Enhanced Optical Sensitivity of Polyvinyl Alcohol – Reduced Graphene Oxide Electrospun Nano ﬁ ber Coated Etched Fiber Bragg Grating Sensor for Detection of Myoglobin a Cardiac Biomarker

label-free, multianalyte, portable, real-time, point-of-care (POC) kits at ambient conditions.

Etched fiber Bragg grating (eFBG) sensors have shown to be highly sensitive with the capability of giving accurate real-time response to a variety of measurands such as pressure, gas, biomolecules, to name a few. These sensors have not shown their mettle as competitive products mainly due to nonreproducibility in results and inefficiency in upscaling for large-scale production; the main reason being nonuniform and complicated coating procedures. Herein, the enhancement in refractive index (RI) sensitivity (%4 times) obtained with electrospinning of polyvinyl alcohol-reduced graphene oxide (PVA-rGO) nanofibers onto eFBG sensor using a customized target and a unique sandwich arrangement is demonstrated. The enhancement in RI sensitivity has led to a lower detection limit and increased sensitivity and linear range for a case study using myoglobin (Mb), an early-stage cardiac biomarker with high reproducible results (standard error ≤AE2.3%). rGO embedded PVA nanofiber electrospun onto an eFBG sensor (PVA-rGO sensor) is the first of its kind and has significant importance in developing cost-effective, label-free, multianalyte, portable, real-time, point-ofcare (POC) kits at ambient conditions. chemiresistive [10,11] sensors with increased performance. None of these methods have yet been commercialized, with some methods requiring sophisticated instruments with dedicated personnel to perform tests and thus becoming expensive while others being nonscalable for mass production. A comparison of their performance is given in Table T1, Supporting Information.
After Iadicicco et al. demonstrated the dependence of refractive index (RI) sensitivity on cladding diameter, [12] the applications of etched fiber Bragg grating (eFBG) sensors as RI sensor, pressure sensor, fuel-level sensor, temperature sensor, gas sensor, and biosensor [13][14][15][16][17][18][19][20] are well documented. Even though these sensors have been a research topic for more than 10 years, translating from an idea to a product has not happened due to nonreproducible results, primarily from the coating's nonuniformity due to their microscale diameter. The methods used to improve eFBG sensors' repeatability have made them an expensive proposition compared with other commercially available sensors. The proposed polyvinyl alcohol-reduced graphene oxide (PVA-rGO) electrospun coating on the eFBG sensor is a cost-effective solution to obtain highly reproducible results at NTP conditions.
The recent applications of functionalized nanocomposites (nanoparticles, polymers) in biological applications have shown promising results. [21][22][23][24][25][26][27] In the present work, a PVA-rGO composite has been electrospun on an eFBG sensor. PVA is a synthetic nonconducting, biocompatible, and biodegradable polymer with hydroxyl groups on its side chains, making them highly hydrophilic. [28] rGO-embedded PVA nanocomposite film has been reported to enhance mechanical property, [29] electrical conductivity, [30] as well as alter the optical conductivity and band structure parameters of PVA, [31] having applications in different sectors. [32][33][34] The electrospinning method is a cost-effective method to generate nanofibers of required diameter using the electrostatic field principle with simple instrumentation [35] and has found application in different sectors. [36,37] PVA with its unique fiber/film-forming capacity, optical property, immense dielectric strength, [38] and low manufacturing cost can be easily coated using the electrospinning method. [39][40][41] Figure 1 shows the proposed scheme to get a uniform coating on the eFBG sensor and an application to study this scheme's efficacy. Here, the eFBG sensor ( Figure 1a) is placed on the customized target (Figure 1b). The shape (cylindrical) and fragility of the eFBG sensor (9 to 13 μm) have made traditional targets (flat sheets or rotating drums), which are typically used in electrospinning, nonviable to obtain a uniform coating. The in situ prepared PVA-rGO solution (Figure 1c) is loaded into the syringe placed in a syringe pump (Figure 1d), and the nanofibers are electrospun onto the sensor using a sandwich protocol. The nanofiber-coated eFBG sensor is removed from the target (Figure 1e) and exposed to glutaraldehyde vapors followed by immersion in tris buffer. The sensor is then incubated in myoglobin specific (anti-Mb) single-strand DNA (DNA/aptamer) for 50 min. As eFBG sensors have constrained area, there is a need to have a sufficient number of capture molecules per unit area, making compact bioreceptors a necessity, and aptamers are one such biomolecule. They are nucleic acids with a minute physical profile, which help achieve high capture molecule concentrations on the sensors. The DNA will be immobilized on the sensor surface at the end of incubation. This step is followed by incubation in Mb. The hybridization of Mb with DNA is shown in Figure 1f. The Bragg wavelength, λ B of the eFBG sensor, is measured using the experimental setup shown in Figure 2.

