Organic Dyes in Dye-Sensitized Solar Cells Featuring Back Reﬂector

: Dye-sensitized solar cells (DSSCs) were fabricated using a photoelectrode covered by a porous layer of titanium dioxide, platinum counter electrode, iodide/triiodide electrolyte and three different dyes: phenylﬂuorone (PF), pyrocatechol violet (PCV) and alizarin (AL). After the adsorption of the dyes on the mesoporous TiO 2 layer, the measurement of absorption spectra of all the tested dyes revealed a signiﬁcant broadening of the absorption range. The positions of highest occupied molecular orbital (HOMO) and lowest occupied molecular orbital (LUMO) levels of dye molecules were determined, indicating that all three dyes are good candidates for light harvesters in DSSCs. The cells were tested under simulated solar light, and their working parameters were determined. The results showed that the implementation of the back reﬂector layer made of BaSO 4 provided an improvement in the cell efﬁciency of up to 17.9% for phenylﬂuorone, 60% for pyrocatechol violet and 21.4% for alizarin dye.


Introduction
The usage of solar energy for production of electricity implies a constant increase in investment in the investigations of photovoltaic (PV) technology. Despite the strong position of silicon solar cells in the market, much attention is attracted by the emerging photovoltaic technologies such as organic photovoltaics (OPV) and dye-sensitized solar cells (DSSCs), as well as perovskite cells (PSCs) that stemmed from DSSCs. Dye-sensitized photovoltaic devices are well known for low-cost production, beneficial energy payback time, easy recycling and simple structure. A dye cell is essentially composed of an anode covered by titanium dioxide nanoparticles with adsorbed photoactive dye molecules, counter electrode and electrolyte inserted in between [1]. The illumination of the photoanode is followed by the excitation of a dye molecule and the injection of the excited electron into the conduction band of titanium dioxide. The electrons can travel through the mesoporous layer consisting of TiO 2 nanoparticles and then by the external circuit to the counter electrode. Here, the electrons reduce the electrolyte, which in turn regenerates the oxidized dye molecules. In the whole operation cycle, the crucial process is the electron transfer from the dye molecule to the TiO 2 conduction band [2,3]. The mechanism of this process depends on the configuration and electronic structure of the adsorbed dyes [4]. The dye is the core component of a DSSC that sensitizes TiO 2 by broadening the spectrum of absorbed light from UV to NIR since the TiO 2 absorbs only UV light. In order to enhance the light harvesting, it is desirable that the absorption spectrum of the dye adsorbed on TiO 2 covers a broad spectral range. The highest occupied molecular orbital (HOMO) energy level of the dye molecule should be lower than the redox potential of the electrolyte, while the lowest unoccupied molecular orbital (LUMO) should be located higher than the edge of the semiconductor conduction band.
Good DSSC performance, confirmed by the efficiency of over 10% [1], was exhibited by the cells sensitized with ruthenium complexes (e.g., N3, N719 dyes); however, their Good DSSC performance, confirmed by the efficiency of over 10% [1], was exhibited by the cells sensitized with ruthenium complexes (e.g., N3, N719 dyes); however, their application is limited due to the scarcity of rare transition elements, high cost of these compounds and environmental hazard.
Another group of sensitizers, characterized by tunable optical, photophysical and electrochemical properties, are porphyrins, which contain an inexpensive metal and are constructed in D-π-A (donor-π-conjugated linker-acceptor) configuration, aiming for efficient intramolecular charge transfer and broad light absorption. The investigations of porphyrin dyes, both theoretical and experimental [5][6][7][8], resulted in an efficiency of 13% with cobalt (II/III) redox shuttle [9].
In recent years, natural dyes such as flavonoids, betalains, tannins, chlorophylls and carotenoids extracted from plants have also been successfully used as light harvesters in DSSCs [19][20][21][22]. Their advantages over synthetic dyes include cost-effectiveness, nontoxicity and possible biodegradation.
This work focuses on three organic metal-free dyes (see Figure 1), namely phenylfluorone (PF), pyrocatechol violet (PCV) and alizarin (AL), and their application in dye cells as photosensitizers. Phenylfluorone (9-phenyl-3-fluorone, fluorone black) consists of fluorine with a phenyl group (Figure 1). Phenylfluorone and its derivatives are widely used as analytical agents in spectrophotometry and fluorescent spectrometry for the detection of metal ions [23]. Pyrocatechol violet is a sulfone phthalein dye, which is one of the organic reagents recommended for the studies of complexation as a ligand with metal ions such as indium and gallium toward the applications in nuclear medicine as well as in metal extraction and determination studies [24]. Alizarin (dihydroxyanthraquinone) is extracted from plant roots but can also be synthesized, and it is usually used as a textile red dye and staining agent in medicine. The application of alizarin dye and its derivatives in DSSCs was investigated theoretically [2,25]. The enediol group of the catechol ring occurring in all the structures of the dyes studied in this work is known for making a chelating bond with titanium in TiO2, giving rise to a new absorption band corresponding to the charge transfer mechanism with very fast electron injection to the TiO2 semiconductor conduction band [2,[26][27][28]. Recently, it The enediol group of the catechol ring occurring in all the structures of the dyes studied in this work is known for making a chelating bond with titanium in TiO 2 , giving rise to a new absorption band corresponding to the charge transfer mechanism with very fast electron injection to the TiO 2 semiconductor conduction band [2,[26][27][28]. Recently, it was demonstrated that upon photoexcitation, the presence of a catechol unit enhances the electron injection to the TiO 2 semiconductor [29].
In addition to the investigation of the aforementioned dyes as photosensitizers in DSSCs, this work presents the influence of the back reflective layer (RL) on the performance was demonstrated that upon photoexcitation, the presence of a catechol unit enhances the electron injection to the TiO2 semiconductor [29].
In addition to the investigation of the aforementioned dyes as photosensitizers in DSSCs, this work presents the influence of the back reflective layer (RL) on the performance of the prepared dye cells, with the aim to exploit the transmitted light (crossing the cell). In this study, barium sulfate (BaSO4) was implemented as the reflective layer on the back side of the cell, with diffuse reflectance exceeding 95% in the wavelength of 200-1800 nm [30]. This kind of material scatters light so diffuse photons can be trapped inside the solar cell structure, which is advantageous for the cell operation.

