The APG molecule (412 mg) was isolated as a yellow amorphous powder. The ¹H NMR and ¹³C NMR spectra exhibited signals due to aromatic systems and sugar moieties. The ¹H NMR spectrum of APG showed a pair of ortho-related protons, indicating the AA'BB' aromatic ring system due to ring B of flavonoid. The ¹H NMR spectrum of APG suggests one flavone unit with signals corresponding to many sugar moieties and one methoxy (-OCH₃) group. It also exhibited one down field peak at δ_H 13.33 ppm for chelated-OH, a pair of doublets (J = 8.8 Hz each) indicating para substituted ring B, and a singlet at δ 6.79 ppm due to H-3 on ring C. We observed another singlet at δ 6.48 ppm, which was assigned to H-6. A singlet observed at δ_H 3.87 ppm confirmed the presence of a methoxy (−OCH₃) group and was assigned to position 4′ on ring B of the flavone unit. The anomeric proton of the first sugar unit appeared as a broad signal at δ_H 4.87 ppm (J = 7.2 Hz, indicating β-configuration), while the remaining sugar protons appeared between δ 3.13 and 4.20 ppm. The ¹³C NMR spectrum of this dye revealed two separate singlets corresponding to C-3 and C-6 positions of flavone nucleus, indicating a substituted C-8. The absence of meta-related coupling suggested a C-8 substituted flavone. Correlation between the anomeric proton δ_H 4.92 of the first glucose sugar unit and C-6 (δ_C 104.9) of the aglycone confirmed the attachment of the anomeric sugar carbon to the C-6 position of the flavone nucleus. Based on the 1D-NMR data and comparison of the data given in the literature (Nawwara et al., 2014), it could be concluded that the structure of APG is a 4′-methoxy derivative of vitexin (apigenin-4′-methylether-8C-glucopyranoside). The proton and carbon-13 NMR results have been summarized in Table 1.

It has been reported that specific functional groups are required for dyes to be effectively adsorbed onto the TiO₂ thin film (Chang et al., 2013). A previous study by Ahmad and Nafarizal (2010) also reported that some functional groups such as hydroxyl groups (–OH) and carbonyl groups (–CO) are important in providing points of attachment to the TiO₂ surface. Table 2 shows the FT-IR spectra with diagnostic absorption bands within the wave band of 4,000–400 cm⁻¹. The FT-IR of the APG isolated from the Hibiscus rosa-sinensis leaf extract revealed the presence of CH₃ and CH₂ vibrations at 2,930 and 2,634 cm⁻¹, respectively. Moreover, vibrations of C=O at 1716 cm⁻¹, C–O at 1,043 cm⁻¹, C=C at 1,651 cm⁻¹, and O–H at 3,311 cm⁻¹ were also observed. Calculated IR data for the APG is presented in Figure 2 alongside the measured data. The computed IR data are in good agreement with the experiment, suggesting that the theoretical method adopted is suitable in the description of the molecular structure.

Initially, it was thought that the experimental photovoltaic parameters (J_sc and V_oc) and the overall efficiencies (%PCE _expt and %η _global) of the fabricated APG cells would increase with increasing dye concentration. Interestingly, what was observed was different from this expectation (Table 3). There was no clear trend in the observed photovoltaic data with respect to concentration. To explain this observation, a parabolic model, from which could be obtained certain physical parameters that may be useful in understanding the underlining photoelectrical processes, was proposed (Eq. 1). J eff = γ + β V − δ V α , (1) V eff = β V − δ V α . (2)

The representative experimental I-V profiles for the APG-based cells are presented in Figure 3, with the solid line representing the fitting curve. The graphs were fitted to the parabolic function described in Eq. 1. The model assumes that the effective potential (V_eff), Eq. 2, is a function of some intrinsic potential V, and some variable factors, α , β , and δ . The intrinsic potential V has been assumed to depend on dye electron injection efficiency (φ _inj), velocity of charge transport on the semiconductor medium (ω), and the amount of dye that is available (n) for sensitization. Both J eff and V eff in this case are unit-less values that describe charge and force factors, respectively. The variables, α , β , γ , and δ , which were obtained by fitting Eq. 1 to the experimental I-V curves (Figure 3), may be described as the diffusion, hole–electron recombination, charge transfer, and surface defect factors, respectively, with all of them influencing the overall efficiencies of the DSSCs (%PCE and %η _global). γ was found to be 1 × 10⁶ fold higher in magnitude than the experimental short circuit current (J_sc) for each of the cells 1–5. The diffusion factor α was observed to increase proportionally with the dye concentration (Table 3). The maximum effective charge factor (J_max,eff) was estimated by assuming that the intrinsic potential has a magnitude equivalent to that of V_oc, that is, the experimental V_oc was taken as V in Eq. 1. J_max,eff values were found to vary with α , β , γ , and δ . The cell’s overall efficiencies were found to depend on the J_max,eff values (Table 3). As indicated in Table 3, cell 4 composing of 4.2 × 10⁻² mol dm⁻³ of APG produced the highest J_max,eff value and global efficiency of 0.061 and 1.98%, respectively. A slight decrease in the value of β was also found to result in a significant increase in the J_max,eff values (Table 3).