Fabrication of rGO, PVA, and PVA-rGO eFBG Sensors
The phase mask technique has been adopted for the inscription of gratings in the core of a single-mode (germania doped) optical fiber. The gratings allow the optic fiber to behave as a wavelengthselective filter by reflecting one particular wavelength (Bragg/ resonant wavelength) λ B while transmitting the others, on incidence with a broadband light source; the Bragg wavelength is defined by the equation λ B ¼ 2Λn eff where Λ is the pitch of grating and n eff is the effective RI of LP 01 mode. [42] The fiber has a core diameter of 9 μm and a clad of 125 μm. Removal of the clad partially by chemical etch permits the evanescent waves from the core-clad interface to interact with surrounding media, governed by the equation Δλ B ¼ 2Λη p0 (n sur À n cl ) where, η p0 is the fraction of the total power (unperturbed mode) that flows in the clad, n sur is the RI of surrounding medium, and n cl is the RI of cladding. [43] Three eFBG sensors (11 AE 0.5 μm) are fabricated in the present work. One sensor is dropped with NH 4 OH: www.advancedsciencenews.com www.adpr-journal.com [1:1:5], followed by rGO dip (allowed to dry %4 h at room temperature) and finally washed in deionized (DI) water. [19] The resultant sensor is referred to as the rGO sensor. The second and third sensors are coated with PVA and PVA-rGO nanofibers, respectively, by electrospinning. The nanofiber coating is exposed to glutaraldehyde (a bi aldehyde) vapor, which acts as a crosslinker between individual nanofibers (aldehyde-hydroxyl bond) and reduces porosity. These are referred to as PVA and PVA-rGO sensors. The PVA-rGO coated sensor's surface morphology is shown in the scanning electron microscope (SEM) images ( Figure 3). The nanofiber coating on the eFBG sensor is shown in Figure 3a. The nanofiber coating post-treatment to glutaraldehyde ( Figure 3b) displays a reasonably uniform thickness (%1.5 μm) along the sensor region.
The sensors are subsequently immersed in tris buffer (prepared in 10 mM tris HCl, 150 mM NaCl, 5 mM MgCl 2 , pH %8) and then removed. The Bragg reading is referred to as "before DNA incubation" (λ B, before ). The buffer salts neutralize the negative charge repulsion between rGO and DNA phosphate backbone. [44] PVA being hydrophilic, swells, and disintegrates, and becomes transparent. The PVA-rGO sensor form islands in the presence of rGO (white spots) (Figure 3c).

Optimization of rGO and DNA Concentration for PVA-rGO Sensors
The sensor (described in Section 2) is incubated with 85 μL of anti-Mb DNA (suspended in 1Â phosphate-buffered saline www.advancedsciencenews.com www.adpr-journal.com [PBS] buffer, pH % 7.4) for 50 min ( Figure S1, Supporting Information) followed by buffer wash to flush out the unbound DNA. The DNA bases bond to aromatic regions of rGO by π-π stacking. [45] At the end of incubation, λ B is measured (λ B, DNA ), and the Bragg shift (Δλ B ¼ λ B, before À λ B, DNA ) is calculated. Δλ B is used to measure DNA immobilization on the sensor. Higher values of Δλ B implies better DNA immobilization. The maximum immobilization of DNA on the sensor is dependent on rGO (wt%) and DNA concentrations (ng μL À1 ) and is shown in Table 1. Maximum DNA immobilization is equated with maximum Bragg shift which occurs (Δλ B|max ¼ 801 pm) for the combination of 0.01 wt% (rGO) and 12 ng μL À1 (DNA). The eFBG sensor is temperature-sensitive with the temperature effect inherent in the measured λ B , thus making its measurement necessary. The temperature response at experimental conditions is determined by multiplexing a second PVA-rGO coated sensor (sensor 2, Figure 2) to sensor 1 (primary sensor), but without further incubations/immersions as sensor 1, and measuring λ B .

