Fabrication of a Novel Electrochemical Sensor Based on Carbon Cloth Matrix Functionalized with MoO3 and 2D-MoS2 Layers for Riboflavin Determination

The preparation and characterization of a hybrid composite, based on carbon cloth (CC) matrix functionalized with two-dimensional (2D) MoS2 flakes and MoO3, and its use for developing an electrochemical sensor for the determination of riboflavin (RF) is here reported. The 2D-MoS2-MoO3CC composite was prepared by depositing 2D-MoS2 nanosheets, obtained by liquid phase exfoliation (LPE), on the surface of a carbon cloth fiber network, previously functionalized with a layer of molybdenum oxide (α-MoO3) by radio-frequency magnetron reactive sputtering technique. The 2D-MoS2-MoO3CC composite was characterized by scanning electron microscopy and energy dispersive X-ray analysis (SEM-EDX), and Raman spectroscopy. An electrochemical sensor has been then fabricated by fixing a slice of the 2D-MoS2-MoO3CC composite on the working electrode of a screen-printed carbon electrode (SPCE). The 2D-MoS2-MoO3-CC/SPCE sensor display good electrochemical characteristics which have been exploited, for the first time, in the electroanalytical determination of riboflavin (RF). The sensitivity to RF, equal to 0.67 µA mM−1 in the linear range from 2 to 40 µM, and a limit of detection (LOD) of 1.5 µM at S/N = 3, demonstrate the promising characteristics of the proposed 2D-MoS2-MoO3-CC/SPCE electrochemical sensor for the determination of riboflavin.


Introduction
The discovery of graphene has been a revolution in the field of nanomaterials because of its inherent two-dimensionality [1]. This characteristic has given the origin at a very impressive research for using graphene in many devices. For example, two-dimensional electron confinement of ultra-thin 2D graphene has improved the electrical properties compared to other nanomaterials [2]. With the time, a lot of works mentioned a major disadvantage in using graphene, essentially the lack of band-gap.
Nowadays, research is focused on other 2D nanomaterials, such as transition metal disulphides (TMDs), due to their amazing properties. For example, they possess sizable band-gaps around 1-2 eV, promising interesting applications in chemical sensors. Molybdenum disulphide (MoS 2 ) is one of the most interesting TMDs nanomaterials. It has an indirect band-gap of 1.29 eV [3] that turns to semiconductor with a direct band-gap of 1.9 eV [4] going from bulk to a monolayer, offering an important charge carrier mobility. MoS 2 nanosheets have been fabricated via liquid phase exfoliation (LPE) by our groups and used for the detection of different biomolecules [5][6][7][8].
In a previous paper [9], we also investigated the electrochemical properties of MoO 3 , a semiconductor oxide with a band gap of 3.2 eV which presents three crystalline structures

Preparation of the 2D-MoS 2 -MoO 3 CCcomposite
The 2D-MoS 2 -MoO 3 CC composite were fabricated as follows: first, pristine carbon cloth substrates were washed for 2 to 3 times with acetone and deionized water under sonication for 2 h, to remove organic residues and other impurities thoroughly. The cleaned carbon cloth substrates were kept in a hot air oven overnight. They are constituted of long nanofibers. Afterwards, the dry substrates were placed in a radio-frequency magnetron reactive sputtering (Huttenger, Germany) in a customized down setup sputtering mode and Ar + (99.999%) is used as working gas and O − 2 (99.999%) as reactive gas in 1:5 ratio. The Mo metal target was fixed in the working pressure of 10 −2 mbar for depositing a thin layer. A α-MoO 3 thin film (around 350 nm thick) was deposited on carbon cloth substrates at deposition rate of 3 Å/s at 450ºC temperature. The surface contamination on the target material was removed by pre-sputtering the target at 0.01 mbar pressure for 10 min. The RF power was set to 150W, the distance between target and substrate was maintained at 50 mm [9]. 2D-MoS 2 nanosheets were prepared by liquid phase exfoliation (LPE) in sodium cholate (SC) watery solutions (1.5 mg/mL) [6]. Solutions were prepared by tip sonication (Branson S250) of MoS 2 powder (particle size < 2µm, Sigma Aldrich) in SC at a concentration of 5 mg/mL for 30 min. Samples were kept in an ice bath to reduce detrimental heating effects during sonication. The dispersions were allowed to decant overnight in a flask. Then, the half top part was centrifuged at 1500 rpm for 90 min and the supernatant, rich of few-layer nanosheets, was collected. The dispersions thus obtained contained MoS 2 flakes with an average number of layers, n = 9, an average lateral of 170 nm, in an estimated concentration of 80 µg/mL. The dispersions were stable for months [6]. Finally, the 2D-MoS 2 suspension was drop casted onto the MoO 3 carbon cloth and left to dry at room temperature.

