM AA, DA, and UA. As seen in Figure 8A, for the CuO/PCL@PPy/ITO electrode, no obvious interference is from the above-observed species, suggesting the good selectivity of the biosensor. The long-term stability of the CuO/PCL@PPy/ITO electrode was also explored by recording the i-t response toward 0.1 mM glucose over 10 days. As displayed in Figure 8B, the amperometric response can retain approximately 88% of the level of the original value after 25-days of monitoring and storage in ambient conditions. It suggests the reasonable stability of the CuO/PCL@PPy/ITO-based sensor. The long-term stability of the CuO/PCL@PPy/ITO electrode was also explored by recording the i-t response toward 0.1 mM glucose over 10 days. As displayed in Figure 8B, the amperometric response can retain approximately 88% of the level of the original value after 25-days of monitoring and storage in ambient conditions. It suggests the reasonable stability of the CuO/PCL@PPy/ITO-based sensor.

Detection of Glucose in Real Samples
As described above, the physiological level of glucose in saliva is in the range of 0.5-20 mg/dL (27.8 µM-1.11 mM). In this work, the saliva sample (1 mL) was centrifuged and the upper side of the liquid was diluted with NaOH solution (0.01 M) up to 10 mL. The spiked samples were obtained by adding the due concentrations of glucose into the treated samples. The results for the detection of glucose in human saliva by using the CuO/PCL@PPy/ITO electrode are summarized in Table 1.
The recoveries are in the range from 96.36% to 105.6%. As compared with the other glucose sensors in the literature, as shown in Table 2, the as-prepared CuO/PCL@PPy/ITO electrode demonstrates excellent performance with a wider dynamic concentration range and sufficient sensitivity. The result suggests that the CuO/PCL@PPy/ITO electrode can be developed to be a promising candidate for sensing low concentrations of glucose in samples such as salivary glucose.

Conclusions
We have successfully constructed a CuO/PCL@PPy/ITO electrode by the electrospinning technique together with the in-situ electrochemical deposition. The resultant CuO/PCL@PPy/ITO with a reasonable stability contributed the excellent electrocatalytic activity for the detection of low concentrations of glucose. The CuO/PCL@PPy/ITO-based glucose sensors can be fabricated in the massive way. The sensor has been successfully utilized to determine the concentration of glucose in human saliva. It is expected that the mass-produced CuO/PCL@PPy/ITO-based glucose sensors made massively can become a non-invasive assay for the detection of salivary glucose at home.