Investigation of UV Dye-Sensitized Solar Cells Based on Water Electrolyte: A New Insight for Wavelength-Selective Greenhouse

Abstract

The optimization of the photoactive electrode based on TiO₂ with a complex architecture for UV dyes along with water-based electrolyte has successfully allowed us (i) to obtain a photovoltaic efficiency of the dye-sensitized solar cell with 1.45 times higher than the best efficiency reported for synthetic dye and 3 times for curcumin dye so far; (ii) transparency on the entire Photosynthetic Active Radiation domain; (iii) preserving high efficiency for lighting 1 sun (summer) and shading, especially for 60 mW/cm², which represents the maximum illumination in the rest of the seasons. Our water-based dye-sensitized solar cells loaded with synthetic and natural UV dyes have revealed that the implementation of a dye-sensitized solar cell in autonomous greenhouses is a viable and inexpensive concept.

1. Introduction

In the context of the expected energy and food crisis, medium- and long-term forecasts have highlighted the need for finding synergetic solutions. An agrivoltaic system is proposed as a possible option that combines agriculture and the generation of photovoltaic energy, alleviating land competition or other spatial restrictions for energy production. A photovoltaic greenhouse must strike a balance between two complementary requirements: (i) maximizing the flow of Photosynthetic Active Radiation (PAR), the light of wavelengths between 430 and 700 nm, which is essential for the growth and photosynthesis of plants, by reducing the shading effect of photovoltaic panels [1], and (ii) further enhancement of the production of energy that will ensure the energetic autonomy of the greenhouse.

The theoretically predicted power conversion efficiency (PCE) of the dye-sensitized solar cell (DSSC) is approximately 20% [2], and in the last few decades, a lot of experiments have focused on continuing increasing the efficiency compared with the used Si-based solar cell, as well as the large modules installations for terrestrial power generation. In recent years, this limited objective has been extended to the advantages over the widely used Si-based solar cell, including higher energy conversion efficiency under weak and indirect illumination conditions. These advantages can compensate for the weaknesses of Si-based solar cells and together with the simple manufacturing process, low fabrication cost, flexibility in scaling, low material usage [3,4,5], and low light level sensitivity, but mainly the variation in color and transparency of the photoanode together with that of the counter electrode [6,7], are essential characteristics that could make these cells the ideal candidate for greenhouse application [8,9].

The selection of the color of DSSC given by the dye can act as a plant growth regulator or serve as a photo selective covering adsorbed with dye to manipulate the light spectrum entering the greenhouse. Thus, this technology could deliver impressive benefits in contrast to conventional photovoltaic panels (the first and second generations of the solar cells) because of its solar radiation manipulation through the optimum choice of photosensitizer.

Making more efficient photoelectrodes is one of the key factors needed for the increased performance of DSSCs, being responsible for the dye loading capacity and the transfer pathway of electrons. For this, several strategies for the TiO₂ photoanode improvement were proposed, including the overlayer of TiO₂ substrate with CdSe particles for a better electron transfer [10], the Ag doping of the TiO₂ material leading to a visible increase in J_SC [11], and a polymer interlayer TiO₂-based DSSC for the prevention of charge recombination and to facilitate the ionic conduction [12].

Our paper aims to propose and experimentally demonstrate an architecture of DSSC which shall be respected simultaneously: (i) partial absorption of solar UV radiation with a positive impact on plant growth and reduced plant pathogenicity [13]; (ii) transparency throughout the PAR domain; (iii) water-based electrolytes, ensuring the realization of safe, less expensive, and environmentally friendly solar cells [14]; and (iv) improvement of the energy performance of DSSC based on UV dyes for full sun and shading conditions. Each of the above requirements has been successfully fulfilled, namely: (i) an affordable commercial dye (DN-F01) and a natural curcumin dye that absorbs UV radiation; (ii) introduction of the water in a low-cost common electrolyte; (iii) design and optimization of the photoactive electrode based on the TiO₂ polymorphs for these dyes and water-based electrolyte; (iv) improving the energy performance of DSSC based on synthetic and natural UV dyes for full sun and shading conditions, without affecting PAR.

