Fresh Bougainvillea glabra and spectabilis flowers were harvested in central Mexico (Queretaro). Violet and red bracts were used from both species. All flowers’ juices were prepared by grinding 12 g of dried flowers in 500 mL of acid water with pH = 5.7; these extracts were filtered and centrifuged separately to remove any solid residue and stabilized at pH = 5.7 by the addition of an aqueous solution 1 N of HCl. Two different extracts from red Bougainvillea glabra (RBG) were used: the first one was taken from the half of the extract and purified using an Amberlited packed column and ammonium acetate (CH₃COONH₄) 1 M in acid water as the eluting solvent to leave only the yellow and red betalain fractions; the second half was used as prepared to determine the efficiency in the sensitization effect that can be reached with minimal chemical procedure and whether the presence of other compounds, in addition to the dye itself, do not interfere with the performed of the dye. The rest of the extracts: Violet Bougainvillea glabra (VBG), Red Bougainvillea spectabilis (RBS), and Violet Bougainvillea spectabilis (VBS), were used without further purification. The dye solutions in acid conditions (pH = 5.7) are stable with a half-time deactivation of more than 12 months when they are properly stored, protected from direct sunlight and refrigerated at about 4 °C [30]. The betaxanthin content was determined using UV-Vis spectroscopy.

The conductive glass plates were indium tin oxide (ITO)-coated glass slides (In₂O₃:SnO₂) with a sheet resistance of 30–60 Ω/cm², 84% transmittance nominal at 550 nm, dimensions (L,W,D) of 25 × 25 × 1.1 mm. Titanium oxide (TiO₂) nanopowder (mesh 320) (Aldrich) and the solvents, ethanol and acetonitrile (Aldrich), were analytic grade; they were used as received. ITO substrates were ultrasonically cleaned in an ethanol-water mixture for 30 min and then heated at 450 °C during 30 min prior to film deposition. The photo-anodes were prepared by depositing two TiO₂ films on the cleaned ITO glass: for the first one the ITO glass was immersed in a sol-gel solution containing a mixture of titanium isopropoxide, water and isopropanol at concentration 2:1:25 vol.; the immersion was at constant speed (1.5 cm/min) [31]; the coated glass was heated at 450 °C during 30 min producing a densified TiO₂ thin film; this isolate the dye-activated porous TiO₂ layer from the conductor glass. Subsequently two edges of the ITO glass plate were covered with adhesive tape (Scotch 3M) to control the thickness of the film; finally a TiO₂ paste was spread uniformly on the substrate by sliding a glass rod along the tape spacer. The TiO₂ paste was prepared by mixing 3.0 g of TiO₂ nano-powder, 10 mL of nitric acid 0.1 N and 4 mL of polyethylene glycol; this suspension was stirred in a closed glass container for 24 h to obtain a smooth paste with the appropriate viscosity. The film was heated at 450 °C for 60 min resulting in a mesoporous film with a thickness of around 8–10 μm and opaque. The TiO₂ photo-anodes were first soaked for 12 h in HCl and then immersed in the natural dye solutions for one night at room temperature, according to published procedures [17]. Later, the photo-anodes were rinsed with distilled water and ethanol and dried. Carbon coated counter electrodes were prepared following a procedure reported elsewhere [32].

An electrolyte solution was prepared as reported elsewhere [17]: 0.1 M of I₂ was mixed with 0.05 M of LiI and 0.05 M of 3-methoxypropionitrile in 50 mL of acetonitrile (C₂H₃N) and stirring for 60 min. This electrolyte solution was poured in the mesoporous TiO₂ film which was previously prepared using paraffin-film as framework to seal the cells to prevent evaporation of the liquids. The counter electrode was pressed against the impregnated anode and clamped firmly in a sandwich configuration. No leaks (solvent evaporation) were detected.

The topology of the TiO₂ films was obtained using a scanning electron microscope (SEM) Jeol JSM-6060LV operated at 20 kV in secondary electron mode with different magnifications; the samples were covered with a gold film. The particle size and the particle size distribution of the commercial TiO₂ particles were determined using a light scattering apparatus Brookhaven Instruments model BI200SM, equipped with a high speed digital correlator PCI-BI9000AT, a solid-state photon detector and a He-Ne laser of 35 mW Melles Griot 9167EB-1 as a light source. The crystalline structure of the TiO₂ thin film was determined using a diffractometer Rigaku model Miniflex+ equipped with a radiation source of 1.54 Å and the angle 2θ was varied from 5° to 80° at a scan of 2°/min. Absorption spectra were obtained using a UV-Vis spectrometer Genesys 2PC. The DSSCs were illuminated using a 75 W Halogen lamp with an incident power of about 100 mW/cm² in an illumination area of 0.16 cm²; UV and IR filter glasses were used in front of the sample. Photocurrent and photovoltage were measured using a Keithley 2400 source meter and the incident light power was determined using a Powermeter Thor Labs S130A (0–300 mw) and a Spectrometer Ocean Optics HR 400.