The fit factors α , β , γ , and δ , especially, were shown to affect the power conversion (%PCE), and the global (%η _global) efficiency values of the cells which have generally followed a noticeable trend for which the values in cell 2 < cell 5 < cell 1 < cell 3 < cell 4. It is worthy of note that cell 5 which contained the highest amount of dye gave a δ value of 3.744. This value is the highest of the five cells (Table 3), suggesting that it has the highest degree of surface defects, which could explain why it has one of the lowest power efficiencies. This high degree of surface defects might have resulted from the aggregation of dye molecules due to increasing dye concentration. With these fitting parameters, it may now be possible to explain I-V curves of different DSSCs.

The observed electronic absorption property of the APG in Ethanol (EtOH) is depicted in Figure 4 along with those obtained by TD-DFT methods in the gas phase, EtOH, and acetonitrile (AcCN). The observed spectrum showed a broad absorption band covering between 320 and 350 nm and peaked at ∼340 nm (λ_max). The computed solution-phase electronic spectra showed similar pattern with the experiment but with slight blue-shifting compared to the latter. The agreement between the solution-phase electronic properties obtained via computation and the one by experiment indicates that the computational method employed is suitable. Solvent effects are noticeable in the computed λ_max when comparing APG in gas and solution phases (Figure 4). The superimposed computed spectra of APG in EtOH and AcCN are red-shifted relative to the gas phase.

It is understood that the lack of sufficient chromophoric groups in the isolated APG would put it at disadvantage for the proposed application, which requires high extinction coefficient over a wide spectral range, covering the visible to the near infrared region. Nonetheless, APG was an ideal choice of sample to validate the previous theoretical model for the prediction of dye’s photovoltaic properties prior to the use of the dye (Sanusi et al., 2019). On the other hand, the BD which has shown relatively higher extinction coefficients (Nazeeruddin et al., 2001), spanning from the visible to the infrared region, is expected to show superior photovoltaic responses compared to APG, and so could serve as a suitable positive reference to validate the theoretical method (Sanusi et al., 2019). The experimental photovoltaic data showed that BD is in manifold better as a photosensitizer than the APG, with the overall efficiency (%η _global) value being 5.25 times higher than that of the best APG-based cell (cell 4) (Table 3).

The computed absorption–emission spectral curves presented in Figure 5 show the longer spectrum range covered by both absorption and emission curves of BD when compared to APG in the three media—gas, EtOH, and AcCN. The calculated photovoltaic parameters, which include light-harvesting (LHE), electron injection efficiency (φ _inj), and incident photon conversion efficiency (IPCE) are mostly higher in BD than APG in the three media, except in the gas phase where the charge collection efficiency (η _c) is higher for APG. The δ _p term (potential gap) which is excessively high for BD in the gas phase could be responsible for the lower η _c. It may thus imply that the excited state BD is highly unstable in the gas phase, but much more stable in solution (Table 4). Overall, the conclusions from the computed data (Table 4) that BD is a better sensitizer for DSSCs do compare favorably with those of experiments (Table 3). This study, by comparing the experimental I-V data of APG and BD, to those predicted by our previously reported theoretical method, has confirmed the validity of the theoretical method. The method predicted that BD is a better sensitizer compared to APG, just as the experiments have done for the considered media phases, except in the gas phase. It is, however, worthy of note that there is no experimental gas phase photovoltaic data for these two sensitizers yet; hence, it would be impossible to know the validity of the theoretical result that shows APG as a better sensitizer in terms of η _c in the gas phase. The free-energy of injection (∆G_inj) is generally negative for both dyes in the three media, suggesting that their electron injection from the LUMO to the TiO₂ CB edge would be spontaneous (Table 4).