Thus, the photoactive TiO₂-based electrode with complex architecture was successfully optimized for UV dyes sensitization and tested using water-based electrolyte, yielding a photovoltaic efficiency 1.45 and 3 times higher than the best efficiency reported for synthetic and curcumin dye so far [15,16]. With a maximum of efficiency under 60 mW/cm² of illumination, excellent performance even under conditions of not optimal lighting was obtained.

2. Materials and Methods

2.1. Preparation of TiO₂ Nanoparticles (TiO₂_NP)

An amount of 2.13 mL of titanium (IV) isopropoxide (Sigma-Aldrich Co., Merck KGaA, Darmstadt, Germany) was added at room temperature and continuously stirred into a solution formed from 1 g of Pluronic P123 (Sigma-Aldrich) dissolved in 10 mL of isopropyl alcohol. The prepared solution was stirred for 3 h at 350 rpm, and after left to age for 24 h, until a solid gel was formed. The as obtained gel was then dried at 60 °C for 24 h and further sintered at 450 °C for 1 h, to obtain the white TiO₂ particles.

2.2. Preparation of TiO₂ Particles for Light Scattering (TiO₂_LS)

In the first step, TiO₂_LS1 were obtained by dissolving 5 mL of titanium (III) chloride solution (Sigma-Aldrich Co., Merck KGaA, Darmstadt, Germany) in 30 mL of water and stirred for 10 min at room temperature. 6 mL of supersaturated NaCl solution was then added to the previously prepared mixture and stirred for another 10 min. The resulting mixture was transferred to a 50% filled Teflon-lined autoclave and heated at 160 °C for 2 h. After the hydrothermal treatment a white precipitate was obtained, which was washed several times by centrifugation at 8000 rpm with Dl water and ethanol and air-dried at 30 °C for 24 h.

For the formation of nanosheet aggregates, another step of hydrothermal treatment was applied to a mixture consisting of 0.1 g of a previously made TiO₂_LS1 precursor, and added into a solution of 5M NaOH, at 140 °C for 24 h. The resulting white sediment was washed by filtration with a 0.1 M HCl solution until an acidic pH of the rinse solution was obtained and further sintered at 400 °C for 30 min, to obtain the TiO₂ nanosheet aggregates (TiO₂_LS2) for light scattering.

The phase purity and crystal structure of the as-obtained powders were characterized by X-ray diffraction (XRD), using the PANalytical PW 3040/60 X’Pert PRO diffractometer (Malvern Panalytical, Malvern, United Kingdom), with a 1.5418 Å wavelength of the Cu-Kα radiation, in the range 2θ = 10–80°. Surface morphology was analyzed with the FEI Inspect S scanning electron microscope (SEM, FEI Company, Eindhoven, The Netherlands). A Lambda 950 UV-Vis-NIR (PerkinElmer, Waltham, CT, USA) spectrophotometer with a 150 mm integrating sphere was used to collect the diffuse reflectance spectra (DSR) in the wavelength range of 300–800 nm at room temperature, and the optical band gap of the oxide powders was estimated. The FTIR spectra were collected in the 4000–400 cm⁻¹ range, with a JASCO-430 (Jasco Co., Tokyo, Japan) Fourier transform spectrometer and by using the KBr pellet technique.

2.3. Preparation of Curcumin Dye

To obtain the curcumin dye, 4 g of turmeric powder was dissolved in 200 mL of ethyl acetate by magnetic stirring for 1 h at 40 °C. The extract was then filtered on filter paper three times to remove any solid particles. The concentration of the curcumin extract was made using a Heidolph Rotary Evaporator-Laborota 4000 (Heidolph Instruments, Schwabach, Germany), at a pressure of 160 mbar and a bath temperature of 40 °C. An amount of 40 mL of the concentrated natural curcumin dye was obtained and used as a photosensitizer in DSSC. The extraction of the natural dye from turmeric powder is described in Figure 1.