Figure 3 shows the absorption spectra of RBS and RBG extracts, both diluted in water at pH = 5.7. In both cases two peaks were found: the first one around 480 nm (480 nm for RBS and 481 nm for RBG) can be associated to the presence of indicaxanthin, while the second one at 535 nm is attributable to the betanin or the bougainvillein-r.

The indicaxanthin concentration in μM was estimated from the absorbance peaks at 481 and 535 nm using the expression [24,33]:

The extracted and diluted VBS and VBG solutions were analyzed with the spectrophotometer at wavelengths between 260 and 700 nm. The absorption spectrum showed two major absorption peaks at 310 and 535 nm for VBS extract and at 307 and 547 nm for VBG extract (Figure 4). Absorbance peaks at 300 and 535 nm are characteristic absorptions for red violet betalain group, betacyanin [34], for bougainvillein-r the absorbance peaks are 280, 320 and 541 nm [23].

The substrates were immersed for 12 h in a 1 M HCl solution to facilitate the absorption of betacyanins on betaxanthins; it is known [17] that the betacyanins absorption is inhibited by the presence of indicaxina. A dip in HCl results in the protonation of the betalainic carboxyl group from its anionic form, promoting its absorption in the TiO₂ film.

The particle size and the particle size distribution of the commercial TiO₂ particles were determined by dynamic light scattering (DLS) resulting of (285 ± 15) nm. The TiO₂ surface morphology is shown in Figure 5a,b. Figure 5a shows a compact (no pores) structure of the first thin layer of TiO₂ obtained by immersing the ITO plates in a titanium alkoxide sol-gel precursor. This thin film acts as an insulator between the ITO layer and the dye-activated porous TiO₂ layer. Some authors [20] have proved the usefulness of the compact layer to achieve a higher efficiency in the performance of the DSSC. On the other side, Figure 5b shows the surface of the thick TiO₂ coating obtained by the screen-printing technique; here it is possible to see a homogeneous mesoporous surface formed by the TiO₂ nanoparticles. The nanoporous structure is useful because it has a high effective surface area where the molecules of the Bougainvillea extracts can be linked.

Five different Bougainvillea extracts were investigated. The first four correspond to extracts obtained from: RBG, VBG, RBS and VBS. They were centrifuged, filtered and used fresh without further purification, as explained above. The fifth one was Red Bougainvillea Spectabilis extract freshly centrifuged and purified by a packed column to remove the bulk of free sucrose compounds; this extract was named PRBS.

Table 1 summarize the results for the five extracts; in all cases low currents were obtained, but comparatively higher compared to those obtained by other authors [17]; this probably partially originated as a result of using ITO-glass with sheet resistance of 30–60 Ohm cm⁻² instead of Fluorine doped tin oxide (FTO) glass with sheet resistance of 10–15 Ohm cm⁻² as reported elsewhere [20]. In this table the short-circuit photocurrent density J_SC, the current density at maximum power J_MP, the open-circuit voltage V_OC, the maximum power Voltage V_MP, the maximum power P_M, the theoretical power P_T, the fill factor FF, and the energy conversion efficiency η, are reported.

It was found that regardless of the species of Bougainvillea used, the extract with the best performance parameters comes from the red bracts (RBG and RBS) as shown in Figure 6. This is because these extracts have both betaxanthin and betacyanins; each with absorptions at different wavelengths, helping the cell to capture photons of two different energies. Figure 7 shows the power curves obtained in terms of voltage; it is possible to observe again the improved sensitivity provided by red bracts compared to the violet ones.

Bougainvillea extracts displayed interesting photoelectrochemical performance parameters: J_SC close to 2.3 mA/cm²; V_OC near to 0.26 V; an excellent fill factor of 0.74; and η in the neighborhood of 0.49% using an electrolyte composed of 0.1 M I₂/0.05 M LiI/0.05 M 3-methoxypropionitrile in acetonitrile (C₂H₃N). The application of a compact TiO₂ under-layer (i.e., overcrowding layer) was necessary for enhancing the cell performance, as compared with previously reported yields [17,20] (Table 1). The filling factor found in this research is promising because, to our knowledge, it is among the highest so far reported with raw natural dyes.

Additionally, a test on the stability of these natural dyes was carried out by monitoring some indicative parameters, such as J_SC and η, under continuous sun illumination (100 mW/cm² and air mass 1.5) in a hermetically sealed solar cell with an electrolyte solution (0.1 M of I₂, 0.05 M of LiI, 0.05 M of 3-methoxypropionitrile in acetonitrile) and without any cooling system during 36 h; no significant changes were observed in this time as reported in Figure 8.