Characterization
Electron dispersive spectroscopy (EDS) analysis is used to determine the elemental composition of PVA-rGO sensors. The atomic % of carbon (C), oxygen (O), and silicon (Si) on the white spots (of Figure 3c) are 26, 56, and 17, respectively, whereas that of the black background are 34, 43, and 21, respectively. There is a decrease in the C content and an increase in the white spots' O content. The actual reason for this needs an understanding of the chemical structure.
Raman spectroscopy has been conducted ( Figure 3d) for rGO, PVA, and PVA-rGO sensors. PVA has an intense peak at 2912 cm À1 , ascribed to─CH 2 stretching vibration. [46] rGO has characteristic peaks at 1358 cm À1 (D) and 1594 cm À1 (G) attributed to induced Raman mode and symmetry allowed tangential mode, respectively. Both these peaks are seen in the PVA-rGO sensor. The increase in I D /I G intensity ratio from 1.1 (rGO) to 1.2 (PVA-rGO) can be explained as the generation of sp 2 carbon domains with a smaller average size.
Fourier transform infrared (FTIR) spectroscopy (Figure 3e) has been performed to study the chemical bonds of PVA-rGO, PVA, and rGO. Both PVA and rGO display a characteristic peak at 1723 cm À1 (C¼O), an indicator of residual oxygencontaining groups. The band between 2840 to 3000 cm À1 (C─H) appears due to glutaraldehyde. The broadband between 3550 to 3200 cm À1 (O─H) in PVA and PVA-rGO is associated with the existence of strong inter and intramolecular hydrogen bonds. [47] The reduction in this peak intensity from PVA to PVA-rGO is attributed to interfacial interaction between PVA and rGO. Similarly, the decrease in intensity in 1150 to 1085 cm À1 (C─O) stretching mode is related to removing oxygencontaining functional groups and forming strong hydrogen bonds between PVA and rGO.
Raman and FTIR results indicate that embedding PVA with rGO and post-treatment with glutaraldehyde results in strong interfacial interaction between PVA and rGO.

Case Study: Myoglobin Sensing
The measured λ B up to DNA immobilization is shown in Figure 4a for the PVA-rGO sensor. All the measurements are made at NTP conditions. Each of the DNA immobilized sensor (rGO, PVA, and PVA-rGO) is initially drop cast with fixed volume (85 μL) of PBS buffer without Mb (0 g mL À1 ) accompanied with Bragg wavelength measurement (λ B , Ref ). This step is followed by incubation in the lowest concentration of Mb aliquot (prepared in PBS buffer) for a total of 10 min (this ensures that the measured λ B curve becomes reasonably flat). These steps are repeated for all the Mb concentrations (10 ag mL À1 up to 100 μg mL À1 ). After a dip in each concentration, the sensor is washed in PBS buffer to flush out the unbound molecules before adding the next higher concentration. Figure 4b shows the plot of λ B versus time for the PVA-rGO sensor. The limit of detection (LOD) of the sensor depends on the signal to noise (S/N) ratio at experimental conditions. The S/N is determined by the interrogator's wavelength repeatability (1 pm) and environmental noise. As the experiments are conducted at NTP, the standard deviation (σ ¼ 5.7 pm) of the sensor signal (0 g mL À1 , Figure 4b) is used for the determination of noise margin ( Figure S2, Supporting Information). Chiavaioli et al. [48] have stated that a sensor's LOD can be considered as 3σ (17.1 pm), which is 100 ag mL À1 (Δλ B ¼ 64 pm) and is well within the clinical detection limit of Mb (90 ng mL À1 ). In the event of MI, Mb (in human blood) increases above the normal value (0-90 ng mL À1 ) within the first 2 h, reaches a peak (%600 ng mL À1 ) between 8 and 12 h, and returns to normal within 24 h. The sensor response is seen to lie well within the required clinical range.
The differences between λ B and λ B , Ref (Δλ B ) for all the Mb concentrations are calculated. The Mb linear plots for PVA-rGO (1 fg mL À1 up to 1 μg mL À1 ) and rGO sensor (1 fg mL À1 up to 100 pg mL À1 ) are shown in Figure 4c with sensitivities of 286 pm 10 À1 g À1 mL and À12 pm 10 À1 g À1 mL, respectively. The PVA sensor did not show any measurable response. Furthermore, an increase in Mb concentration causes a decrease in λ B for the rGO sensor, whereas it increases for the PVA-rGO sensor. The reason for this can be explained as follows: the DNA immobilized on the sensor surface has a net negative charge with a fixed electron (e À ) transfer rate between DNA and buffer. The hybridization of Mb (heme group þ globin protein) with DNA is accomplished by reduction of iron (center atom of heme group) from ferric to ferrous (Fe 3þ þ e À ↔ Fe 2þ ) with a net decrease in e À transfer rate. The reduction will cause a decrease in the bulk dielectric constant (ε) with a consequent decrease in RI (n) of the Mb solution (n α ε 1/2 ). The higher Mb concentration will increase hybridization with a further decrease in RI of the solution and lower values of λ B for rGO sensor and higher values of λ B for PVA-rGO sensors.
The higher value of λ B for lower RI indicates a PVA-rGO sensor with a negative RI sensitivity. The RI experiment, which includes the sensor to be dipped in each concentration (85 μL) of the prepared sucrose solution (Section 7.1) for 10 min (accompanied by λ B measurement) followed by DI wash, confirms this hypothesis. Figure 4d shows the λ B plots for sucrose solutions of different RI for both rGO and PVA-rGO sensors. A linear regression fit for the plots gives a downshift of 1.45 nm for every one unit decrease in RI for rGO sensor and an upshift of 6.9 nm in λ B for every one unit decrease in RI for PVA-rGO sensor.
The reproducibility in the sensor performance is established by repeating the Mb experiment on three similarly prepared DNA immobilized PVA-rGO coated sensors (Sections 2 and 4.2). The mean and σ at each concentration are calculated. The σ is shown as the error bar (Figure 4c) and gives a standard error |max ≤AE2.3% (10 pg mL À1 ).
The sensor response to temperature changes at experimental conditions needs to be determined, as stated earlier. The temperature sensor has a standard deviation of 1.73 pm ( Figure S3, Supporting Information), which lies well within the sensor's noise margin (17 pm) and makes Mb sensor response (Figure 4b) independent of temperature effects.