The absorbance spectrum was collected using a Lambda 950 UV-Vis-NIR (PerkinElmer, Waltham, CT, USA) spectrophotometer in the wavelength range of 250–800 nm at ambient temperature.

2.4. Fabrication of DSSCs

TiO₂_NP and TiO₂_LS pastes were prepared as in our previous study [17] using ethyl cellulose, α-terpineol, glacial acetic acid, and Dl water as paste constituents.

For the optimized photoanode realization, multiple layers of paste were deposited using the doctor blade technique on the previously cleaned and UV-Ozone-treated FTO glass substrate (Sigma Aldrich, with a surface resistivity of ~13 Ω sq⁻¹). First, a TiO₂_NP layer was deposited, followed by 2 layers of TiO₂_LS (TiO₂_LS1 or TiO₂_LS2), after each deposited layer, a sintering process was applied at a temperature of 500 °C for 1 h with a 1 °C min⁻¹ heating rate. Finally, the double-layered paste configuration was immersed in a 40 mM TiCl₄ solution at 70 °C for 1 h and sintered at 450 °C for 1 h with a 1 °C min⁻¹ heating rate.

The optimized configuration was determined by testing different photoanodes prepared with different functionalized layers. Photoanode 1 (PhA 1) consists of a single layer of TiO₂_NP, photoanode 2 (PhA 2) consists of TiO₂_NP and TiO₂_LS1 layers, photoanode 3 (PhA 3) consists of TiO₂_NP and TiO₂_LS2 layers, and photoanode 4 (PhA 4) consists of a TiO₂_NP layer, a TiO₂_LS2 layer, and TiCl₄ treatment. For a better understanding of the layer configuration, a schematic representation is presented in Figure 2.

Two dyes were used for the sensitization of the photoanodes, a synthetic dye consisting of a 0.3 mM solution of DN-F01 (Dyenamo Yellow) in absolute ethanol (5 h immersion time) and a natural curcumin extract (1 h immersion time). The obtained sensitized photoanodes were fixed together with the counter electrodes produced by treating a H₂PtCl₆ solution at 400 °C, using a Meltonix 1170-60 thick spacer.

The electrolyte used and injected into the space between the electrodes consists of a solution of 0.6M 1-butyl-3-methyl-immidazolium iodide, 0.03 M I₂, 0.10 M guanidinium thiocyanate and 0.5 M 4-tertbutylpyridine in acetonitrile/valeronitrile (85:15, v/v), hereinafter referred to as E. The water effect on the DSSC performance was also studied by replacing 10% of the solvent medium in the E electrolyte with water, named E10%.

The DSSCs produced based on the photoanodes will henceforth be called DSSC 1, DSSC 2, DSSC 3, and DSSC 4, respectively.

For determining the photovoltaic performances of the DSSCs, the photocurrent density vs. photovoltage curves were recorded on a Keithley 2450 source measure unit, keeping a scan rate equal to 1 mV and under different illumination conditions (20–100 mW cm⁻²). Electrochemical impedance spectroscopy (EIS) measurements were performed using a Voltalab potentiostat model PGZ 402, with VoltaMaster 4 software (version 7.09), under different illumination conditions and a frequency range from 0.001 to 10 kHz, using 10 mV of the magnitude for the modulation signal. The applied potentials were taken as the V_OC values determined for the DSSC4 configuration tested with and without water content in the electrolyte, at two illumination intensities (60 and 100 mW cm⁻²), and by using both dyes’ sensitization (synthetic and natural).

3. Results

The XRD pattern of the TiO₂ nanoparticles obtained by a simple sol–gel procedure is shown in Figure 3a. All diffraction peaks are indexed as TiO₂ crystallized in anatase form, having a tetragonal structure (space group: I41/amd; JCPDS No. 01-073-1764). No other TiO₂ crystalline phases such as rutile or brookite can be observed. The XRD patterns of the obtained microparticles used as scattering layers in the DSSC configurations are presented in Figure 3b,c. In the case of the TiO₂_LS1 sample, a crystallization in rutile form with the tetragonal structure of the TiO₂ can be observed (space group: P42/mnm; JCPDS No. 00-021-1276). Furthermore, a mixture of polymorphic phases of ~46% rutile and ~54% anatase was obtained in the TiO₂_LS2 sample.