Characterization
SEM images were acquired by means of a Zeiss CrossBeam 540 apparatus equipped with EDX detector. Raman spectroscopy was carried out with a XploRA micro-spectrometer (Horiba Scientific), with excitation at 638 nm. The laser beam was focused with a 100X objective (NA 0.9, WD 0.21 mm) on a diffraction limited spot. The Raman signal was collected through the same objective in a backscattering configuration and dispersed by an 1800 lines/mm grating onto a charge-coupled device (CCD) detector (Syncerity, Horiba Scientific). The laser power was set at 0.2 mW in order to avoid sample damaging and the signal was integrated over 30 s.

Modified Electrode Fabrication
To fabricate the modified SPCE, slices of pristine CC, MoO 3 -CC and MoS 2 -MoO 3 -CC were cut in order to cover entirely the surface of the working electrode of SPCE. Then, 6 µL of Nafion solution 5% was dropped on that for enhancing the adhesion on the surface of SPCE and left to dry at room temperature.

Electrochemical Measurements
Electrochemical analyses (Cyclic Voltammetry and Linear Sweep Voltammetry) were performed by using a DropSens µStat 400 Potentiostat empowered by Dropview 8400 software for data acquisition. The sensors were characterized by cyclic voltammetry (CV) and linear sweep voltammetry (LSV) in aerated 1M Phosphate-buffered saline (PBS) electrolyte. CV and LSV tests were carried out at a scan rate of 50 mV/s in the potential range from −0.8 to 0 V, by varying the concentration of the investigated analyte. The calibration curves were obtained by plotting the Faradaic current vs. analyte's concentration. The sensitivity was computed as the slope of the calibration curve.

2D-MoS 2 -MoO 3 -CC/SPCE Characterization
A picture of the modified 2D-MoS 2 -MoO 3 -CC/SPCE is reported in Figure 1a. The sensitive area is 0.125 cm 2 . The network morphology of the 2D-MoS 2 -MoO 3 -CC fixed on the working electrode of SPCE is shown in Figure 1b. The large surface of carbon cloth fibers is beneficial for growing the MoO 3 layer and successively to anchoring the 2D-MoS 2 nanosheets (Figure 1c). This open network structure facilitates the electrolyte ions diffusion during the electrochemical tests. The change of morphology subsequent to the CC surface modification is shown in Figure 2.   centered at 1330 cm −1 and the G-band around 1600 cm −1 . These two peaks correspond to the amorphous carbon phase present in the carbon cloth [24]. mental mapping analysis (Figure 2d,e,f) of the MoS2-MoO3-CC composite-based SPCE confirms the presence and distribution of the main O, Mo, and S elements on the electrode surface.
Raman spectra at 638 nm have been acquired first on the bare carbon cloth SPCE ( Figure 3, black line), then on the carbon cloth added with molybdenum oxide SPCE (Figure 3, blue line) and finally on the MoS2-MoO3-CC/SPCE ( Figure 3, red line). On the CC/SPCE, only the Raman fingerprint of carbon is detected, namely the D-band peak centered at 1330 cm −1 and the G-band around 1600 cm −1 . These two peaks correspond to the amorphous carbon phase present in the carbon cloth [24]. On the MoO3CC/SPCE (blue line) we observe the superposition of the carbon bands and the MoO3 vibrational modes which are attributed to the α-MoO3 orthorhombic phase [25]. The spectrum of the MoS2-MoO3-CC/SPCE is dominated by the MoS2 signal. At 638 nm the photons are almost resonant with the B-excitonic transition of MoS2 [26], leading to resonant Raman (RR) effects [26,27]. When the frequency of the incoming light comes close to the specific frequency needed to drive the transfer of an electron from an occupied state to an unoccupied state, the absolute Raman intensity can change by several orders of magnitude. Besides the amplification of first-order transitions, RR scattering causes second-order Raman processes to be particularly amplified. These exhibit features coming from different crystalline momenta, potentially from the entire Brillouin zone (BZ), with the only constraint of negligible momentum of the phonon pairs involved in the two-phonon process. The strong second order transitions in MoS2 cover the 500 -850 cm −1 fre-  [26], leading to resonant Raman (RR) effects [26,27]. When the frequency of the incoming light comes close to the specific frequency needed to drive the transfer of an electron from an occupied state to an unoccupied state, the absolute Raman intensity can change by several orders of magnitude. Besides the amplification of first-order transitions, RR scattering causes second-order Raman processes to be particularly amplified. These exhibit features coming from different crystalline momenta, potentially from the entire Brillouin zone (BZ), with the only constraint of negligible momentum of the phonon pairs involved in the two-phonon process. The strong second order transitions in MoS 2 cover the 500-850 cm −1 frequency range. Table 1 summarizes the Raman modes observed on our sample, their symmetry and transition order. The position of the first-order transitions E 1 2g (383 cm −1 ) and A 1 g (408 cm −1 ) is typical of the 2H phase on few layers 2D-MoS 2 [6,27]. The 466 cm −1 peakhas been attributed to either an A 2u (Γ) vibration [28] or to an overtone of a longitudinal acoustic (LA) phonon [29].

