Enhanced Performance of Carbon–Selenide Composite with La0.9Ce0.1NiO3 Perovskite Oxide for Outstanding Counter Electrodes in Platinum-Free Dye-Sensitized Solar Cells

For large-scale applications, dye-sensitized solar cells (DSSCs) require the replacement of the scarce platinum (Pt)-based counter electrode (CE) with efficient and cheap alternatives. In this respect, low-cost perovskite oxides (ABO3) have been introduced as promising additives to composite-based CEs in Pt-free DSSCs. Herein, we synthesized composites from La0.9Ce0.1NiO3 (L) perovskite oxide and functionalized-multiwall-carbon-nanotubes wrapped in selenides derived from metal-organic-frameworks (f-MWCNT-ZnSe-CoSe2, “F”). L and F were then mixed with carbon black (CB) in different mass ratios to prepare L@CB, F@CB, and L@F@CB composites. The electrochemical analysis revealed that the L@F@CB composite with a mass ratio of 1.5:3:1.5 exhibits better electrocatalytic activity than Pt. In addition, the related DSSC reached a better PCE of 7.49% compared to its Pt-based counterpart (7.09%). This improved performance is the result of the increase in the oxygen vacancy by L due to the replacement of La with Ce in its structure, leading to more active sites in the L@F@CB composites. Moreover, the F@CB composite favors the contribution to the high electrical conductivity of the hybrid carbon nanotube–carbon black, which also offers good stability to the L@F@CB CE by not showing any obvious change in morphology and peak-to-peak separation even after 100 cyclic voltammetry cycles. Consequently, the corresponding L@F@CB-based device achieved enhanced stability. Our work demonstrates that L@F@CB composites with a low cost are excellent alternatives to Pt CE in DSSCs.


Introduction
Dye-sensitized solar cells (DSSCs) are one of the third-generation photovoltaic technologies that have acquired extensive consideration because of their cheap manufacturing cost and potential to reach a power conversion efficiency (PCE) of over 14% up to now [1]. Generally, a DSSC is comprised of four principal parts, namely a photoanode, a dyesensitizer, an electrolyte containing a redox mediator, and a counter electrode (CE) [2]. The

Synthesis of L
The synthesis of L perovskite oxide nanoparticles was performed following previous literature, by a hydrothermal synthesis method using a GNP [19,20]. First, 20 mmol of La(NO 3 ) 3 ·6H 2 O (8.6602 g), 2.2 mmol of Ce(NO 3 ) 3 ·6H 2 O (0.9553 g), and 22 mmol of Ni(NO 3 ) 2 ·6H 2 O (6.3976 g) were added to 50 mL of DI water and stirred for 15 min. Then, 110.5 mmol of aqueous glycine (8.2952 g) was added dropwise to the mixture solution and stirred for 10 min. The mixture was finally adjusted to a pH of 7.7 with 1 M NH 4 OH and stirred for an additional 10 min. When the pH was adjusted, the color of the mixture changed from pale green to an intense blue. Finally, we transferred the mixture into a Teflon-lined autoclave and heated it for 12 h at 180 • C. The obtained pale blue precipitate was washed with water and ethanol several times for 10 min each time, then dried at 80 • C for 12 h. After grinding the product into powder, it was calcinated to 700 • C at 5 • C·min −1 for 2 h in an air atmosphere oven.

Functionalization of MWCNT
The functionalization of MWCNT was performed according to the previous report [29]. In brief, 1.5 g of raw MWCNT was refluxed in a mixture solution of HNO 3 /H 2 SO 4 combined in a 1:3 ratio to create a solution with a final volume of 50 mL. The above-mixed solution was treated by vigorous magnetic stirring at 70 • C for 12 h. The resulting product was cooled and centrifuged at 4000 rpm for 30 min, followed by successive washings with distilled water and ethanol at 4000 rpm for 10 min each time. The final product was dried at 90 • C and named f -MWCNT.

Preparation of f-MWCNT-ZnSe-CoSe 2
For the preparation of f -MWCNT-ZnSe-CoSe 2 , we performed the selenization of f -MWCNT/ZIF-8@ZIF-67 composite based on a previously reported hydrothermal synthesis of CoSe 2 . [27] First, we prepared a selenium (Se) precursor solution by dissolving NaOH (19.2837 g) in 30 mL of deionized (DI) water, then selenium powder Se (1.5427 g) was added and completely dissolved by heating the mixture at 80 • C. Secondly, we dissolved f -MWCNT/ZIF-8@ZIF-67 powder (2.3234 g) in an EDTA-2Na aqueous (aq.) solution (10 mL, 80 mM). After heating at the same temperature both precursor solutions of f -MWCNT/ZIF-8@ZIF-67 and Se, we mixed them, then stirred while heating the mixture for 10 min. Finally, we transferred the mixture into a Teflon-lined autoclave for 12 h heating at 170 • C. After several centrifugations (8000 rpm, 10 min), and washing with water and ethanol of the cooled mixture, a black precipitate of the f -MWCNT-ZnSe-CoSe 2 composite was obtained, dried for 12 h at 60 • C, and named F.