Materials and Methods
All chemicals were purchased commercially and used without further purification. The glass electrodes with fluorine tin oxide (FTO) conductive coverage, the counter electrodes covered with platinum black and the I -/I3 -liquid electrolyte were purchased from the Greatcell Solar company. The glass FTO electrodes of 2 × 2.5 cm size and 18 Ω/sq sheet resistance were cleaned in acetone and ethanol, dried in nitrogen gas flow and employed as substrates for the layer of the titanium dioxide paste. The preparation of paste included several steps [31] in which the TiO2 nanoparticles P25 from Degussa were used. The TiO2 paste was deposited on glass electrodes with doctor blade method and then annealed at 450 °C to sinter the nanoparticles. The electrodes covered with TiO2 were immersed in 2 mM solutions of phenylfluorone, pyrocatechol violet, and alizarin dyes in absolute ethanol for 1, 2, 24 or 48 h in darkness at room temperature, and then the sensitized electrodes were rinsed with absolute ethanol. The TiO2 working electrode and counter electrode were assembled with Surlyn separator and sealed in a Gestigkeit PZ 28-3T (Inverness, UK) high-temperature titanium hot plate with a lid. The electrolyte was introduced through the holes in the Pt-coated electrode. The active area of the cells was 0.8 cm 2 . Figure 2 shows the schematic illustration of the dye cell. The cells were examined in a solar light simulator (Abet Technologies SUN 3000, Sydney, Australia) under 100 mW/cm 2 illumination, and I-V curves were obtained on a Keithley 2440 Source Measure Unit. The measurements were repeated 3-6 times for a minimum of 3 cells of the same type. The absorption and reflectance spectra of dyes or electrodes were measured by using a Thermo Scientific Evolution 220 UV-Vis (Darmstadt, Germany) spectrophotometer equipped with ISA-220 integrating sphere. The morphology of the TiO2 layer was determined by using a Hitachi SU 3500 scanning electron microscope (SEM).  In order to investigate the effect of the back reflective layer, the prepared semitransparent cells were placed on a BaSO 4 powder layer, and the measurements of I-V curves were performed under the simulated light.

Characterization of Photoelectrode
The morphology of the TiO 2 nanoparticle layer deposited on the glass electrode and sintered at high temperature is shown in Figure 3a. The SEM image reveals the mesoporous structure of the layer, which is essential to ensure penetration of the dye solution and electrolyte between nanoparticles. The coverage of TiO 2 by the adsorbed dye is visible in Figure 3b.
In order to investigate the effect of the back reflective layer, the prepared semitransparent cells were placed on a BaSO4 powder layer, and the measurements of I-V curves were performed under the simulated light.

