Fabrication and Electrochemical Behavior Investigation of a Pt-Loaded Reduced Graphene Oxide Composite (Pt@rGO) as a High-Performance Cathode for Dye-Sensitized Solar Cells

A platinum-reduced graphene oxide thin film composite (Pt@rGO, 100 nm) was prepared on a fluorine-doped tin oxide(FTO-) coated glass substrate by a screen printing method using a Pt@rGO screen printing paste (0.12% Pt; Pt/rGO = 1:5w/w). The asprepared electrode (denoted as Pt@rGO/FTO) was used as the cathode for the assembly of dye-sensitized solar cells (DSSCs). It showed a well-dispersed and high loading of Pt on rGO surface with a particle size distributed around 10 nm. The redox behavior of ferrocene was performed at Pt/FTO, Pt@rGO/FTO, and rGO/FTO electrodes by a cyclic voltammetry (CV) method. The kinetic parameters, in particular, the standard reduction potential (E0, V), the transfer coefficient (α), the heterogeneous rate constant (k0, cm·s ), and the diffusion coefficient (D, cm s), were determined by CV data treatment using convolutiondeconvolution and fitting methods. The values of E0, α, k0, and D at Pt@rGO/FTO electrode were, respectively, 326mV, 0.471, 3.33 cm·s, and 4.19 cm·s, equivalent to those of Pt/FTO electrode (340mV, 0.474, 3.18 cm·s, and 4.19 cm·s). The Pt@rGO/FTO electrode exhibited excellent electrocatalytic activity compared to that of Pt thin film (Pt/FTO electrode) prepared from Pt commercial paste. The heterogeneous electron transfer rate constant k0 (cm·s ) for I3 /I at Pt@rGO/FTO is 1.3 times faster than that at Pt/FTO. The energy conversion efficiency of the DSSCs assembled from Pt@rGO-DSSC cathode reached 7.0%, an increase of 20.7% over the commercial Pt-based cathode (Pt-DSSC, 5.8%). The rGO component in the Pt@rGO composite plays two important roles: (i) facilitating the electron transfer between Pt NPs catalyst and the FTO substrate via the bandgap effect and (ii) the enlargement catalytic surface area of Pt NPs via the loading effect. The rGO material has, therefore, potential to replace the Pt content and improve the performance of the DSSC device.