Counter Electrodes' Preparation
For the preparation of F, L, and CB-based CEs, 0.06 g of each compound was separately mixed with 0.02 g of PEG-20000, and then we added 4 mL of ethanol while milling in a mortar to form a paste. As for the preparation of L@F@CB composites based on CEs, we prepared 0.06 g in the mass ratios of 1:4:1, 1.5:3:1.5, and 2:2:2 of L: F: CB in 0.02 g of PEG-20000, corresponding respectively to 1-L@F@CB, 2-L@F@CB, and 3-L@F@CB. Then, each dispersion was spin-coated onto FTO glass substrates and dried for 20 min at 120 • C, followed by heating at 200 • C for 30 min. We also prepared Pt CEs for comparison via the thermal decomposition of H 2 PtCl 6 (10 mM) in ethanol at 380 • C followed by the drop-casting of the Pt solution onto FTO glass and heating at 400 • C for 20 min [30].

Fabrication of TiO 2 Photoanode and Full DSSCs
Based on previous literature, we prepared TiO 2 pastes [31]. First, we washed and rinsed FTO glass substrates in several ultrasonic baths for 15 min each, using a water-Nanomaterials 2022, 12, 961 5 of 18 detergent solution, deionized water, and ethanol, respectively. Then, the cleaned FTO glass was heated to 450 • C on a hot plate, and we sprayed by airbrush a 10% (v/v) alcoholic solution of titanium isopropoxide onto its conductive side, to form a blocking compact layer. This was followed by the screen printing onto the blocking layer of commercial TiO 2 pastes. In brief, the working electrode was made of 5 layers of a transparent TiO 2 paste (18 nm) coated on top of the blocking layer, and 3 layers of scattering TiO 2 paste (400 nm) coated on top of the transparent TiO 2 layers. Then the electrode was annealed at 120 • C for 10 min each time after the printing of each layer of TiO 2 paste. The final TiO 2 films having an active area of 0.16 cm 2 were heated under an airflow oven system by progressive sintering at 325 • C for 5 min, 375 • C for 5 min, 450 • C for 15 min, and 500 • C for 30 min, respectively, and cooled in air at room temperature. After cooling, the electrode was TiCl 4 treated at 70 • C for 30 min in a TiCl 4 (aq.) (20 Mm) bath, then rinsed in deionized water and ethanol, respectively. Before dye loading, we heated the TiO 2 electrode at 500 • C for 30 min, cooled it, and immersed it for 24 h in a dye solution of N719 (0.3 mM). We assembled the DSSCs by clipping the sensitized TiO 2 photoanode onto a prepared CE. The photoanode and CE were separated by a plastic joint spacer of 25 µm thickness and an inner dimension of 5 × 6 mm. Then, the liquid electrolyte was injected into the gap between the photoanode and the CE.

Characterization Methods
X-ray diffraction (XRD) was employed to analyze the structure of our compounds. The measurements were made on a Bruker D8 advance X-ray diffractometer with monochromatized Cu Kα radiation. The chemical states and nature of chemical bonds were analyzed via X-ray photoelectron spectroscopy (XPS), measured at a base pressure of <10 −9 Torr on a VG Multi-Lab 2000 spectrometer. To investigate the surface area and pore size distribution of the prepared nanoparticle powders, we employed the N 2 gas adsorption/desorption method on a Micromeritics ASAP 2020 HD88 system. The surface area was measured using the Brunauer-Emmett-Teller (BET) method, while the pore size distribution was measured using the Barrett-Joyner-Halenda (BJH) method. A field emission scanning electron microscope (FESEM, JSM 7100F), and a high-resolution transmission electron microscope (HRTEM, JEM-2100F) using an energy dispersive spectrum (EDS) analyzer were used to examine the morphology of the different CEs. For the electrochemical analysis, we used a CHI760D electrochemical workstation to perform cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), and Tafel polarization plot measurements. The CV system was made of a reference electrode Ag/Ag + (10 mM AgNO 3 in acetonitrile), a Pt wire counter electrode, and a working electrode composed of the CE to be tested. The three electrodes were immersed in an acetonitrile electrolyte solution of 0.1 M LiClO 4 ·3H 2 O, 10 mM LiI, and 1 mM I 2 , at a scan rate of 50 mV·s −1 and a potential range of ±1.0 V. The EIS measurements were conducted at a 0 V bias voltage using an amplitude of 5 mV and a frequency range from 100 mHz to 100 KHz in dark conditions. The obtained data was analyzed and fitted by Zview−Version 3.5g software, using the related equivalent circuits as models. Moreover, a 10 mV·s −1 scan rate was used to record the Tafel plots. Both the EIS and Tafel samples consisted of two identical CEs and the liquid electrode (CE/electrolyte/CE). The J-V photocurrent density voltage curves of full DSSCs were finally obtained under AM 1.5 G, 0.1 W.cm −2 light radiation intensity, by the Newport solar simulator, USA (450 W Xe source), with a Keithley 2400 source meter.