Characterization of Photoelectrode
The morphology of the TiO2 nanoparticle layer deposited on the glass electrode and sintered at high temperature is shown in Figure 3a. The SEM image reveals the mesoporous structure of the layer, which is essential to ensure penetration of the dye solution and electrolyte between nanoparticles. The coverage of TiO2 by the adsorbed dye is visible in Figure 3b.  The TiO2 layer exhibits the ability to scatter the incident light, which is beneficial for enhancing the light harvesting. The diffuse reflectance of TiO2 layer shown in Figure 4 achieves over 40% in the range of 400-530 nm. Immersion of fabricated photoelectrodes in solutions of dyes results in the intensive tint of the titanium dioxide layer, which is visible in Figure 5. Each electrode shown in Figure 5 was immersed in a dye solution for the optimal time selected for a given dye, leading to the best photovoltaic performance (v.i.).
In order to investigate the effect of the back reflective layer, the prepared semitransparent cells were placed on a BaSO4 powder layer, and the measurements of I-V curves were performed under the simulated light.

Characterization of Photoelectrode
The morphology of the TiO2 nanoparticle layer deposited on the glass electrode and sintered at high temperature is shown in Figure 3a. The SEM image reveals the mesoporous structure of the layer, which is essential to ensure penetration of the dye solution and electrolyte between nanoparticles. The coverage of TiO2 by the adsorbed dye is visible in Figure 3b.  The TiO2 layer exhibits the ability to scatter the incident light, which is beneficial for enhancing the light harvesting. The diffuse reflectance of TiO2 layer shown in Figure 4 achieves over 40% in the range of 400-530 nm. Immersion of fabricated photoelectrodes in solutions of dyes results in the intensive tint of the titanium dioxide layer, which is visible in Figure 5. Each electrode shown in Figure 5 was immersed in a dye solution for the optimal time selected for a given dye, leading to the best photovoltaic performance (v.i.). Immersion of fabricated photoelectrodes in solutions of dyes results in the intensive tint of the titanium dioxide layer, which is visible in Figure 5. Each electrode shown in Figure 5 was immersed in a dye solution for the optimal time selected for a given dye, leading to the best photovoltaic performance (v.i.).

Photoelectrical Properties of the Dyes
The sensitizing compound used in a DSSC should absorb light, preferably in a broad range of the visible spectrum. In order to verify the coverage of the light spectrum by the investigated dyes, especially after the process of adsorption on TiO2 nanoparticles, the absorption measurements were performed. The absorption spectra in Figure 6 show that free PF dye solution in ethanol exhibits only a narrow absorption band around 520 nm; however, upon adsorption on TiO2 nanostructural coverage of electrodes, a significant broadening of the absorption range occurs. In consequence, the absorption covers a variety of wavelengths from around 400 to 600 nm. Below 400 nm, the absorption of bare TiO2 is observed. The absorption spectrum of PCV in free form in an ethanol solution consists of two bands: one with a maximum around 450 nm and the second, much weaker, band at 575 nm ( Figure 7). The adsorption of PCV on TiO2 leads to the broadening of the absorption range that covers the visible spectrum nearly to 700 nm.

Photoelectrical Properties of the Dyes
The sensitizing compound used in a DSSC should absorb light, preferably in a broad range of the visible spectrum. In order to verify the coverage of the light spectrum by the investigated dyes, especially after the process of adsorption on TiO 2 nanoparticles, the absorption measurements were performed. The absorption spectra in Figure 6 show that free PF dye solution in ethanol exhibits only a narrow absorption band around 520 nm; however, upon adsorption on TiO 2 nanostructural coverage of electrodes, a significant broadening of the absorption range occurs. In consequence, the absorption covers a variety of wavelengths from around 400 to 600 nm. Below 400 nm, the absorption of bare TiO 2 is observed.