Introduction
The pioneering work of O'Regan and Grätzel [1] launched a low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiO 2 films, named dye-sensitized solar cell (DSSC). Considered as the third generation of solar cells, the DSSC has attracted a lot of attention because of the simplicity of its manufacturing processes, respect for the environment, and good plasticity [2][3][4]. DSSC consists of three main components: a photo-sensitized anode, a liquid electrolyte, and a platinum-based cathode. The operating principle of DSSC is based on the process of mimicking natural photosynthesis using photosensitive dyes. Of which, in the first step-the current generation step-the dye absorbs photon (hν) from sunlight and turns into the excited state (S * ), the excited electron is then injected into the porous TiO 2 layer and flows out of the anode to the cathode resulting in the oxidation of the excited state of the dye to an oxidation state (S + ); in the second step-the regeneration step-the oxidation state of the dye will be reduced to ground state (S) by a dissolved ion mediator Iin the liquid electrolyte to form I 3 -; in the last step-shuttle redox generation-the oxidation oxidized redox mediator, I 3 − , was reduced to I − ions at the cathode surface driven by Pt catalyst [4,5]. It is important to note that the platinum-based cathode is an expensive material in the DSSC construction. Many efforts have been made to reduce Pt content in the DSSC cathode [3]. As such, recent developments of carbon-based materials, such as graphene and graphene oxide, appear as a potential material for Pt replacement [4]. Graphene is an advanced one-atom-transparent material with sp 2 -hybridized carbon atoms packed into a hexagon lattice structure. Graphene has an extremely large surface area, high conductivity, high carrier mobility, and chemical inertness [4]. The production of graphene, however, requires special manufacturing methods. Graphene oxide (GO) is a material that contains functional groups on the carbon hexagon lattice structure. GO is simpler to fabricate than graphene, and GO can be reduced to graphene oxide (rGO) with a comparable structure to graphene by a controlled reduction process [6]. In addition, the high level of defects on the rGO may provide more sites for anchoring as well as stabilizing the nanoparticles [4]. Earlier reports have mentioned the partial reduction of Pt in the cathode of DSSC by graphene and graphene oxide. Guai et al. prepared a graphene-Pt\ITO cathode by electrochemical deposition of a porous graphene film on a low loaded Pt\ITO electrode. They reported that the as-prepared cathode reduced more than 60% of Pt while achieving better performance (7.1%) attributed to the lower charge transfer resistance and the large conductive interfacial surface of the graphene coating [7]. Bolhan et al. reported the platinum-rGO cathode for DSSCs by mixing rGO and commercial Pt paste, the photoperformance of DSSCs showed an efficiency of 5-6% [8]. Similarly, Gong et al. mentioned a high electrocatalytic activity (7.6%) DSSC based on a Pt-rGO hybrid cathode prepared via a sequential spin coating and thermal reduction of GO and Pt precursors [9]. Although it has been shown that graphene and rGO provide a large and superior conductive interface between the platinum nanoparticles and the conductive glass electrode [7,10], the electrochemical behavior of the composites/hybrids (Pt-rGO) and their electron transfer kinetics within the DSSC have not been systematically studied. In addition, controlling the uniform distribution and the high loading of Pt on the surface of rGO are the key factors in reducing Pt content while maintaining the catalytic performance of the cathode. Lastly, the preparation of a platinum-based composite screen printing paste for the manufacture of the cathode is an important demand to realize and evaluate the possibility of reducing Pt content in the DSSC device.
Herein, we reported a facile and rapid route to prepare a platinum-reduced graphene oxide screen printing paste from H 2 PtCl 6 precursor and rGO as starting materials.Using this paste, a platinum-reduced graphene oxide thin film composite (Pt@rGO) was prepared on a fluorine-doped tin oxide-(FTO-) coated glass substrate via a two-step screen printing/thermal reduction method. The as-prepared electrode was used as the cathode for the assembly of dyesensitized solar cells (see Scheme 1). The morphologies, structures, electrochemical behavior, electron transfer kinetics of composite materials, and the characteristic of the associated DSSC were detailed.

Fabrication of Pt@rGO Composite Screen Printing Paste
2.2.1. Preparation of rGO. Reduced graphene oxide (rGO) was synthesized from graphite (Gr) as a starting material via a two-step protocol. In the first step, graphene oxide (GO) was synthesized from Gr by using the modified Hummer method [11]. Accordingly, Gr (3.0 g, powder) was dispersed in a concentrated binary mixture of H 2 SO 4 (360 mL) and H 3 PO 4 (40 mL) at ambient temperature. To this mixture, 18 g KMNO 4 was added, the reaction mixture was then heated to 50°C for 12 h under continuous stirring. After cooling, 500 mL H 2 O and 15 mL H 2 O 2 (30% v/v) were added to the reaction mixture. The resulting solid was then separated by centrifugation, washed with distilled water, and dried at 50°C for 12 h. The resulting oxidized graphite powder was then dispersed in distilled water (1 g/L), and the exfoliation of the graphite layer was performed under sonication was coated on FTO substrate by screen printing with a mesh size of 43 T using transparent 18NR-T titania paste (two layers) and active opaque 18NR-AO titania paste (one outer layer) prepared according to our previous paper [12,14]. The electrodes were dried at 120°C after each printed layer and finally heated at 500°C under airflow for 30 minutes to obtain photoanode TiO 2 .

Cathode Preparation.
The Pt@rGO composite paste was printed on FTO substrate using screen printing with a mesh size of 90 T (3 layers, heating at 120°C for 15 minutes after each printed layer, denoted as Pt@rGO/FTO). For comparison, the commercial Pt paste and rGO paste were also used to fabricate the electrodes at the same conditions (respectively, denoted as Pt/FTO and rGO/FTO). The prepared electrodes were annealed at 450°C for 30 minutes which were to be used as cathodes in DSSCs.