Electrochemical Characterization
A series of electrochemical tests has been carried out to characterize the bare screenprinted carbon electrodes and the modified ones. Preliminary CV tests have been conducted in 1 M PBS. Figure 4a compares the CV spectra of the bare SPCE and modified CC/SPCE at 0.05 V/s. The strong enhancement of the background current of the CC/SPCE, is due to its large electrochemical double-layer capacitance (EDLC). Cyclic voltammograms of the fabricated electrodes at different scan rates have been also collected. Each point related to sensor response reported is the average of three independent measurements. The standard deviation associated with these measurements has been evaluated to be less than 10%, which is good for these not yet optimized sensors. As expected, the scan rate amplifies the capacitive current. This is well evident for the modified electrodes (compare Figure 4b-e), due to the exposure of more active sites on the working electrode's surface. Plotting the scanning rate versus the current for all electrodes (Figure 4f), a series of straight lines were obtained, allowing us to estimate the EDLC for these electrodes from the slope and the geometrical area. These data show that the EDLC of the CC/SPCE is increased largely compared to the bare SPCE (black line). The EDLC is increased by nearly a factor 2 when the MoO 3 layer is grown on CC/SPCE (dark green line) and undergoes a further strong enhancement when the MoS 2 nanosheets are deposited (blue line). The 2D-MoS 2 -MoO 3 CC/SPCE composite electrode shows the highest EDLC (blue line). This suggests that the network structure of CC is able to provide a larger surface area compared to bare SPCE, which is further increased in the presence of the layered MoO 3 and hybrid 2D-MoS 2 -MoO 3 structure.
The electron transfer capability of the various electrodes has been tested with [Fe(CN) 6 ] 4− as analyte (10 mM in 1 M PBS) by varying the scan rate from 0.05 to 0.4 V/s. In order to provide a quick comparison among the fabricated electrodes, Figure 5 displays the cyclic voltammograms obtained at a scan rate of 0.05 V/s.
As can be noted, both the current peak intensity (Ip) and the peak-to-peak separation (∆Ep) depend on the investigated electrodes. ∆Ep and Ip are helpful parameters to provide a qualitative estimation of the electron transfer rate due to the redox process at the electrode's surface. In Table 2     As can be noted, both the current peak intensity (Ip) and the peak-to-peak separation (∆Ep) depend on the investigated electrodes. ∆Ep and Ip are helpful parameters to provide a qualitative estimation of the electron transfer rate due to the redox process at the electrode's surface. In Table 2 we report the values of the anodic (Ipa) and cathodic (Ipc)  Lower ∆Ep values and higher Ipa and Ipc is measured using the modified electrodes. These values suggest a faster electron transfer in the carbon cloth composite electrodes compared to bare SPCE likely resulting from a larger reaction surface area. Furthermore, the ∆Ep values as well as the current peaks increase linearly with the square root of the scan rate (see Figure 6 for the anodic peak), indicating a diffusion-controlled process.