Characterizations of L and F
FESEM was used to analyze the morphology of the prepared CEs. Figure 1a-c shows that L, F, and CB are made of porous nanocubes, amorphous nanoparticles, and roundgranular nanoparticles, respectively. This is further confirmed in Figure 1f where the 2-L@F@CB composite also contains three kinds of nanoparticles. We also observed the FESEM cross-sectional images in Figure S1 (ESI) which display good adhesion between FTO Nanomaterials 2022, 12, 961 6 of 18 glass substrates and the coated nanoparticles. Furthermore, we could observe the average film thicknesses of 0.34 µm, 2.04 µm, 12.21 µm, 3.66 µm, and 0.44 µm for L, F, CB, 2-L@F@CB and Pt, respectively. Therefore, 2-L@F@CB has a thicker film, which generally contributes to better overall performance as CE in DSSCs [32]. The detailed microstructure of L and F was investigated using HRTEM. As displayed in Figure 2, L and F are made of porous nanocubes and amorphous nanoparticles with the presence of MWCNTs, respectively. The lattice fringes in HRTEM show a fringe spacing of 0.38 nm and 0.19 nm (Figure 2i,j), which can be assigned to rhombohedral LiNO 3 (101) and cubic CeO 2 (220), respectively. Moreover, the fringes of 0.35 nm and 0.25 nm (Figure 2k,l) can be assigned to cubic ZnSe (111) and orthorhombic CoSe 2 (111), respectively. . These patterns indicate that LaNiO3 and CoSe2 with high purity were formed respectively during the synthesis of perovskite oxide and the selenization process. The XRD pattern of 2-L@F@CB confirms that it is a new composite prepared from the synthesized L and F, which did not display any obvious peak shifts in the pattern.   An EDS analysis was performed to evaluate the distribution of different atoms in the L and F nanostructures. The EDS mapping in Figure S2a-f displays well-distributed atoms of La, Ce, Ni, and O over the entire structure of L nanocubes. As for the EDS mapping in Figure S2g-l, it displays a homogeneous distribution over the entire structure of F of Co, Se, and C atoms, but not for Zn. Indeed, the lower melting point (419.5 • C) of Zn metal compared to Co metal (1495.1 • C) gives rise to its phase-separation from other elements during the selenization process, causing the separation of ZnSe in the composite [27].
The BET and BJH were used to assess the surface area and pore size distribution of the nanoparticles. In Figure 3a,b, each prepared sample displays a standard type-IV isotherm with an H3 hysteresis loop at high relative pressure (above P/P 0 = 0.9). This indicates the mesoporous nature of our samples [33]. Furthermore, the BET surface areas were calculated to be 15.33, 20.90, 421.33, 211.95, 213.18, and 119.29 m 2 . g −1 , respectively, for L, F, CB, L@CB, F@CB, and 2-L@F@CB (Table 1). We can observe an improvement in the surface areas of F@CB and 2-L@F@CB composites compared to the surface areas of L and F. Moreover, Figure 3c,d shows pore size diameters mostly ranging between 2 and 50 nm for the prepared samples, which confirms their mesoporous structure. High surface area porous materials give rise to higher electrocatalytic activity, which is of benefit for CE applications in DSSCs [34].