DSSC Assembly.
The DSSC assembly was performed in the nitrogen-filled glove box to avoid oxygen and water. Firstly, the DSSCs were assembled by placing a hot-melt Surlyn film (25 μm) between the photoanode and the cathode and then pressed with a thermopress at 170°C in 15 seconds. Secondly, the N719 dye solution (10 mM in dimethylforma-mide) was continuously injected into the cell and then aspirated through the holes at the back of the cathode using a syringe (four times, soaking 20 minutes in the fourth times), following by cleaned with acetonitrile solvent (three times). The HSE redox electrolyte solution was then injected into the cells using the same dye injection method. Finally, the holes were covered with a thin glass slide and sealed with a Surlyn film by hot pressing at 170°C for 15 seconds to obtain DSSC devices.

Characterizations
2.4.1. Structural Characterization. Fourier transform infrared (FTIR) analysis (range 500 to 4000 cm -1 ) was performed on a Bruker FT-IR Alpha. Raman spectroscopy measurements were carried out on a LabRAM HR 800 Raman Spectrometer (HORIBA Jobin Yvon) with the green line of an argon-ion laser (514 nm) as the excitation source, and the Raman bands were collected in the wavelength range of 200-3000 cm -1 at room temperature. The crystalline structures of materials were characterized by X-ray diffractometer D2 Phaser (Bruker, Germany) with a copper anode (λ = 1:54 Å). The XRD patterns were acquired in the 2θ range of 10°-80°(0.02°per second). The morphology and chemical composition of Pt@rGO composite were analyzed by a transmission electron microscopy (TEM) method on a TEM 1400 instrument (JEOL, Japan) and by a scanning electron microscope (SEM) method with an energy dispersive X-ray (EDX) detector using a Jeol 6490-JED 2300 (Japan).

Electrochemical
Characterization. The electrocatalytic behavior of rGO, Pt, and Pt@rGO thin films was investigated by a cyclic voltammetry (CV) method using a conventional three-electrode electrochemical cell. The working electrodes (WE) are Pt/FTO, rGO/FTO, and Pt@rGO/FTO; a platinum mesh and an Ag/AgCl (ALS, Japan) are used as the counter electrode (CE) and the reference (RE) electrode, respectively. The CV measurements were carried out under a nitrogen atmosphere in a solution of 5 mM ferrocene and 0.01 M of LiClO 4 electrolyte in acetonitrile using an Autolab 302 N (Ecochemie, Netherlands) instrument equipped with NOVA software. The electrochemical kinetic reaction of the ferrocene redox reaction was thoroughly investigated via convolution (semiintegration), deconvolution (semidifferentiation) [15,16], and fitting the CV data using EC-Lab software. The diffusion coefficient (D, cm 2 ·s -1 ) was calculated from convolution-deconvolution transforms using equation (1) and (2) [17,18].
where I lim is the limiting value achieved for the convoluted current when the potential is driven to a sufficiently extreme value past the peak, n is the number of exchanged electrons, F is Faraday's constant (96485 Cmol -1 ), A is the surface area (cm 2 ), e p is the peak height (A) of the forward deconvolution 3 International Journal of Photoenergy sweep, α is the transfer coefficient (obtained from fitting results), ν is the potential scan rate (Vs -1 ), C b is bulk concentration of electroactive species (M), R is the universal gas constant (8.314 Jmol -1 K -1 ), and T is absolute temperature (K).

DSSC Characterizations.
The photocurrent-voltage curves (I-V curves) of the DSSCs were measured using a Keithley model 2400 multisource meter under one sun condition using an Oriel Sol1A class ABB solar simulator (Oriel-Newport-USA, Model No. 94061A, with a 1000 W Xe lamp and an AM 1.5 filter, 100 mW·cm -2 ). The electrochemical impedance spectroscopy (EIS) of fabricated DSSCs was measured by using Autolab 302 N. The EIS measurement was carried out under light illumination of 100 mW cm -2 at open-circuit voltage with the alternating voltage amplitude 10 mV and frequency range from 0.01 Hz to 100 kHz.