Electrochemical Behavior in Presence of Riboflavin
We have exploit the enhanced electrochemical performances of the 2D-MoS2-MoO3 layer on CC to develop an electrochemical sensor for biomolecules detection. Here, we tested the new electrode on riboflavin (RF) at a concentration of 100 µM. Figure 7a shows the remarkable enhancement of the CV signal measured with the modified 2D-MoS2-MoO3CC/SPCE (red lines) compared to the bare SPCE (black lines) and MoO3CC/SPCE. This behavior highlight the strong effect of the MoS2 layer on RF electrocatalysis. Figure 7b presents the CV curves of the 2D-MoS2-MoO3CC/SPCE in absence (black dotted line) and in presence of RF (orange line). At the starting potential of -0.8 V, RF exists in its reduced form. At -0.55V, RF exhibits an oxidation peak followed by a reduction peak at -0.72V on the back forward scan. This electrochemical redox process involves the transfer of two protons and two electrons [29]. We have subsequently checked the effect of the loading of MoS2 on the MoO3CC fiber network. As shown in Figure 7c, the current increases as a function of the MoS2 quantity on the MoO3CC matrix, thus prov-

Electrochemical Behavior in Presence of Riboflavin
We have exploit the enhanced electrochemical performances of the 2D-MoS 2 -MoO 3 layer on CC to develop an electrochemical sensor for biomolecules detection. Here, we tested the new electrode on riboflavin (RF) at a concentration of 100 µM. Figure 7a shows the remarkable enhancement of the CV signal measured with the modified 2D-MoS 2 -MoO 3 CC/SPCE (red lines) compared to the bare SPCE (black lines) and MoO 3 CC/SPCE. This behavior highlight the strong effect of the MoS 2 layer on RF electrocatalysis. Figure 7b presents the CV curves of the 2D-MoS 2 -MoO 3 CC/SPCE in absence (black dotted line) and in presence of RF (orange line). At the starting potential of −0.8 V, RF exists in its reduced form. At −0.55V, RF exhibits an oxidation peak followed by a reduction peak at −0.72V on the back forward scan. This electrochemical redox process involves the transfer of two protons and two electrons [29]. We have subsequently checked the effect of the loading of MoS 2 on the MoO 3 CC fiber network. As shown in Figure 7c, the current increases as a function of the MoS 2 quantity on the MoO 3 CC matrix, thus proving the fundamental electrocatalytic role of MoS 2 in enhancing the response to RF. Based on these results, we have evaluated the response of the 2D-MoS2-MoO3CCbased sensor at different RF concentrations. In Figure 8a, is reported the linear sweep voltammetry (LSV) analysis of solutions containing increasing concentrations of RF, from 0 to 40 µM, evidencing the associated augmentation of the peak current value. In Figure 8b (black dots), is shown the calibration curves for RF, plotting the peak current as a function of the analyte concentration. The sensitivity, namely the slope of the calibration curve, is computed from a linear fit of the data (red line), and is 0.67 µA µM −1 . LOD was 1.5 µM, as calculated by comparing the signals (S) from the samples with known and low concentrations of RF with those of blank samples (N) and by establishing the minimum concentration at which the RF signal is three times as high as noise (S/N = 3).  Based on these results, we have evaluated the response of the 2D-MoS 2 -MoO 3 CCbased sensor at different RF concentrations. In Figure 8a, is reported the linear sweep voltammetry (LSV) analysis of solutions containing increasing concentrations of RF, from 0 to 40 µM, evidencing the associated augmentation of the peak current value. In Figure 8b (black dots), is shown the calibration curves for RF, plotting the peak current as a function of the analyte concentration. The sensitivity, namely the slope of the calibration curve, is computed from a linear fit of the data (red line), and is 0.67 µA µM −1 . LOD was 1.5 µM, as calculated by comparing the signals (S) from the samples with known and low concentrations of RF with those of blank samples (N) and by establishing the minimum concentration at which the RF signal is three times as high as noise (S/N = 3). , x FOR PEER REVIEW 11 of 15 The performances (linear range, sensitivity and limit of detection) of the proposed 2D-MoS2-MoO3CC/SPCE based sensor have been compared to most of the recently reported riboflavin electrochemical sensors (see Table 3).  The performances (linear range, sensitivity and limit of detection) of the proposed 2D-MoS 2 -MoO 3 CC/SPCE based sensor have been compared to most of the recently reported riboflavin electrochemical sensors (see Table 3). From this comparison it can be deduced that our sensor platform displays a wide linear range and a high sensitivity respect to the other sensors. We repeated this test after about one year using a new prepared 2D-MoS 2 nanosheets suspension for replicating the fabrication of a new 2D-MoS 2 -MoO 3 CC/SPCE sensor in the same conditions of the first one. The calibration curves obtained by these tests are compared in Figure 9. It can be clearly observed that the two set of data points can be fitted almost well from the same linear relationship. The reported findings suggest that the electrochemical properties of the 2D-MoS 2 -MoO 3 CC network structure as well as the fabrication procedure of 2D-MoS 2 -MoO 3 CC/SPCE platform can be replicated very well, leading to different sensor devices with reproducible response. clearly observed that the two set of data points can be fitted almost well from the same linear relationship. The reported findings suggest that the electrochemical properties of the 2D-MoS2-MoO3CC network structure as well as the fabrication procedure of 2D-MoS2-MoO3CC/SPCE platform can be replicated very well, leading to different sensor devices with reproducible response. The effect of some interferent biomolecules has been also investigated. Folic acid (FA), which is another vitamin of B group, and ascorbic acid (AA), a vitamin of C group, are considered as the main interferent analytes in the determination of RF. These two biomolecules are present in human body so it's mandatory to verify if their presence affect the RF detection. Preliminary tests have shown that these compounds, in absence of RF, show no redox peak in the potential range where the RF peak is present. This is an expected result because, similarly to redox processes of most organic compounds, they take place in the positive potential range.
The effect of FA and AA on the determination of riboflavin has been investigated with our 2D-MoS2-MoO3CC/SPCE sensor. The test has been carried out at different concentrations of RF in presence of the two analytes. LSV curves measured in a solution of RF mixed with 100µM AA and 100µM FA are reported in Figure 10. The test shows clearly how the presence of the two interferent analytes decreases the RF current peak. This behavior can be explained assuming that FA and AA compete with RF for the interaction with a significant fraction of the active site on the sensing layer. The effect of some interferent biomolecules has been also investigated. Folic acid (FA), which is another vitamin of B group, and ascorbic acid (AA), a vitamin of C group, are considered as the main interferent analytes in the determination of RF. These two biomolecules are present in human body so it's mandatory to verify if their presence affect the RF detection. Preliminary tests have shown that these compounds, in absence of RF, show no redox peak in the potential range where the RF peak is present. This is an expected result because, similarly to redox processes of most organic compounds, they take place in the positive potential range.
The effect of FA and AA on the determination of riboflavin has been investigated with our 2D-MoS 2 -MoO 3 CC/SPCE sensor. The test has been carried out at different concentrations of RF in presence of the two analytes. LSV curves measured in a solution of RF mixed with 100µM AA and 100µM FA are reported in Figure 10. The test shows clearly how the presence of the two interferent analytes decreases the RF current peak. This behavior can be explained assuming that FA and AA compete with RF for the interaction with a significant fraction of the active site on the sensing layer. The above results indicate that the effects of interferents substances on the RF sensor response needs to be checked. As it is well known, the electrochemical characteristics of The above results indicate that the effects of interferents substances on the RF sensor response needs to be checked. As it is well known, the electrochemical characteristics of the biomolecules here investigated are largely dependent on pH [35][36][37]. This implies that it is possible to optimize the conditions at which riboflavin has the larger interaction with the electrode surface, which means that the sensor response will be probably less influenced by the presence of these interferent analytes. This is a general strategy which can be used for monitoring riboflavin in real samples [38,39], thus we have planned to investigate in detail this aspect aim at applying our developed sensor in real applications.