Enhancement of PVA-rGO Sensor Response
The PVA-rGO nanofiber sensor has increased by %4 times in the RI sensitivity than the rGO dip sensor. This increase has led to an increase in the LOD (%90%) (LOD of rGO sensor is 1 fg mL À1 ) and a phenomenal increase in Mb sensitivity (%24 times) and linear range (%10 K times). The explanation for this enhancement can be attributed to two factors: one is the strong interfacial bond between PVA and rGO (as determined by the FTIR and Raman results), and the second is the uniformity of the PVA-rGO nanofiber coating (as determined by SEM results). The electrospinning of nanofibers (PVA-rGO) using the customized www.advancedsciencenews.com www.adpr-journal.com target and sandwich arrangement provides the initial uniformity in a coat over the eFBG sensor. Postprocessing of the nanofibers with glutaraldehyde makes the PVA matrix a strong adhesive for rGO entrapment on the sensor surface. The rGO is responsible for DNA-Mb hybridization. The thickness of the PVA-rGO nanofiber coating can be precisely controlled (by fine control of the electrospinning parameters) to obtain reproducible sensors. The PVA-rGO sensor performance for Mb sensing is better than the other techniques listed to date (Table T1, Supporting  Information), showing the least LOD and highest dynamic range.

Cross sensitivity and Stability of PVA-rGO Sensor
The cross-sensitivity of the PVA-rGO sensor has been determined by conducting experiments similar to Mb sensing (Section 4.2) with aliquots (10 pg mL À1 to 100 μg mL À1 ) of other proteins commonly present in human plasma (hemoglobin and hematin) and bovine serum albumin (BSA, a protein structurally different from Mb). The Δλ B has been depicted as the height in Figure 4e, displaying a very low cross-sensitivity of the sensor for other analytes. The long-term stability of the PVA-rGO sensor has also been determined. One of the sensors used for Mb sensing (Section 4.2) (last measured value is λ B, 100 μ for 100 μg mL À1 Mb) is immersed (after 1 month of the initial experiment) in 100μg mL À1 (85 μL) of Mb for 10 min at NTP conditions and is accompanied by λ B (λ ' B, 100 μ ) measurement. The error is calculated as (λ ' B, 100 μ À λ B, 100 μ )/ λ B, 100 μ . The experiment is repeated for the same sensor in 3, 6, 9, and 12 months. The sensor shows a maximum error of <0.5% after 12 months.