Results and Discussion
3.1. Morphological and Structural Analysis. Figure 1 presents the FTIR spectra of GO, rGO, and Pt@rGO composite. The FTIR spectrum of GO (Figure 1(a)) shows six characteristic vibrating modes specified for the functional groups on GO, notably the stretching of C-O (1050 cm -1 ) and C-OH (1240 cm -1 ), the bending of the O-H (1400 cm -1 ) and the C=O (1625 cm -1 ), the stretching of the C=O (1720 cm -1 ) and the O-H (3427 cm -1 ) [6,11,19,20]. FTIR results clearly show that the graphite powder has been successfully oxidized and exfoliated to GO sheets upon oxidation-sonication process and conversed to rGO after treatment with ascorbic acid. In fact, the GO functional group peaks disappeared after the reduction process, and a new peak appeared at 1567 cm -1 assigned to the skeletal vibrations of rGO backbone chain (Figure 1(b)). This peak is shifted to a higher wavenumber at 1641 cm -1 when platinum nanoparticles (Pt NPs) are incorporated onto rGO sheets (Figure 1(c)). The structural order of rGO was further characterized by Raman spectra. Figure 2 illustrates the Raman spectra of graphite, GO, rGO, and Pt@rGO. The Raman spectrum of graphite appears three well-known peaks at 1332 cm -1 (D-band), 1580 cm -1 (G-band), and 2685 cm -1 (2D-band) of graphite materials [19][20][21]. The G-band and the D-band are attributed to the in-plane stretching (E 2g mode) and out of plane vibrations (the symmetry A 1g mode) of the sp 2 hybridized carbon atoms in graphite lattice. It is believed that the D-band is assigned to the vacancies in the graphite layer or at its edge and is also related to the defects in graphite material. The 2D-band or D * -band (originally called G ' -band) in the range 2500-2800 cm -1 corresponds to the overtone of the D-band and related to the number of graphite layers, which is the signature of the graphitic materials [22]. A sharp reduction in the intensity of the 2D-band in the GO and rGO spectra made it possible to define a very thin layer structure (several layers) of these materials. Both D-band and G-band of GO are broadened compared to that of graphite, indicating increased disorder. In addition, the D-band peak of GO is located almost at the same frequency as that of the graphite while the peak of the G-band is shifted by 20 cm -1 towards the high frequencies (blue shift) to 1600 cm -1 . The G-band of rGO returned close to the initial position of graphite which confirmed the reduction of GO after treatment with citric acid. Otherwise, the intensity ratio of D-band to G-band (I D /I G ) can be used to evaluate the concentration of defect or disorder degree of graphitic carbon structure [13,20,23]. The I D /I G ratio increased from 0.33 (graphite) to 1.32 (GO), 1.21 (rGO), and 1.30 (Pt@rGO) related to the increment of defects in the carbon network. Note that the reduced process preparing rGO multiplied the sp 3 hybridizations in carbon sheets from the sp 2 hybridization in graphite. Moreover, the nano Pt particles also contribute in enhancing the disorder of rGO network due to the incorporation in carbon lattice [19,20]. To further investigate the effect of the reduction process and the Pt-loading on the degree of crystallization of rGO, the crystalline domain sizes of the GO and rGO were calculated by using equation (3) [22].