Photoelectrical Properties of the Dyes
The sensitizing compound used in a DSSC should absorb light, preferably in a broad range of the visible spectrum. In order to verify the coverage of the light spectrum by the investigated dyes, especially after the process of adsorption on TiO2 nanoparticles, the absorption measurements were performed. The absorption spectra in Figure 6 show that free PF dye solution in ethanol exhibits only a narrow absorption band around 520 nm; however, upon adsorption on TiO2 nanostructural coverage of electrodes, a significant broadening of the absorption range occurs. In consequence, the absorption covers a variety of wavelengths from around 400 to 600 nm. Below 400 nm, the absorption of bare TiO2 is observed. The absorption spectrum of PCV in free form in an ethanol solution consists of two bands: one with a maximum around 450 nm and the second, much weaker, band at 575 nm ( Figure 7). The adsorption of PCV on TiO2 leads to the broadening of the absorption range that covers the visible spectrum nearly to 700 nm. The absorption spectrum of PCV in free form in an ethanol solution consists of two bands: one with a maximum around 450 nm and the second, much weaker, band at 575 nm ( Figure 7). The adsorption of PCV on TiO 2 leads to the broadening of the absorption range that covers the visible spectrum nearly to 700 nm. Free alizarin absorbs light up to 500 nm with the maximum at about 440 nm, which is shown in Figure 8. Binding of AL to TiO2 causes the shift to 500 nm and broadening with the band wing extending to over 600 nm. The observed broadening of absorption spectrum upon adsorption of the tested dyes on nanoparticles implies that they may exhibit prominent optical properties as sensitizers in DSSCs.
Each of the Figures 6-8 presents several curves (colored solid lines) corresponding to the absorption spectra of the dye/TiO2 complex for different times of soaking the TiO2 electrode in the dye solution. The possible concentration-dependent band shifts determined for the spectra with increasing dye load did not exceed 3 nm for all three investigated dyes, which does not exceed the experimental accuracy with these broad bands. No significant spectral effects occur in the examined electrodes, and thus the aggregation of dye molecules was not considered. Free alizarin absorbs light up to 500 nm with the maximum at about 440 nm, which is shown in Figure 8. Binding of AL to TiO 2 causes the shift to 500 nm and broadening with the band wing extending to over 600 nm. Free alizarin absorbs light up to 500 nm with the maximum at about 440 nm, which is shown in Figure 8. Binding of AL to TiO2 causes the shift to 500 nm and broadening with the band wing extending to over 600 nm. The observed broadening of absorption spectrum upon adsorption of the tested dyes on nanoparticles implies that they may exhibit prominent optical properties as sensitizers in DSSCs.

TiO2/Dye/Electrolyte Interface
Each of the Figures 6-8 presents several curves (colored solid lines) corresponding to the absorption spectra of the dye/TiO2 complex for different times of soaking the TiO2 electrode in the dye solution. The possible concentration-dependent band shifts determined for the spectra with increasing dye load did not exceed 3 nm for all three investigated dyes, which does not exceed the experimental accuracy with these broad bands. No significant spectral effects occur in the examined electrodes, and thus the aggregation of dye molecules was not considered. The observed broadening of absorption spectrum upon adsorption of the tested dyes on nanoparticles implies that they may exhibit prominent optical properties as sensitizers in DSSCs.

TiO2/Dye/Electrolyte Interface
Each of the Figures 6-8 presents several curves (colored solid lines) corresponding to the absorption spectra of the dye/TiO 2 complex for different times of soaking the TiO 2 electrode in the dye solution. The possible concentration-dependent band shifts determined for the spectra with increasing dye load did not exceed 3 nm for all three investigated dyes, which does not exceed the experimental accuracy with these broad bands. No significant spectral effects occur in the examined electrodes, and thus the aggregation of dye molecules was not considered.

TiO 2 /Dye/Electrolyte Interface
The electron transfer from the sensitizing dye to TiO 2 nanoparticles requires the LUMO level of the dye to be more reducing than the conduction band of a semiconductor. Analogously, in the process of the regeneration of dye ground state, the dye cation is required to be more oxidizing than the I − /I 3 − redox couple. Using the oxidation potential of the PF, PCV and AL molecules [25,32], the HOMO levels were determined in eV according to the absolute energy scale. The LUMO orbital (E LUMO ) was calculated as a sum of HOMO energy (E HOMO ) and transition energy (E 0−0 ) at the absorption maximum of the dye adsorbed on a TiO 2 nanoparticle, according to the following relation: Particular energy levels, the positions of which are crucial for the operation of the dye-sensitized cell, are depicted in Figure 9. The arrangement of the HOMO and LUMO levels of the investigated dyes in relation to the bottom of the TiO 2 conduction band and redox potential of the electrolyte is suitable for an efficient transfer of electrons in the working cycle of the dye cell. The differences between the positions of HOMO, LUMO, E CB and electrolyte redox level E el enable the determination of the parameters that quantitatively describe the electron transfer reactions: free energy driving force ∆G inj and electron regeneration driving force ∆G reg (Table 1), according to the following equations [33]: Energies 2021, 14, 5529 7 of 18 The electron transfer from the sensitizing dye to TiO2 nanoparticles requires the LUMO level of the dye to be more reducing than the conduction band of a semiconductor. Analogously, in the process of the regeneration of dye ground state, the dye cation is required to be more oxidizing than the I -/I3 -redox couple.
Using the oxidation potential of the PF, PCV and AL molecules [25,32], the HOMO levels were determined in eV according to the absolute energy scale. The LUMO orbital (ELUMO) was calculated as a sum of HOMO energy (EHOMO) and transition energy (E 0-0 ) at the absorption maximum of the dye adsorbed on a TiO2 nanoparticle, according to the following relation: Particular energy levels, the positions of which are crucial for the operation of the dye-sensitized cell, are depicted in Figure 9. The arrangement of the HOMO and LUMO levels of the investigated dyes in relation to the bottom of the TiO2 conduction band and redox potential of the electrolyte is suitable for an efficient transfer of electrons in the working cycle of the dye cell. The differences between the positions of HOMO, LUMO, ECB and electrolyte redox level Eel enable the determination of the parameters that quantitatively describe the electron transfer reactions: free energy driving force ΔGinj and electron regeneration driving force ΔGreg (Table 1), according to the following equations [33]: Figure 9. Scheme of the energy levels of PF/TiO2, PCV/TiO2, AL/TiO2, TiO2 and I -/I3 -electrolyte redox potential. The photoelectric properties of the studied dyes were also evaluated by the determination of the chemical reactivity parameters, including chemical potential, global hardness and electrophilicity (Table 2). Chemical potential µ reflects the escaping tendency of an electron from equilibrium and is connected with rearrangement of electrons in chemical phenomena [34]: µ = −χ, where χ is the electronegativity, E is the total energy of an atom or molecule in the ground state and N is the number of electrons.
Global hardness η g expresses the resistance to electron transfer (equilibrium of the system corresponds to maximum hardness) and can be calculated according to the following formula: The approximation based on Koopman's theorem, successfully used for different kinds of organic and inorganic molecules [34,35], leads to the following formulas expressing µ and η g , which are related directly to the HOMO and LUMO energies: Electrophilicity index ω is also one of the descriptors of reactivity that corresponds to the balance energy of a system saturated by electrons; thus, it is a measure of energy lowering due to transfer of electrons between donor and acceptor [34,36,37]: Global hardness is the highest for PF; however, the differences in values for other compounds are small. The chemical potential values of PF and PCV are around 4.3 eV, close to TiO 2 anatase, which is described further in Section 4.