The crystallite sizes of the obtained TiO₂ powders were calculated using the Scherrer formula [18]:

$$D_{hkl}=\frac{k\lambda }{Bcos\theta }$$

(1)

where k is a constant (∼1), B is the full width at half maximum (FWHM), λ the wavelength of the X-ray and θ is the diffraction angle. The obtained crystallite sizes were of ~15 nm for the TiO₂_NP sample, of ~25 nm for the TiO₂_LS1 sample, and of ~33 nm and ~23 nm for the anatase and rutile phases of the TiO₂_LS2 sample, respectively.

Figure 4 presents the SEM images of the two architectures proposed for the light scattering effect, TiO₂_LS1 showing a solid microsphere structure, and a dendritic structure containing agglomerated nanorods was obtained for the TiO₂_LS2 sample.

The FT-IR spectra of the samples measured in the range of 4000–400 cm⁻¹ are presented in Figure 5. The first two peaks observed at 3438 and 1630 cm⁻¹ correspond to the stretching vibrations of the hydroxyl group O-H of interlayer water molecules and the bending mode of water molecules δ(H₂O), respectively [19]. The broad and intense band between 881 cm⁻¹ and 408 cm⁻¹ with maximum at 481 cm⁻¹ assigned to Ti-O and Ti-O-Ti stretching and bending vibrations is characteristic of the pure anatase polymorph of TiO₂ [20]. In case of TiO₂_LS1, the intense band in the range of 881–452 cm⁻¹ with a maximum at 620 cm⁻¹ is attributed to the Ti-O vibration of pure rutile [21]. In accordance with the XRD pattern of TiO₂_LS2, the band widening is extended in the range 1025–408 cm⁻¹ with two specific maxima for both anatase and rutile polymorphs. The high NaOH molarity used in TiO₂_LS2 preparation caused the formation of surface hydroperoxo species, TiOOH, that exhibits an absorption band at 970 cm⁻¹ due to vibration modes [22].

From the diffuse reflectance spectra, presented in Figure 6, the optical direct band gaps of the TiO₂ photoelectrode materials were estimated by using the Tauc plot [23], and intercepting the extrapolated linear fit for the plotted experimental data of (αhν)² versus incident photon energy (hν) near the absorption edge. The obtained values of 3.12 eV and 3.09 eV for TiO₂_NP and TiO₂_LS1, respectively, are comparable to the data reported in the literature for anatase and rutile polymorphs of TiO₂ structures [24]. The value of 3.36 eV obtained for the TiO₂_LS2 sample is much higher than previously reported values of mixed anatase-rutile TiO₂ nanoparticles [25]. In general, the value of Eg of a mixed TiO₂ nanoparticles with rutile and anatase structures decreases continuously with the increase in rutile content. In the case of TiO₂_LS2, a significant increase in the band gap was highlighted and could be correlated with the presence of the surface titanium hydroperoxo species revelated by FTIR spectra.

The UV-Vis spectrum of curcumin in ethyl acetate solution showed a broad absorption at around 300–450 nm with a maximum absorption band at a wavelength of 416 nm attributed to the electronic transitions n → π *, and a weak absorption band at 251 nm to the electronic transitions π → π *, respectively. The groups responsible for the absorption of the dye are the carbonyl groups in the structure of curcumin and the accumulation of groups grafted on the aromatic nucleus producing a bathochrome shift. The DN-F01 dye is similarly characterized by a broad absorption at around 350–480 nm with a maximum absorption band at a wavelength of 421 nm (Figure 7).