Conclusion
PVA-rGO nanofibers have been coated uniformly with the required thickness on the eFBG sensor using an electrospinning technique with a customized target and sandwich arrangement. The electrospinning parameters have been optimized to achieve low sample volume (<0.1 mL) and spinning time (%8 min). Postprocessing with glutaraldehyde vapors has resulted in PVA acting as a strong adhesive to hold rGO onto the sensor surface.
The functionalization steps adopted are simple, using inexpensive analytes used in small quantities (85 μL). This protocol has led to a strong interfacial bond between PVA and rGO, characterized using FTIR and Raman, and is demonstrated with RI sensitivity enhancement. A case study of Mb detection has displayed a phenomenal increase in detection range, sensitivity, and LOD compared to the rGO dip coat sensor. The sensor has also shown quick detection time, good reproducibility, and stability. In addition to these, eFBG sensors' inherent qualities such as small footprint, multiplexing capability for multianalyte sensing, portability, and immediate quantitative measurements at NTP make PVA-rGO eFBG sensor an attractive contender as a cost-effective point of care, multianalyte kit for large-scale production. The Mb stock solution of 1 mg mL À1 was prepared using PBS (1Â, pH % 7.4) from dehydrated 10 mg mL À1 (laboratory synthesized) Mb. Aliquots of Mb ranging from 100 μg mL À1 to 10 ag mL À1 (in steps of tenfold decrease) were prepared using PBS buffer for dilution.
Drop Cast (Dip/Immerse) Experimental Setup: The 3D-printed sensor holder included four separate pieces (two bases and two plates) fitted together ( Figure 2). The analyte sensor (sensor 1) (3 mm) was multiplexed with the temperature sensor (sensor 2), and the combined sensors (with extra fiber length) were positioned between the two plates such that sensor 1 was accessible through a hole (4 mm diameter) provided in both the plates whereas sensor 2 was inaccessible. The liquids were pipetted through the hole onto sensor 1. A base with a groove/well (4 mm diameter, 1 mm depth) was placed below the hole to hold the liquids and a second base for support. The interrogator (Micron Optics SM 130) had a tunable laser source (1510 to 1590 nm) with an inbuilt receiver and spectrum analyzer. The interrogator was connected to the sensor (through a patch cord) and laptop to capture and save λ B .
The record of λ B (baseline) was initiated on the placement of the sensor in the holder. A 85 μL (to cover the sensor completely) of HF was dropped on the sensor. The sensor was removed from HF when λ B fell %0.9 nm below the baseline and was washed in DI water to remove excess HF.
In situ Preparation and Electrospinning Process: PVA-rGO sensor required lab prepared rGO (prepared using green synthesis method starting from graphene oxide) [50] of required wt% to be dispersed in DI water (2%) and sonicated (1 h) and was transferred into a beaker along with remaining DI (98%) and kept on a hot plate (70 C). PVA (12 wt%) was added in small amounts with continuous stirring (4 h) followed by mechanical stirring (8 h, without heat). The prepared PVA-rGO solution (Figure 1c) was loaded into a syringe (1 mL) and held in a syringe pump (Figure 1d) with the flow rate set at 280 μL h À1 . The needle (0.40 mm diameter) was connected to the anode of a high voltage supply and the aluminum (Al) sheet (foil) (prepared as per 3D-printed customized plastic mold) to the cathode (Figure 1b). On applying a high voltage (13 kV) between the anode and cathode, the nanofiber layer was formed and collected (4 min) across the Al bridge. The eFBG sensor was placed perpendicular to this nanofiber layer and coated again (4 min). The result was an eFBG sensor sandwiched between two nanofiber layers (with <0.1 mL solution). See ST6, Supporting Information, for optimization of electrospinning parameters.
PVA sensor required the PVA (12 wt%) to be added slowly to DI water and stirred continuously for 4 h by maintaining the temperature at 70 C, after which it was stirred for 8 h at room temperature.
Raman and FTIR Spectra: Raman spectra were recorded (HORIBA Scientific instrument) using 532 nm laser excitation with 100Â objective and averaging to improve the S/N ratio. The FTIR experiments had been performed (Perkinelmer, Frontier) to capture samples' spectra in 4000 to 650 cm À1 range with 4 cm À1 resolution using ATR (Attenuated total reflection) mode. The average (32) of collected spectra was calculated to improve the S/N ratio.
UV Spectrophotometer Studies: The DNA had a maximum absorbance at 260 nm, and the concentration of DNA solutions (IN and OUT of Table 1) was measured using microvolume (2 μL) mode of DeNovix DS-11 FX spectrophotometer. PBS (1Â) buffer was used for blank. "IN" is the concentration of the prepared DNA solution. "OUT" is the collected DNA

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.