Measurements of Electrochemical Properties of CEs
We then assessed the electrocatalytic activities of various CEs of Pt, L, F, CB, L F@CB, 1-L@F@CB, 2-L@F@CB, and 3-L@F@CB for triiodide ion (I ) reduction by th technique (Figure 4a). For each sample, we can see two pairs of standard oxidatio duction current density peaks, respectively named (Ox1, Red1) for the lower pot region related to the chemical reaction in Equation (1, and (Ox2, Red2) for the high p tial region corresponding to the chemical reaction in Equation (2).
The redox peaks located at a more negative potential allow assessment of the ele  XPS was employed to study the nature of chemical bonds and chemical states of L, F, and 2-L@F@CB ( Figure S3). The survey spectrum in Figure S3a reveals peaks such as Zn 2p, Ce 3d, Ni 2p, La 3d, Co 2p, O 1s, C 1s, Se 3d, and Co 3p. The spectra of Zn 2p ( Figure S3b) display two strong peaks located at~1044 eV and~1021 eV, related to the binding energies of Zn 2p 1/2 and Zn 2p 3/2 , respectively. These two strong peaks are consistent with Zn 2+ in the binding energies of Zn-Se. [35] The Ce 3d spectra in Figure S3c present for 2-L@F@CB and L, different peaks located at~915 eV,~899.7 eV (Ce 3+ :~904.4 eV, and Sat.:~896.18 eV), and~880.4 eV (Sat.:~886.4 eV). These peaks can be assigned to the binding energies of Ce 4+ 3d 3/2 , Ce 3+/4+ 3d 5/2 , and Ce 3+ 3d 5/2 found in CeO x , respectively [36]. The Ni 2p spectra in Figure S3d display peaks at~871 eV and~858.6 eV (Sat.:~865.4 eV and~863.9 eV) for 2-L@F@CB, and peaks at~870.5 eV and~856.3 eV (Sat. 864.9 eV and 861 eV) for L. In both samples, the two categories of peaks can be ascribed respectively to the characteristics of Ni 2+/3+ for Ni 2p 1/2 and Ni 2p 3/2 occurring in LaNiO 3 [37]. We can also observe half of the La 3d spectrum in Figure S3d located at~849.5 eV (Sat.~853.5 eV) and corresponding to the characteristics of La 3+ for La 3d 5/2 occurring in LaNiO 3 [38]. The La 3d spectra in Figure S3e display peaks located at~849.5 eV (sat.~853.5 eV), and~833 eV (sat.~837 eV) which can be ascribed to the binding energy of La 3d 5/2 and La 3d 3/2 , respectively. Both types of peaks are consistent with the results for La 3+ in a La (III) oxidation state. [38] The spectra of Co 2p ( Figure S3f) is made up of two spin-orbit doublets and two shake-up satellites (Sat.). In F and 2-L@F@CB, the peaks located at~796 eV with shoulders at~802 eV (Sat.) can be ascribed to Co 2p 1/2 , while the peaks found at~781 eV with a shoulder at~785 eV (Sat.) can be assigned to Co 2p 3/2 . Both kinds of peaks match well with the characteristics of Co 2+ in CoSe 2 [39]. Furthermore, we can observe in F, two weak peaks at 793.2 eV and~776.5 eV which correspond to Co 3+ of Co 2p 1/2 and Co 2p 3/2 , respectively. The presence of Co 3+ peaks in the Co 2p spectrum of this sample is due to the partial surface oxidation under an air atmosphere. However, the main valence state of Co in both samples is the divalent metal cation. As for the shake-up satellites at higher energy, it is due to the antibonding orbital occurring between Se and Co atoms [40]. Figure S3g [15,16]. The spectra of C 1s ( Figure S3h) reveal for F and 2-L@F@CB peaks at around 290 eV, 287 eV, 285.3 eV (shifting to~283.8 eV in 2-L@F@CB), and 283.9 eV (shifting to~282.8 eV in 2-L@F@CB) corresponding to O-C=O, C-O, C-C, and C=C bonding, respectively. Furthermore, we can observe at~282 eV a supplementary peak visible in F and corresponding to C-metal (C-Zn/Co). The peaks related to C=C bonds display quite a high intensity, while those corresponding to C=O bonds show low intensity. This reveals the formation of graphitic carbon during the selenization process of F [27]. In Figure S3i, we can observe for F and 2-L@F@CB, two kinds of deconvoluted peaks at 55 eV and~54.3 eV (shifting to 53.2 eV in 2-L@F@CB), matching well with Se 3d 3/2 and Se 3d 5/2 , respectively. These peaks are both consistent with the metal-Se interaction occurring in CoSe 2 . As for the broad peaks located within 62-57 eV, they are consistent with Se-O bonding on the surface, and Co 3p [27]. Therefore, the 2-L@F@CB composite is made up of the same chemical bonds as those observed in L and F, with very slight peak shifts due to changes in the chemical environment. Moreover, we could observe in 2-L@F@CB lower peak intensities compared to L and F for almost all peaks except for C 1s (especially C=C) and O 1s (especially O 2− 2 /O). The increase in the intensity of C=C can be attributed to the graphitic carbon-rich environment favored by F and CB in the 2-L@F@CB composite. Moreover, the increase in the intensity of O 2− 2 /O ( Figure S4) in 2-L@F@CB can be ascribed to the contribution of L to increasing the oxygen vacancy percentage, made possible thanks to the replacement of La sites with Ce in its structure. Abundant oxygen vacancies lead to more active sites, giving rise to minimized efficiency losses in DSSCs [17,21]. Figure S5a . 01-075-1621). These patterns indicate that LaNiO 3 and CoSe 2 with high purity were formed respectively during the synthesis of perovskite oxide and the selenization process. The XRD pattern of 2-L@F@CB confirms that it is a new composite prepared from the synthesized L and F, which did not display any obvious peak shifts in the pattern.