The successful implementation of DSSC in an energetically independent greenhouse is conditioned by four main requirements, namely eco-friendly materials, transparency in the entire PAR domain, high UV absorption, good photovoltaic efficiency, and sustainability throughout the year. To simultaneously fulfil the above-mentioned objectives, four different architectures of TiO₂ photoactive electrodes (PhA 1, PhA 2, PhA 3, and PhA 4) using synthetic and natural dyes, and water-based electrolytes were designed and tested.

From the SEM images presented in Figure 8a–d, the uniformity and surface morphology of the as-deposited TiO₂ thin films can be observed. As can be seen, the particle morphologies were preserved during the paste formation, deposition, and thermal treatment of photoanodes.

To investigate the light scattering effect of the TiO₂ particles synthesized for this purpose, the reflectance spectra of all photoelectrode configurations were studied (Figure 9). A higher reflectance for the PhA 2, PhA 3, and PhA 4 photoelectrodes in the wavelength range of 400–800 can be observed compared with that of the TiO₂_NP particle film (PhA 1), confirming that spherical TiO₂ aggregate films have a higher light scattering capacity, due to their larger diameters, which are comparable to the wavelengths of visible light. The distance traveled by the incident light within the photoelectrode is significantly extended by the better scattering effect, with this increasing the photon harvesting by the dye molecules and leading to a relatively higher photocurrent [26].

The J-V characteristics of the DSSCs using E and E10% electrolytes are presented in Figure 10 and the detailed photovoltaic parameters (J_SC, V_OC, FF, and η) for both dyes are summarized in

Table 1. Photovoltaic parameters of the DSSCs tested using E and E10% electrolytes and both synthetic and natural dyes.

Sample	Dye	Electrolyte	JSC (mA)	VOC (mV)	FF (%)	η (%)
DSSC 1	DN-F01	E	4.15	738	42.7	1.307
DSSC 2			4.2	743	49.7	1.550
DSSC 3			5.76	778	43.8	1.962
DSSC 4			6.38	741	65.3	3.087
DSSC 1		E10%	4.35	726	50.0	1.579
DSSC 2			4.32	703	47.7	1.448
DSSC 3			6.06	761	50.2	2.315
DSSC 4			6.78	792	64.7	3.474
Best record a	DN-F01	0.6 M TBAI, 0.1 M LiI, 0.05 M I2, and 0.5 M 4-tert-butylpyridine (4-TBP) in acetonitrile	5.23	710	64	2.39
DSSC 4	Curcumin	E	3.37	570	62.9	1.208
		E10%	2.73	574	61.8	0.968
Best record b	Curcumin from turmeric powder	I−/I3− redox based electrolyte	0.720	432	40.0	0.41

^a Ref. [15], ^b Ref. [16].

. In the case of the synthetic dye and E electrolyte, a significant increase in the efficiency is correlated with the addition of the LS2 layer (50% more), and together with TiCl₄ treatment, 136% is achieved relative to DSSC1. The microsphere rutile structure of TiO₂_LS1 had a minor effect on the photovoltaic efficiency caused by the higher light-scattering capacity compared with TiO₂_NP. A positive effect of TiO₂_LS2 on the photovoltaic parameters is provided by the dendritic morphology presented in the mixture of rutile and anatase polymorphic phases along with the surface hydroperoxo species, TiOOH. The increase in J_SC (1.5 times) is directly determined by both the rutile phase of the TiO₂_LS2 microdendritic structure and the high refractive index (n_rutile = 2.8736) which reflects the incident sunlight onto the dye, increasing the light-absorbing capacity of the dye. Additionally, the presence of rutile content prompts the electron transfer from rutile to the anatase trapping sites; this synergistic effect inhibits the electron-hole recombination occurrence, thus leading to an increase in DSSC performance [27]. The high anchoring of the dye molecule on the TiO₂ surface is due to the hydroxyl group (−OH) given by the TiOOH surface species and in addition to the TiCl₄ treatment have improved the dye loading capacity of the DSSC 4. Furthermore, the improvement of V_OC (more than 40 mV) is the result of the high bandgap energy (Eg) of the TiO₂_LS2 compared with the anatase phase (Figure 6) which is only reflected in the most negative conduction band (CB) level, the valence band (VB) level being similar for both crystal phases, rutile and anatase polymorphs. A similar beneficial evolution of the photovoltaic parameters was observed for PhA 4 loaded with the curcumin dye, obtaining a photovoltaic efficiency three times higher than the best efficiency reported for this dye so far [16].