Photovoltaic Performance
The performance of the dye-sensitized solar cell is characterized by such electric parameters as short-circuit current (I SC ), open-circuit voltage (V OC ), current and voltage at maximum power (P MP ) and fill factor (FF) and efficiency (η) of the cell. The last two parameters are calculated according to the following formulas [38]: where V MP is voltage at maximum power point, I MP is current at maximum power point and P in is the incident light power. Figures 10-12 present the density of current vs. voltage (J-V) characteristics, including the measurements performed with the BaSO 4 reflective layer (RL) on the back side of the dye cells. The electric parameters derived from the characteristics of the investigated photocells are listed in Table 3. The results presented in Table 3 include the parameters obtained for different times of the photoelectrode immersion in the dye solution. The optimal time in terms of efficiency was determined for each type of dye and is indicated in Table 3. The maximum efficiency of the cells with PF and AL is one order of magnitude greater than that for PCV. The implementation of the back reflective layer provides better results for all tested cells. In the case of the PF dye, the improvement of even 17.9% in efficiency (for 2 h dipping time) is observed due to the reflection of light from the BaSO 4 layer. The same tendency is visible for the cells with PCV, the efficiency of which is 60% better (for 1 h dipping time) than in the measurements without reflective layer. For alizarin, the best parameters of the cells were obtained after 2 h soaking time, and a 21.4% increase in efficiency was observed with the use of the BaSO 4 reflective layer.
Energies 2021, 14, 5529 9 of 18 side of the dye cells. The electric parameters derived from the characteristics of the investigated photocells are listed in Table 3. The results presented in Table 3 include the parameters obtained for different times of the photoelectrode immersion in the dye solution. The optimal time in terms of efficiency was determined for each type of dye and is indicated in Table 3. The maximum efficiency of the cells with PF and AL is one order of magnitude greater than that for PCV. The implementation of the back reflective layer provides better results for all tested cells. In the case of the PF dye, the improvement of even 17.9% in efficiency (for 2 h dipping time) is observed due to the reflection of light from the BaSO4 layer. The same tendency is visible for the cells with PCV, the efficiency of which is 60% better (for 1 h dipping time) than in the measurements without reflective layer. For alizarin, the best parameters of the cells were obtained after 2 h soaking time, and a 21.4% increase in efficiency was observed with the use of the BaSO4 reflective layer.       The values of open-circuit voltage are in the range of 387-437 mV for PF, 119-267 mV for PCV and 128-318 mV for AL. The shape of the curves is properly bent; however, the internal resistances occurring in the cells lead to an efficiency decrease.
The performance parameters of the investigated cells can be related to the values reported previously for the cell prepared according to the same methodology based on the N719 dye [39]. The cell sensitized with N719, which can be recognized in this work as the reference cell, provided J SC = 12.29 mA/cm 2 , V OC = 713 mV and the efficiency of 5.03% in the measurements without the reflective BaSO 4 layer. The implementation of the BaSO 4 reflective layer in the reference cell based on N719 resulted in the improvement of the following photovoltaic parameters: J SC = 14.71 mA/cm 2 , V OC = 722 mV and the efficiency of 5.82%, which confirmed the beneficial role of the reflective layer also in the cells with the well-known, efficient sensitizer.
The values of standard deviations of efficiency, determined for the best cells (indicated in bold in Table 3) are as follows: 0.001466 (PF no RL) and 0.000358 (PF with RL), 0.00046 (PCV no RL) and 0.00024 (PCV with RL), 0.00029 (AL no RL) and 0.00022 (AL with RL).
In order to compare the performance of the studied dyes, the absorption of light was analyzed in relation to J SC , which is strongly influenced by the amount of absorbed photons. Figure 13 presents the dependency of absorbance at the absorption maximum, measured for the electrodes covered by TiO 2 and immersed in dye solution for the optimized time (2 h in case of PF and AL, 1 h in case of PCV), versus the short-circuit current density obtained for the cells prepared with these electrodes. The plot in Figure 13 shows that the higher absorbance is accompanied by the greater photocurrent, and the best-performing dye, PF, stands out in terms of both parameters. The values of standard deviations of efficiency, determined for the best cells (indicated in bold in Table 3) are as follows: 0.001466 (PF no RL) and 0.000358 (PF with RL), 0.00046 (PCV no RL) and 0.00024 (PCV with RL), 0.00029 (AL no RL) and 0.00022 (AL with RL).
In order to compare the performance of the studied dyes, the absorption of light was analyzed in relation to JSC, which is strongly influenced by the amount of absorbed photons. Figure 13 presents the dependency of absorbance at the absorption maximum, measured for the electrodes covered by TiO2 and immersed in dye solution for the optimized time (2 h in case of PF and AL, 1 h in case of PCV), versus the short-circuit current density obtained for the cells prepared with these electrodes. The plot in Figure  13 shows that the higher absorbance is accompanied by the greater photocurrent, and the best-performing dye, PF, stands out in terms of both parameters. For the cells with optimized immersion time, the parasitic resistances were also determined. The series resistance is in the range of 542-602 Ω for cells sensitized with PF, 2617-2954 Ω for PCV and 1416-1611 Ω for AL, with RL and without RL. Shunt resistance value is 1562 Ω for cells sensitized with PF, 6250-6666 Ω for PCV and 5000-6944 Ω for AL.