Measurements of Electrochemical Properties of CEs
We then assessed the electrocatalytic activities of various CEs of Pt, L, F, CB, L@CB, F@CB, 1-L@F@CB, 2-L@F@CB, and 3-L@F@CB for triiodide ion (I − 3 ) reduction by the CV technique (Figure 4a). For each sample, we can see two pairs of standard oxidationreduction current density peaks, respectively named (Ox1, Red1) for the lower potential region related to the chemical reaction in Equation (1, and (Ox2, Red2) for the high potential region corresponding to the chemical reaction in Equation (2).  An EIS was also carried out to deeply study the electrocatalytic activity of CEs from symmetric dummy cells made of two identical CEs. Figure 4b presents the Nyquist plots for our CEs. We can notice that Pt, F, and L-based CEs show two semicircles, while CB, F@CB, L@CB and all L@F@CB-based CEs display three semicircles, resulting from their porous nature [27]. In standard Nyquist plots, the series resistance (Rs) is the intercept of the first semicircle with the horizontal axis; the charge-transfer resistance (Rct) and the capacitance of the CE/electrolyte interface are linked to the first semicircle (high frequency region). As for the middle semicircle (related to the middle frequency), it is linked to the adsorption of iodine. Finally, the last semicircle (low frequency region) is related to the Nernst diffusion impedance of the triiodide/iodide redox couple within a thin layer of the electrolyte (W) [27]. We obtained the EIS data by fitting the Nyquist plots with an equivalent circuit (Figure S7), and the related data are listed in Table 2. The Rs values for F, The redox peaks located at a more negative potential allow assessment of the electrocatalytic performance and can be assigned to the chemical reaction of Equation (1) [41]. Therefore, we examined the electrocatalytic parameters such as peak-to-peak separation (∆E pp ) and reduction peak current density (J Red1 ) only for the low potential region. ∆E pp is directly linked to the kinetic redox capability of the CE in the I − /I − 3 redox couple. The smaller the ∆E pp value, the higher the electrocatalytic ability of the corresponding CE [42]. Table 2 summarizes the main CV parameters obtained for our different CEs. ∆E pp values show the trend of 1-L@F@CB (389 mV) < F (413 mV) < 2-L@F@CB (414 mV) < 3-L@F@CB (470 mV) < L (480 mV) < F@CB (511 mV) < Pt (604 mV) < L@CB (607 mV) < CB (647 mV). In this trend, L@F@CB composites show higher electrocatalytic activities, which can be related to the higher electrocatalytic of both F composites and L, and the excellent electrical conductivity of CB [43]. We can also observe for the reduction peak current density (|J Red1 |) the following trend of L@CB (1.104 mA·cm −2 ) > CB (1.073 mA·cm −2 ) > F@CB (0.921 mA·cm −2 ) > 1-L@F@CB (0.735 mA·cm −2 ) > 2-L@F@CB (0.730 mA·cm −2 ) > F (0.606 mA·cm −2 ) > 3-L@F@CB (0.561 mA·cm −2 ) > Pt (0.551 mA·cm −2 ) > L (0.004 mA·cm −2 ). These values suggest that L@F@CB composites have high reduction ability, in good agreement with the results of ∆E pp . Furthermore, based on the trends of ∆E pp and |J Red1 |, it can be concluded that 1-L@F@CB and 2-L@F@CB demonstrate better electrocatalytic activity than Pt. Moreover, we assessed the ratio J Ox1 /|J Red1 | for the low potential region, very relevant for the catalytic activity and gives information about the reversibility of the redox reaction [27]. We can see from Table 2 that the J Ox1 /|J Red1 | for 2-L@F@CB is closer to 1 than all the other samples, including Pt (1.624), indicating a more reversible reaction for 2-L@F@CB [44]. We also analyzed, for the 2-L@F@CB-based CE, the CV plots obtained for various scan rates and noticed that increasing the scanning speed leads to an increase in the anodic and cathodic peak current densities. Consequently, we found a direct link between the peak current densities and the square root of the scan rate, which means that no chemical reaction occurs between the redox couple and the 2-L@F@CB-based CE ( Figure S6a,b). Furthermore, a direct connection indicates that the cathodic/anodic reactions have restricted the diffusion of the flow of iodide species in the cell toward the CE. This results from the narrowing of the diffusion layer occurring with the increase of the scanning speed and the electrochemical polarization, giving rise to high over-potential and reduced reversibility [27]. In addition, we also investigated the electrochemical stability of 2-L@F@CB-based CE through consecutive CV measurements at the scan rate of 50 mV.s −1 (Figure S6c,d). Indeed, electrochemical stability is highly important for the potential application of the CE in DSSCs. Even after 100 cycles, no obvious change was observed in the morphology of J Red1 and ∆E pp . The stability of the 2-L@F@CB-based CE is the result of the high surface area and good conductivity of F, L, and CB. Furthermore, the binding properties of CB contribute to the strong adhesion of this CE.