Achieving photovoltaic efficiencies higher than the best efficiency reported so far for both dyes, the complex architecture of PhA4 is validated for the future optimization of the electrolyte, another important component of DSSC.

In an attempt to reduce the volatility, toxicity, flammability of the electrolyte, and at the same time, overcome this poor stability on moisture, one of the main steps to eco-friendly materials, the effect of water (Table 1), has been studied in detail for the configuration of DSSC 4. A clear increase in efficiency can be observed with the addition of water in the electrolyte for the DN-F01 dye, which is especially attributable to the increases in V_OC with 51 mV by more than in the electrolyte without water. Due to the higher solubility in the water of I⁻, compared with that of I₃⁻, the addition of water to the electrolyte leads to a positive shift of the potential [28]. If no negative effect of water was observed in the case of the synthetic dye, the curcumin dye was affected and a decrease in the cell photocurrent, mostly due to the dye detachment from the semiconductors surface promoted by the water, was highlighted.

The transmittance spectra (Figure 11) were similar for DSSC 4, characterized by a high absorption of UV radiation, close to 98% for DSSC 4 with E10% electrolyte and a transparency of the DSSCs on the entire PAR domain, in accordance with those reported in the literature [29].

One of the most important advantages of the DSSCs is sustainability over the whole year, a prerequisite for successful implementation in a wavelength-selective greenhouse. The effect of different light intensities (ranging from 20 to 100 mW/cm²) on the photovoltaic performance of DSSC 4 was investigated (

Table 2. The photovoltaic parameters of DSSC4 under different illumination intensities.

Cell	Pin (mW/cm2)	Electrolyte	Dye	JSC (mA)	VOC (mV)	P (µW)	FF (%)	η (%)
DSSC 4	20	E	DN-F01	0.85	648	387	70	1.927
	30			0.90	647	409	70	1.358
	40			1.25	661	579	69.7	1.439
	50			1.46	670	671	68.4	1.338
	60			3.89	729	1926	67.8	3.204
	70			4.25	728	2101	67.8	2.996
	80			5.00	732	2435	66.5	3.042
	90			5.56	736	2681	65.3	2.969
	100			6.38	741	3087	65.3	3.087
DSSC 4	20	E10%	DN-F01	0.80	668	368	68.8	1.838
	30			0.82	659	361	68.5	1.233
	40			1.21	680	564	68.2	1.402
	50			1.43	697	681	68.0	1.355
	60			4.06	768	2105	67.5	3.507
	70			4.23	771	2202	67.6	3.149
	80			4.86	776	2509	66.4	3.130
	90			5.87	786	3032	65.6	3.362
	100			6.78	792	3474	64.7	3.474
DSSC 4	20	E	Curcumin	0.56	529	426	64.4	0.953
	30			0.60	532	213	67.0	0.713
	40			0.80	537	291	67.6	0.726
	50			0.87	537	313	66.9	0.625
	60			2.05	570	747	64.0	1.246
	70			2.06	564	765	65.8	1.092
	80			2.43	565	859	62.6	1.074
	90			2.72	567	1004	65.2	1.117
	100			3.37	570	1208	62.9	1.208
DSSC 4	20	E10%	Curcumin	0.38	537	125	60.7	0.619
	30			0.43	532	149	64.7	0.493
	40			0.64	538	220	64.0	0.551
	50			0.68	539	239	64.9	0.476
	60			1.65	560	581	62.8	0.967
	70			1.88	568	672	62.8	0.958
	80			2.04	567	732	63.1	0.912
	90			2.35	569	852	63.7	0.946
	100			2.73	574	969	61.8	0.968

, Figure 12 and Figure 13). To give a better understanding of the charge dynamics involved in our DSSCs and the effect of the water, electrochemical impedance spectroscopy (EIS) analysis was performed at V_OC and under 1 sun illumination. In the equivalent circuit presented in the insets of Figure 14, R1 represents the intrinsic resistance of the assembled cells, R2 is the charge transfer resistance at the CE/electrolyte interface, and R3 represents the charge transfer resistance at the TiO₂ photoelectrode/dye/electrolyte interface.