Discussion
The performance of DSSCs depends on numerous factors, including the types of materials used and the structure of the cells. This work is devoted to the application of three different sensitizers and the implementation of the back reflective layer, which raises the chances of absorbing the light that passes through the cell. It is advisable to return the light back to the cell when the cell absorption is relatively poor and also to use the light out of the absorption range.
The photoelectrodes used in the investigated DSSCs, prepared by deposition of the TiO2 paste, exhibited porous structures suitable for adsorption of the dye. The observed light-scattering ability of the TiO2 layer can enable better capturing of the light. The diffuse reflectance of the prepared TiO2 layer is relatively high in comparison to a lower value of 25% or narrow reflectance range in the literature [40,41]. For the cells with optimized immersion time, the parasitic resistances were also determined. The series resistance is in the range of 542-602 Ω for cells sensitized with PF, 2617-2954 Ω for PCV and 1416-1611 Ω for AL, with RL and without RL. Shunt resistance value is 1562 Ω for cells sensitized with PF, 6250-6666 Ω for PCV and 5000-6944 Ω for AL.

Discussion
The performance of DSSCs depends on numerous factors, including the types of materials used and the structure of the cells. This work is devoted to the application of three different sensitizers and the implementation of the back reflective layer, which raises the chances of absorbing the light that passes through the cell. It is advisable to return the light back to the cell when the cell absorption is relatively poor and also to use the light out of the absorption range.
The photoelectrodes used in the investigated DSSCs, prepared by deposition of the TiO 2 paste, exhibited porous structures suitable for adsorption of the dye. The observed light-scattering ability of the TiO 2 layer can enable better capturing of the light. The diffuse reflectance of the prepared TiO 2 layer is relatively high in comparison to a lower value of 25% or narrow reflectance range in the literature [40,41].
The dyes introduced in the dye cells as sensitizers should adsorb on titanium dioxide nanoparticles making a durable bond. The UV-Vis spectroscopy measurements presented in this work confirmed the ability of phenylfluorone, pyrocatechol violet and alizarin to chelate titanium. Similar effects were observed in the studies on the creation of complexes with other metal ions. Phenylfluorone was reported to create a complex with antimony, the features of which were investigated by UV-Vis and Raman spectroscopy, revealing the possibility of application for detection and quantitative determination of the antimony content in toxicology [42]. The material functionalized with PF was also used for cadmium and copper determination in water samples [43]. PCV as a catechol-like compound was proven to be a good ligand for metal ions; e.g., the ligand-exchange reaction of PCV coordinated to Zr(IV) has been observed [44]. In this kind of reaction, the yellow color changes to blue depending on pH, and the absorption is shifted from 440 to 630 nm. Not only Zr but also Ti was reported to form stable complexes with PCV. The titanium alkoxide compound allowed simple preparation of HF-sensitive PCV-Ti complex that changes color upon exposure to the HF vapor [45]. In the literature, the spectra of alizarin-(TiO 2 ) 15 complex were modeled, and the ultrafast electron transfer dynamics of alizarin adsorbed to a colloidal titanium oxide nanocrystallite was investigated theoretically [46].
In order to achieve fast electron injection and regeneration of the oxidized dye, an efficient sensitizer is supposed to possess the ground state below the redox potential of the electrolyte and excited state over the edge of TiO 2 conduction band. This fundamental requirement is fulfilled by all three studied dyes (PF, PCV, AL). Moreover, all three dyes bind to TiO 2 in the same bidentate chelating configuration by the enediol ring; this circumstance provides a similar energy barrier for electron injection in each case and leaves the energetics of electron transfer as the main determinant of the process.
In general, it is assumed that greater ∆G inj leads to more effective electron transfer, contrary to ∆G reg for which lower values are beneficial [47,48]. Taking into account these rules, ∆G inj and ∆G reg of PCV and PF have the best values. However, the photovoltaic performance of PCV is the worst among the studied dyes. The possible aggregation of PCV molecules, revealed especially for longer dipping times, that results in worsened photovoltaic parameters, may suppress the effective electron injection process. In the literature, the ∆G inj value greater than 0.2 eV is recommended [25]; however, for one of the most efficient sensitizers, N719, the driving force is equal to 0.15 eV [38]. In terms of ∆G reg , the value of 0.2-0.3 eV is required for electron recombination for ruthenium dyes by oxidation of iodide/triiodide electrolyte [38].