An EIS was also carried out to deeply study the electrocatalytic activity of CEs from symmetric dummy cells made of two identical CEs. Figure 4b presents the Nyquist plots for our CEs. We can notice that Pt, F, and Lbased CEs show two semicircles, while CB, F@CB, L@CB and all L@F@CB-based CEs display three semicircles, resulting from their porous nature [27]. In standard Nyquist plots, the series resistance (R s ) is the intercept of the first semicircle with the horizontal axis; the charge-transfer resistance (R ct ) and the capacitance of the CE/electrolyte interface are linked to the first semicircle (high frequency region). As for the middle semicircle (related to the middle frequency), it is linked to the adsorption of iodine. Finally, the last semicircle (low frequency region) is related to the Nernst diffusion impedance of the triiodide/iodide redox couple within a thin layer of the electrolyte (W) [27]. We obtained the EIS data by fitting the Nyquist plots with an equivalent circuit ( Figure S7), and the related data are listed in Table 2 170 Ω). We can observe that upon the addition of F and L to CB, the R ct value strongly decreases gradually from 12.550 Ω, 16.170 Ω and 4.996 Ω, respectively, for F, L, and CB, to 0.685 Ω and 0.848 Ω for 1-L@F@CB and 2-L@F@CB. These values are significantly lower than the 3.004 Ω obtained for Pt CE. Indeed, the lower the R ct value, the faster the electron transfer from the surface of the CE to the electrolyte, leading to less interfacial recombination [45]. Therefore, 1-L@F@CB and 2-L@F@CB have the best catalytic activity among our CEs, in confirmation of the previous CV analysis.
Tafel polarization was also investigated to determine the catalytic activity of various CEs. A typical Tafel polarization plot has three regions: the polarization region (|V| < 120 mV), the Tafel region (120 mV < |V| < 400 mV), and the diffusion region (|V| > 400 mV). For investigation of the catalytic performance of various CEs, we studied the diffusion region. The corresponding two key parameters, namely the exchange current density (J 0 ) and the limiting diffusion current density (J lim ), are connected to the catalytic ability of each sample and can be obtained, respectively, from the Tafel and diffusion regions. Moreover, J 0 and J lim can also be obtained by calculation using Equations (3) and (4) [27].
where R stands for the gas constant, T for the temperature in Kelvin, F for the Faraday constant, n for the number of electrons transferred per chemical reaction between the electrolyte and the CE interface (n = 2), Dn for the diffusion coefficient of I − 3 , l for the thickness of the diffusion layer, and C for the concentration of I − 3 . Moreover, Dn can be further obtained from the Randles-Sevcik theory [46], (Equation (5), with J red corresponding to the peak current density, K = 2.69 × 10 5 , A being the active area of the symmetric cell, and υ representing the applied scan rate of CV plots. J red = Kn 1.5 ACD 0.5 n υ 0.5 In Table 2, we can see a noticeable increase in the J 0 values after the mixing of L and F into CB. These values increase from 0.094, 7.656 and 9.484 mA·cm −2 for L, F, and CB, respectively, to 23.768 and 22.646 mA·cm −2 for 1-L@F@CB and 2-L@F@CB, respectively. Both values are superior to the J 0 value of 19.275 mA·cm −2 obtained for the Pt-based CE. In addition, the small increase in the J 0 value observed from 2-L@F@CB (22.646 mA·cm −2 ) to 1-L@F@CB (23.768 mA·cm −2 ) indicates that 2-L@F@CB is the best CE as mentioned in the previous CV investigation. Moreover, Table 2  > Pt (1.049 × 10 −7 cm −2 ·s −1 ) > L (5.528 × 10 −12 cm −2 ·s −1 ). We can see the high Dn values of 1-L@F@CB (1.866 × 10 −7 cm −2 ·s −1 ), and 2-L@F@CB (1.841 × 10 −7 cm −2 ·s −1 ) CEs compared to Pt (1.049 × 10 −7 cm −2 ·s −1 ) CE. This indicates the high velocity of these materials, which is advantageous for improving the photovoltaic performance of the DSSCs. Table 3 presents the photovoltaic parameters of various CEs in DSSCs, and the corresponding J-V plots are visible in Figure 4d. We can observe that when used separately as CEs for DSSCs, F and L allow the achievement of PCEs of 4.73% and 0.42%, respectively. These values were significantly improved with the L@F@CB composites. Indeed, when properly adjusting the mass ratio (1.5:3:1.5) of L:F:CB, the PCE of the corresponding DSSC (2-L@F@CB) significantly increases, up to 7.49%. This value is superior to the value obtained for the Pt-based DSSC (7.09%) counterpart, leading to better electrocatalytic activity in the composite. These outcomes were first favored by the increase in the oxygen vacancy percentage offered by L thanks to the replacement of La sites with Ce. This gives rise to more active sites in the final L@F@CB composite. Secondly, this high electrocatalytic activity and good performance result from the joint benefits of L, F, and CB holding high electrocatalytic activities. Moreover, this achievement is justified by the high electrical conductivity of the carbon-based materials created by the F@CB composite, which boosts the conductivity with the contribution of the hybrid CNT@CB [23]. Finally, the strong adhesion of CB with FTO glass contributes, in agreement with the above CV stability test results, to the stability of the 2-L@F@CB based DSSC. All this leads to lower R ct and lower diffusion resistance in the 2-L@F@CB based CE, confirming the previous morphological and electrochemical analysis. However, the lowering of PCE observed in the 3-L@F@CB composite-based DSSC is the result of the agglomeration of particles in the composite, giving rise to reduced performance.