International Journal of Photoenergy
The obtained values of the crystalline domain sizes for GO, rGO, and Pt@rGO samples are about 12.5, 14, and 13 nm. The L a value of rGO has increased compared to that of GO, again, attributed to the increase of the sp 2 domain in rGO after the reduction process. The decrease in the L a value of rGO in Pt@rGO may be related to the anchoring of Pt NPs to rGO sheets via the covalent bond to the oxygen group on the rGO structure [24,25]. In addition, Raman spectra of Pt/FTO, rGO/FTO, and Pt@rGO/FTO were also provided (see Figure S2).
The powder X-ray diffraction (PXRD) patterns of GO, rGO, and Pt@rGO are illustrated in Figure 3. The PXRD pattern of the GO shows a diffraction peak at 2θ value of 13.68°w hich corresponds to the (001) crystalline planes of graphite oxide. The corresponding interlayer distance (d 001 ) calculated from Bragg's law (0.64 nm) shows a significant increase in spacing from that of graphite (d 002 = 0:335 nm) due to the intercalation of the oxygen functional groups in the graphite interlayer and confirming a high degree of oxidation in the graphite oxide structure. In addition, two weak diffraction peaks at about 30°and 42.5°can be observed due to the incomplete oxidation of graphite (Figure 3(a)). When GO is reduced to rGO, the peak at 13.68°disappeared, and a new broad peak appeared at about 24.8°(d 002 = 0:36 nm) corresponding to the (002) plane (Figure 3(b)). Such decreased d spacing value in rGO demonstrates the complete removal of the oxygen-functional groups (a epoxide group, a carboxylic group, and a hydroxyl group) during the reduction process.
The PXRD pattern of Pt@rGO composite showed the (002) diffraction peak at the 2θ value of 24.8°indicating the crystalline nature of the rGO in the composite. In addition, there are three diffraction peaks at 39.  [8]. The broadening of the diffraction peaks is considered as the nanocrystalline characteristic of Pt. The average crystalline size of Pt particles is calculated from the full width of the half maximum (FWHM) of most intensity peak (111) through the Debye-Scherrer equation (4) [13,20].
where d is the average size (nm), k is the constant depending on the crystallite shape (0.9), λ is the wavelength of copper Kα X-ray radiation (0.154 nm), β is the FWHM of the most intense peak (in radians), and θ is the diffraction angle. The FWHM of (111) peak is about 0.014 rad (fitted from origin software), and the corresponding particle size calculated from equation (3) is about 10 nm. Similarly, the stacking height (L c ) of the rGO crystallite determined from the FWHM of (002) peak using equation (4) is about 0.76 nm, which corresponds to the number of layer (N) in rGO of about 3 layers (calculated from equation L 002 = ðN − 1Þ · d 002 ) [26,27]. This result is in good agreement with the Raman analysis.
The microstructures of Pt@rGO composite and Pt@rGO thin film were analyzed by TEM and SEM/EDX methods. As shown in Figure 4, the TEM image of Pt@rGO clearly shows two phases, the broad and bright phase is the rGO sheets, whereas the metallic phase of Pt NPs is decorated on rGO sheets as black spots. The spherical Pt nanoparticles are well distributed on the surface of the rGO layers with a particle size in a range of 10 nm. In addition, the thick dark lines observed in the rGO phase indicated the multilayer structure of rGO (Figures 4(a) and 4(b)). It can be seen clearly from the TEM image that the density of Pt NP loading on rGO is very high due to its nanosize characteristic (100 Pt NPs per 1 μm 2 rGO). The SEM image of Pt@rGO thin film prepared on an FTO substrate by a screen printing method (one layer, see Figure 4(c)) using a Pt@rGO-based paste clearly shows the FTO substrate (gray, flat), rGO (dark, clouds: sharp), and Pt NPs (bright, clusters: shape). For comparison, the SEM images of Pt/FTO and Pt@rGO/FTO were also provided (see Figure S3). The EDX spectrum of the corresponding thin film indicates the presence of the elements C, O, and Pt in the composition of Pt@rGO composite, the mass ratio of Pt/rGO being approximately 1.56%, close to the mass ratio used to prepare the paste (1.5%). The cross-sectional TEM image of a thin Pt@rGO film prepared on FTO (three layers) shows a thickness of about 100 nm (Figure 4(d)).