Nevertheless, the consideration of various literature data and the results of this study do not provide a clear and strict recipe that determines the best configuration of HOMO and LUMO positions of dye molecules in DSSCs. Taking into account the driving force values mentioned above and their connection with the Marcus theory of electron transfer [49], it can be assumed that an increase in ∆G inj is beneficial to some upper limit. Further increase in the difference between the LUMO of the dye and semiconductor bandgap edge can result in the suppression of the DSSC efficiency according to the Marcus inverted regime. However, Marcus theory is devoted to proteins and dye molecules taking part in photosynthesis, and its reference to the dye molecules adsorbed on solid state may not be straightforward.
The chemical reactivity parameters determined in this work present beneficial values for PF, since the greater global hardness and the smaller electrophilicity lead to the higher efficiency of DSSCs [34]. In general, the photosensitizers best suited for titanium dioxide surface should present good matching of both electronegativity and the global hardness to the parameters of TiO 2 anatase, equaling 4.3 and 1.6 eV, respectively [50], which is achieved in this work.
The photovoltaic performance of the solar cells sensitized with three tested dyes exhibits the best values for the PF compound. The value of V OC over 0.4 V achieved in this work is satisfactory taking into account V OC of 0.33-0.64 V for natural dyes [22] or 0.254-0.691 V for the recently designed ruthenium complexes [51]. The J-V curves of dye cells with PF and AL exhibit also the proper shape, which is reflected in fill factor value of up to 39.14% (AL) and over 42% (PF) which is a good result compared to 27-69% for natural sensitizers [22] or 45-70% achieved for Ru complexes [51]. In general, any deterioration of the fill factor decreases the output power. The achieved values of V OC and FF are satisfactory; however, the current and in consequence efficiency of the investigated cells deserve improvement. In the literature, ZnO-based DSSCs sensitized with fluorescein, the molecular structure of which is similar to phenylfluorone, show the efficiency of 6.54% and 0.55% with fluorinated fluorescein [52,53].
For comparison, the literature study on pyranoflavylium salts possessing a catechol linker unit in the sensitizer molecule led to 1.15% efficiency in a DSSC with TiO 2 and iodine electrolyte [29]. The trials involving other catecholate-functionalized compounds in DSSCs resulted in an efficiency of 0.92% [54].
Alizarin was applied in a DSSC before; however, it was a quasi-solid-state DSSC with a polymer electrolyte. After titanium tetrachloride treatment of the TiO 2 layer, which improved the electron transport and dye molecule anchoring, an enhancement of the efficiency was achieved from 3.57 to 5.12% [55]. In the literature, the exposition of a photoanode, prepared by using P25 Degussa TiO 2 nanoparticles sensitized with alizarin, to microwave radiation improved the efficiency of the cells from 0.14 to 0.16% [56]. The application of natural dyes such as alizarin usually leads to low efficiencies of 0.01-0.6% [57], 0.2-0.18% [58] and recently 0.197-0.878% in the investigation of natural plant extracts rich in luteolin [22]. The review of natural sensitizers shows the efficiency values of 0.008% for the ZnO-based cells, 0.0083% and 0.301% for the TiO 2 -based cells, 0.516% for the cells with graphene-coated electrodes and 1.5% in the cells sensitized with Rhus fruit extract [59]. The application of the anthocyanin dye provided a PCE of 2.9% [59]. Low efficiency of 0.1-0.14% was also delivered by the cells sensitized with the dye extracted from chokeberry; however, the reference cell with N719 fabricated in that research provided 5.06% [60]. This value of efficiency is very close to the efficiency of the N719 reference cell used in this work, which amounts to 5.03%.
The efficiency of solar energy conversion of the dye cells was enhanced in this study by applying the back reflective layer of BaSO 4 . The exhibited high efficacy of the reflected light may indicate that the electron transfer to the TiO 2 nanoparticles located closer to the electrolyte is more efficient. The reason may be the faster regeneration of dye molecules that are in good contact with the electrolyte. In deeper parts of the TiO 2 layer, near FTO, the reduction of dye molecules is limited due to the difficulty in electrolyte exchange by diffusion.
Back reflectors were typically introduced in thin-film silicon or copper gallium diselenide (CIGS) cells in order to improve light trapping in the cell. In thin-film silicon cells, other materials were also applied as a reflective layer: porous silicon dielectric layer [61], nanostructured ZnO and Ag [62] and polystyrene microspheres enhancing the short-circuit current and efficiency by over 30% [63]. Commercial white paint containing TiO 2 pigment also was used on the back side of a thin silicon cell, providing a short-circuit current density boost of 41% [64]. In ultrathin CIGS solar cells, Ag back mirror improved short-circuit current density by 20% [65]. In the literature, the use of reflected light in DSSCs is reported rarely. However, some modifications such as reflecting counter electrode [66], photoanode including nanostructured light scattering layer [67] and combination of upconversion nanoparticles and light-reflecting silver particles were introduced [68], resulting in the enhancement of the electrical parameters of DSSCs.