Table 3. EIS parameters obtained from fitting the Nyquist plots of DSSC 4 with and without water content using both DN-F01 and curcumin dye, and under 60 mW/cm² and 100 mW/cm² of the illumination intensity.

Cell	Dye	Pin (mW/cm2)	Electrolyte	R1 (Ω)	R2 (Ω)	R3 (Ω)
DSSC 4	DN-F01	60	E	7.15	3.2	12.4
		100		7.14	2.26	7.56
		60	E10%	6.82	1.64	14.51
		100		6.82	1.27	8.27
	Curcumin	60	E	7.09	3.08	21.24
		100		7.09	2.14	13.24
		60	E10%	7.65	3.79	32.96
		100		7.8	2.51	14.67

summarizes the fitted values of the resistance associated with each interfacial process in the DSSC.

In case of the synthetic dye (Figure 12), under the illumination with different light intensities, the J_SC value varies almost identically for both DSSCs, without and with water, from approximately 0.8 to 6.7 mA/cm² with a steep increase to 60 mW/cm². EIS analysis highlighted that the evolution of J_SC is directly correlated with R2 and R3 (Table 3), for example, in the case of 60 mW/cm² and 100 mW/cm², increasing J_SC is determined by reducing the charge transfer resistance at the TiO₂ photoelectrode/dye/electrolyte interface and the improved catalytic activity, more drastically for E10%. The open-circuit potential slightly increases with increasing illumination intensity and becomes almost constant after 60 mW/cm². The voltage dependence of the output power (P = IV) under different illumination intensities is shown in Figure 12. The power conversion increases with increases in the light intensity until 3087 µW for DSSC 4_E and 3474 µW for DSSC 4_E10%, respectively, at 100 mW/cm². The stability of our DSSCs with increasing light intensity is highlighted by the near-constant evolution of the fill factor, even with a 10% water content. DSSC 4_E10 demonstrates nearly 3.5% photovoltaic efficiency for illumination intensity between 60 mW/cm² and 100 mW/cm² with a maximum at 60 mW/cm².

In the case of the curcumin dye (Figure 13), although the evolution of the photovoltaic parameters (J_SC, V_OC, ff, η) depending on the illumination intensity is similar to that of the synthetic dye, the water slightly affects the performance of the DSSC (increasing R2 and R3), but not the stability under illumination. The photovoltaic efficiency remains almost constant for illumination intensities between 60 mW/cm² and 100 mW/cm² with a maximum at 60 mW/cm².

Our water-based dye-sensitized solar cells loaded with synthetic and natural UV dyes have revealed excellent performance even under conditions of not-optimal lighting, which can usually occur in real outdoor applications, such as a greenhouse during of the day, month, or season.

4. Conclusions

In conclusion, a photoanode with a complex architecture consisting of a TiO₂_NP layer, a TiO₂_LS2 layer, and TiCl₄ treatment was designed, built, and optimized for loading with an affordable commercial UV synthetic and natural dye using a water-based electrolyte. The best water-based dye-sensitized solar cell has demonstrated nearly 3.5% of photovoltaic efficiency at an illumination intensity between 60 mW/cm² and 100 mW/cm². Highlighting the maximum efficiency during all four seasons, along with selective absorption of UV, transparent PAR, and improved eco-friendly characteristics of the materials used, has demonstrated experimentally that the implementation of water-based dye-sensitized solar cells in a wavelength-selective and autonomous greenhouses is a viable and inexpensive concept for an agrivoltaic system.