Electrochemical
Behavior. The electrocatalytic activity of Pt/FTO, rGO/FTO, and Pt@rGO/FTO electrodes toward the ferrocene redox reaction was carried out by CV measurements. As can be seen in Figure 5, all the CV curves showed a symmetry redox pair corresponding to the reversible electron transfer in ferrocene redox reaction. It important to note that both Pt and rGO showed good catalyst activity for the ferrocene redox reaction. From the CV curves, the Pt@rGO/FTO electrode exhibited the highest faradic current, while the rGO/FTO showed the highest capacitive current.
In order to further figure out the electrochemical behavior of these electrodes, the CV curves were further analyzed by convolution-deconvolution transforms and a fitting 5 International Journal of Photoenergy method using the Nova software and EC-Lab software (see Figure S1). The kinetic parameters such as standard reduction potential (E 0 , V), the transfer coefficient (α), the heterogeneous rate constant (k s, cm·s -1 ), and the diffusion coefficient (D, cm 2 ·s -1 ) are presented in Table 1. The diffusion coefficient of ferrocenium estimated by different methods gives a similar value and in the range of 3 × 10 −5 to 5 × 10 −5 cm 2 · s −1 depending on the composition of the electrodes, which is in good agreement with other reports [28]. The k s value of ferrocene redox reaction on the Pt@rGO/FTO electrode (3:33 × 10 −3 cm · s −1 ) was found to be higher than that of Pt/FTO (3:18 × 10 −3 cm · s −1 ). The fast electron transfer on Pt@rGO/FTO can be attributed to the bandgap effect (waterfall effect) in the FTO/rGO/PtNPs system. In fact, the transfer of electrons from the FTO (E CB = −4:4 eV) to the Pt NPs (E F = −5:6 eV) via a middle energy level of the bridge rGO (E F = −4:6 eV) is much facile than the direct transfer of electrons between the FTO substrate and the Pt NPs (see Figure 6), resulting in a 4.7% increase in the value of k s . The ratio of peak current (i pa /i pc ) obtained for both Pt@rGO/FTO and Pt/FTO is close to 1, and the difference in the peak potentials (ΔE p~2 00 mV) 6 International Journal of Photoenergy higher than the standard value (59 mV) of one-electron Nernstian' process exhibited the quasireversible characteristic of the ferrocene redox reaction on these electrodes. Interestingly, the peak anodic current (i pa ) increased by about 4%, from 342 mA·cm -2 (on Pt@rGO/FTO) to 356 mA·cm -2 (on Pt/FTO), indicating the large surface area of Pt NPs in Pt@rGO/FTO compared to Pt/FTO. This result suggests that rGO helps improve the loading efficiently and thus improves the performance of the Pt catalyst on the FTO electrode. The bandgap-high loading synergic effect when using rGO in the composite Pt@rGO is the key factor contributing to the high catalyst activity and, therefore, the high performance of the DSSCs. It is important to note that rGO also has moderate catalyst activity toward ferrocene redox reaction (k s = 2:78 cm s −1 ). However, it caused a capacitive current due to the layer structure of graphitic materials. In addition, the diffusion coefficient of ferrocene ions on Pt@rGO/FTO is the same with Pt/FTO (4:19 × 10 −5 cm 2 s −1 ) showed that the redox reaction of ferrocene occurred mainly on the surface of Pt@rGO/FTO electrode.