Device Stability
We also examined for 360 h (15 days) the device stability of the DSSC made of 2-L@F@CB CE. The DSSC was kept in dark conditions after each IV measurement, and the photovoltaic performance was tested under 100 mA cm −2 illumination at regular intervals of time. Figure 5 shows the photovoltaic parameters recorded for the analyzed time frame. At the 360th h (15th day), the DSSC shows a PCE of 6.50% with 14.25 mA·cm −2 (Jsc), 0.721 V (Voc), and 62.93 (FF). This result indicates about a 13% decrease in the PCE after 360 h of testing, which is quite good performance under laboratory conditions. These results also confirm the performance of the 2-L@F@CB CE for DSSCs. photovoltaic performance was tested under 100 mA cm −2 illumination at regular intervals of time. Figure 5 shows the photovoltaic parameters recorded for the analyzed time frame. At the 360th h (15th day), the DSSC shows a PCE of 6.50% with 14.25 mA·cm −2 (Jsc), 0.721 V (Voc), and 62.93 (FF). This result indicates about a 13% decrease in the PCE after 360 h of testing, which is quite good performance under laboratory conditions. These results also confirm the performance of the 2-L@F@CB CE for DSSCs.

Effect of 2-L@F@CB Film Thickness on DSSCs' Performance
2-L@F@CB composite CEs with various film thicknesses ranging from 3.6 to 10.4 µm were prepared to investigate their effects on DSSCs' performance. We also prepared Pt CE for comparison. Figure 6a shows the CV curves of 2-L@F@CB CEs with different film thicknesses. As displayed in Table S1, the peak current density ranges from 0.730 to 1.470 mA·cm −2 , with the corresponding peak-to-peak separation also increasing from 414 to 787 mV, respectively, for 3.6 and 10.4 µm thickness. These results suggest that despite the increase in surface area, thicker 2-L@F@CB films give rise to lower catalytic activity, which can be due to the increase in internal resistance [25,47]. We also recorded the EIS data for 2-L@F@CB CEs with the various film thicknesses. Figure 6b shows the Nyquist plots with the corresponding parameters listed in Table S1. There is no obvious variation in the Rs values for all 2-L@F@CB CEs with different thicknesses. However, there is a significant increase in the R ct values from 0.848 to 6.357 Ω, respectively, for 3.6 and 10.4 µm film thickness. This increase in R ct resistance indicates a significant lowering of the electrocatalytic activity with the increase of 2-L@F@CB film thickness, in agreement with the previous CV analysis. Tafel polarization curves were also analyzed as shown in Figure 6c, and the resultant parameters are listed in Table S1. The J o and J lim values of the 2-L@F@CB CEs greatly decrease with the increase in film thickness. These results indicate an increase in the diffusion resistance of the redox couple in the electrolyte with 2-L@F@CB film thickness. Therefore, the DSSCs made of 2-L@F@CB CEs with thinner films are expected to show optimum performance comparable or higher than Pt-based cells. We finally investigated the effect of 2-L@F@CB CE film thicknesses on the performance of DSSCs as shown in Figure 6d, and the corresponding photovoltaic parameters are reported in Table S2. We could observe a slight decrease of about 12% (from 7.49% to 6.55%) in the PCE with the increase in film thickness from 3.6 to 10.4 µm. This outcome can be attributed to the slight lowering in J SC and FF values resulting from the increase in internal resistance as demonstrated in the previous analysis. Therefore, the optimum efficiency was achieved for DSSCs prepared with a 3.6 µm film thickness.