Conclusions
The performance of dye-sensitized solar cells is influenced by numerous factors. This work addresses two aspects: proper selection of the sensitizing dye and application of a back reflector that allows harvesting the light transmitted through the semitransparent dye cell.
In the literature, great effort has been devoted to the development of new effective sensitizers, natural dyes among them. In the presented work, three organic metal-free dyes were chosen to serve as sensitizers in dye-sensitized solar cells: phenylfluorone, pyrocatechol violet and alizarin. The molecules of the investigated dyes possess an enediol group in their structure, which chelates with titanium and can ensure fast electron transfer from the excited sensitizer to the titanium dioxide nanoparticle layer that constitutes the photoactive coverage of illuminated electrodes in DSSCs. The prepared TiO 2 layer showed porous morphology confirmed by scanning electron microscopy and high reflectance of 30-45% in the range of 400-600 nm, favoring the harvesting of diffused light.
All the investigated dyes exhibited a significant broadening of absorption bands upon the adsorption on TiO 2 . The observed improvement of the light-harvesting properties together with the adequate distribution of the HOMO and LUMO positions qualified phenyl fluorine, pyrocatechol violet and alizarin for applications in DSSCs. The dye cells were fabricated, and their performance parameters were determined. Phenylfluorone was found as the best-performing sensitizer among the tested dyes, which was indicated by the beneficial values of chemical reactivity parameters. Phenylfluorone also presents the highest value of the absorbance at the absorption maximum and in consequence the highest value of the photocurrent. The observed differences in photocurrent obtained with the studied dyes are significant, which suggests that not only the light absorption but also the efficiency of electron injection differs among the dyes. However, internal resistances deteriorate the overall performance of the tested cells. Prospective improvement of the dye cells should include the reduction in resistance of metal contacts and better control of defects that contribute to leakage current.
In order to improve the efficiency, the reflective layer made of barium sulfate was implemented on the back side of the semitransparent cells. It was found that the application of back reflector is beneficial, and a significant improvement in the efficiency of the cells was observed. The achieved increase in the efficiency of the cells enriched with reflective layer was up to 60%.
Although further optimization of dye cells is still necessary to gain better photovoltaic parameters, the application of the BaSO 4 reflective layer is a very promising solution that can be introduced in all kinds of semitransparent cells.