DSSC Performances.
To study the effect of rGO on the performance of DSSC, we assembled the DSSCs using three different cathodes Pt@rGO/FTO, Pt/FTO, and rGO/FTO, respectively, denoted as Pt@rGO-DSSC, Pt-DSSC, and rGO-DSSC; the other components of DSSCs such as photoanode TiO 2 films, electrolyte, and dye N719 were prepared with the same process. Figure 7 shows the I-V curves of the Pt@rGO-DSSC, Pt-DSSC, and rGO-DSSC; the corresponding photovoltaic parameters are given in Table 2. The fill factor (FF) and the energy conversion efficiency (η) are calculated from the short-circuit current (J sc ), the opencircuit voltage (V oc ), the maximum power output (P max ), and the input optical power (P in ) using equations FF = ðP max Þ/ðJ sc V oc Þ and η = ðP max 100Þ/ðP in Þ. Compared with other DSSCs, the Pt@rGO-DSSC shows the highest shortcircuit current (12.1 mA/cm 2 ), high open-circuit voltage (726 mV), and fill factor (0.61) and, consequently, the highest energy conversion efficiency (7.0%). This result is comparable to the recent reports in cathode for DSSC. Wang et al. reported that the monodispersed Co 9 S 8 nanocrystals (NCs) on reduced graphene oxide (rGO), prepared by hydrothermal route, showed an excellent catalytic activity and fast electron transfer ability in cathode for DSSC with a 7.31% of overall photoelectrical conversion efficiency [29]. The author Oh and coworker prepared Ag nanoparticles doped on Graphene-Ba 2 GaInO 6 (GBGI@Ag) by hydrothermal. Ag noble metallic nanoparticles improved chemical and electronic properties that led an excellent power conversion efficiency at 9.90% [30]. The author Vijaya and coworker prepared MoS 2 nanosheet as counter electrode, the fabricated dye-sensitized solar cells showed an open-circuit voltage of 795 mV, short circuit current of 19.6 mA/cm 2 , fill factor of 0.36, and power conversion efficiency of 6.6% [31]. Clarity, using an appropriate amount of rGO to fabricate the Pt@rGO composite-based cathode, helps to enhance the DSSC energy    International Journal of Photoenergy conversion efficiency. Similar observations have been mentioned in many other previous reports [10,13,32]. Interestingly, as a part of this work, a 20.7% increase in the energy conversion efficiency of DSSC was investigated when using the Pt@rGO composite (Pt/rGO = 1:5 w/w) to partially replaced the Pt in the conventional cathode.
To clarify the enhancement mechanism by rGO, EIS measurements of the DSSCs were carried out under illumination at open-circuit voltage. As represented by the Nyquist plots in Figure 8, both Pt@rGO-DSSC and Pt-DSSC three semicircles in the frequency range of 0.01 Hz-100 kHz appeared. The small semicircle at a high frequency is attributed to the charge transfer for the reduction reaction of I 3 − at the counter electrode (C CE , R CE ). The large semicircle at the intermediate-frequency region reflects the electron transport in the mesoscopic TiO 2 film associated with the back reaction at the TiO 2 /electrolyte interface (C CT , R CT ). The response in the low-frequency region is the finite Warburg impedance (R D , C D ) of the I − /I 3 − in the electrolyte. The equivalent circuit model fitting the experimental EIS data is shown in Figure 8, the R s accounts for the resistance of the electrolyte (25 Ω) [33]. The fitted data of the C CE , R CE , C r , R r , C d , and R d obtained from the Nyquist plots of Pt@rGO-DSSC and Pt-DSSC are shown in Table 3. The R CE of Pt@rGO-DSSC is smaller than that of Pt-DSSC which indicated the fast electron transfer process in Pt@rGO composite (decrease by 23%). Clearly, the presence of rGO in the composite reduced the charge transfer resistance between the counter electrode and the electrolyte solution [7]. The electrochemical rate constant k o (cm·s -1 ) values for the I − /I 3 − couple on the cathodes can be calculated from charge transfer resistance R CE using equation R CE = RT/ðn 2 F 2 AC k o Þ [34,35]. Using this equation, the ratio of k o between two cathodes (I and II) can be calculated via the ratio of As a result, the k o ratio value between Pt@rGO/FTO and Pt/FTO is about 1.3, which means that the electron transfer rate on Pt@rGO/FTO is 1.3 times faster than that of Pt/FTO. In addition, the triiodide diffusion coefficient, D I can be calculated from the peak frequency of the low-frequency arc (ω max = 1/R d C d ) using the equation D I = ð1/2:5Þδ 2 ω max , where δ is the thickness of the Surlyn film (25 μm) [36,37]. The estimated triiodide D I values for Pt@rGO-DSSC and Pt-DSSC are 0:6 × 10 −5 and 0:8 × 10 −5 cm 2 s −1 , respectively, and in good agreement with other reports [8,36]. These EIS results are consistent with the high photovoltaic performance of Pt@rGO-DSSC and the    Overall, the high performance of the DSSC-based Pt@rGO or based on other Pt-graphitic nanosheet composites/hybrids can be explained by consideration of two key factors: (i) the large catalytic surface area of Pt NPs upon loading on the graphitic nanosheets (loading effect) and (ii) the reduction of electron transfer resistance due to the presence of graphitic nanosheets (bandgap effect). The electron transfer mechanism involving the Pt@rGO cathode in the DSSC is proposed in Figure 6. Accordingly, when the DSSC illuminated under sunlight, the photoelectrons from the photoanode are injected into the conducting band (CB) of the FTO current collector at the cathode (FTO-CB); due to the lower energy (Fermi level) of rGO than of FTO-CB, the photoelectrons can be collected on the rGO nanosheets, then easily transferred into Pt NPs to reduce the I 3 − to I − at a high rate, which leads to an improvement of the photocurrent in DSSCs. In addition, the concentration of electrons on rGO sheets will lead to the enlargement of the V OC of the cell, as shown in Table 2.

Conclusion
A screen-printable paste based on Pt NPs uniformly loaded on rGO nanosheet composite is prepared to fabricate the cathodes in DSSC. The composite cathode exhibited an encouraged photoperformance with a significant increase in the energy conversion efficiency of 20.7% over conventional platinum cathode DSSC. The electrochemical studies evidenced the small interfacial cathode/electrolyte resistance that benefited the electron-transfer process on the cathode side. The high performance of the DSSCs based on Pt@rGO cathode attributed to the bandgap-high loading synergic effect.

Data Availability
The data used to support the findings of this study are included within the article.

Conflicts of Interest
The authors declare that they have no conflicts of interest.