Discussion on the Potential Cost of Prepared CEs' Materials
In this section, we compare the potential prices of the prepared CEs' materials. Although the price per gram of Pt is always fluctuating because of widespread demand, it is generally high as given in Table S3. In Figure S8a, we can observe the very low cost of the materials involved in our work compared to Pt, and this leads to L@F@CB composites being cheaper than Pt, with a cost-saving percentage of more than 30% ( Figure S8b). This further confirms the potential of prepared L@F@CB CEs over Pt CE.

Conclusions
L perovskite oxide, F, and L@F@CB composites have been prepared with success and analyzed for the first time as CEs for DSSCs. L perovskite oxide was synthesized using the GNP method, while F was obtained from the hydrothermal selenization of f-MWCNT/ZIF-8@ZIF-67. The electrochemical analysis reveals that the optimized L@F@CB CE exhibits better electrocatalytic activity than the Pt CE. Moreover, the related DSSCs achieved a superior PCE of 7.49% than the corresponding Pt-based DSSC (7.09%) prepared under similar conditions. The enhanced performance is due to the increase in the oxygen vacancy percentage offered by L thanks to the replacement of La sites with Ce. This gives rise to more active sites in the final L@F@CB composite. Moreover, the F@CB composite favors the contribution to the high electrical conductivity of the hybrid carbon nanotube-carbon black (CNT@CB), which also offers stability to the L@F@CB CE. Consequently, the corresponding L@F@CB-based device achieved enhanced stability. Our work demonstrates that L@F@CB composites with a low cost are excellent alternatives to Pt CE in DSSCs.

Discussion on the Potential Cost of Prepared CEs' Materials
In this section, we compare the potential prices of the prepared CEs' materials. Although the price per gram of Pt is always fluctuating because of widespread demand, it is generally high as given in Table S3. In Figure S8a, we can observe the very low cost of the materials involved in our work compared to Pt, and this leads to L@F@CB composites being cheaper than Pt, with a cost-saving percentage of more than 30% ( Figure S8b). This further confirms the potential of prepared L@F@CB CEs over Pt CE.

Conclusions
L perovskite oxide, F, and L@F@CB composites have been prepared with success and analyzed for the first time as CEs for DSSCs. L perovskite oxide was synthesized using the GNP method, while F was obtained from the hydrothermal selenization of f-MWCNT/ZIF-8@ZIF-67. The electrochemical analysis reveals that the optimized L@F@CB CE exhibits better electrocatalytic activity than the Pt CE. Moreover, the related DSSCs achieved a superior PCE of 7.49% than the corresponding Pt-based DSSC (7.09%) prepared under similar conditions. The enhanced performance is due to the increase in the oxygen vacancy percentage offered by L thanks to the replacement of La sites with Ce. This gives rise to more active sites in the final L@F@CB composite. Moreover, the F@CB composite favors the contribution to the high electrical conductivity of the hybrid carbon nanotube-carbon black (CNT@CB), which also offers stability to the L@F@CB CE. Consequently, the corresponding L@F@CB-based device achieved enhanced stability. Our work demonstrates that L@F@CB composites with a low cost are excellent alternatives to Pt CE in DSSCs.
Author Contributions: A.R.T. contributed to drafting the manuscript, performing electrochemical testing, and analyzing some testing data. W.X. contributed to drafting the manuscript and analyzing all testing data. S.W., B.L. and Q.L. contributed to carrying out the synthesis of L and F. M.Z. contributed to the XPS testing. M.G., F.V. and J.W. contributed to the synthesis of L. A.T. contributed to the CV analysis and drafted the manuscript. X.Z. contributed to drafting the manuscript and analyzing all testing data. All authors have read and agreed to the published version of the manuscript.