Conclusions
A 2D-MoS 2 -MoO 3 CC nanocomposite electrode was prepared and tested for the electrochemical detection of riboflavin in PBS. The modified platform was fabricated by layers of MoO 3 and 2D-MoS 2 nanosheets on the fiber of CC network structure. The MoO 3 layer was grown on CC fibers through radio-frequency magnetron reactive sputtering. An additional layer of 2D-MoS 2 nanosheets was casted, starting from a solution of MoS 2 few-layers flakes produced by LPE.
The structure was used for fabricating a novel electrochemical platform with enhanced properties with respect to the conventional commercial devices and those prepared with the single constituents. Riboflavin electroxidation was observed at negative potential (around −0.55 V) and this can be regarded as an advantage, because most of the redox processes of organic compounds which can interfere with it take place at more positive potentials. Interference from folic and ascorbic acid has been observed to decrease the sensitivity of the sensor for riboflavin, although not dramatically.
The enhanced sensitivity of the 2D-MoS 2 -MoO 3 CC/SPCE sensor was attributed to the improved electron transport provided by the conductive CC structure network, as well as the enhancement of specific active sites for the electrocatalytic reduction/oxidation of RF on MoS 2 . This unique 2D-MoS 2 -MoO 3 CC/SPCE electrochemical platform resulted in a promising sensor device for the electroanalytical determination of riboflavin.