A theoretical investigation of decorated novel triazoles as DSSCs in PV devices

Some novel metal-free 1,2,4-triazole compounds A1-A8, based on the 3,5-bis(2-hydroxyphenyl)-1,2,4-triazole derivatives were examined for photovoltaic properties using density functional theory (DFT) and time-dependent density functional theory (TD-DFT) calculations to test their suitability as metal-free organic dyes for use in dye-sensitized solar cells (DSSCs). Through deductive logic, the fluorescence emission (Φf) and charge collection (ηc) efficiencies of these compounds as dyes were obtained and used to determine each dye’s incident conversion efficiency (IPCE). From the analyses, A2 displayed the highest IPCE value, followed by A6 and A1. This technique is restricted to evaluating compounds for potential metal-free organic dyes only.


Introduction
The rise in energy demands and the concern over global warming have led to a shift away from fossil fuels in recent years [1]. Renewable energy sources are attractive alternatives since they are naturally replenished continuously. As they do not emit CO 2 and other harmful gases into the atmosphere, they can be referred to as green or clean energy [2]. Recently, photovoltaic (PV) systems that generate power from solar radiation have gained popularity. This technology relies entirely on the photoelectric effect. The first generation of solar cells in PV systems contained silicon and is still primarily used due to its hole transporting and charge carrier mobility [3]. Since fabricating silicon solar cells is not easy and requires a high cost, it is challenging to distribute PV systems. The low cost, facile fabrication, and high power conversion efficiency of dye-sensitized solar cells (DSSCs) make them suitable candidates to replace silicon in PV systems. Moreover, these cells can also function in low light conditions [4]. For a long time, Ru-based complexes held the record for the highest efficiency in DSSCs of over 11%. However, their high cost, non-renewability, and difficulty in purification made them unsuitable for commercial usage. Seeking more worthwhile dyes, researchers stumbled across zinc porphyrin complexes in 2014 with an efficiency of 13.00%. Soon afterwards, a DSSC with two co-sensitized metal-free organic dyes achieved a higher efficiency of 14.30% in 2015 [5]. Thus, it is more suitable to use organic dyes because of their efficiency, lower cost, and simple synthetic procedures. Other advantages that they offer are high molar absorption coefficients and variable structure adjustability [6,7]. A DSSC usually consists of a chemisorbed material onto a mesoporous material, generally titanium dioxide (TiO 2 ). A metal-free organic dye is traditionally divided into donor-linker-acceptor (D-π-A) fragments where intramolecular charge transfer (ICT) occurs from D to A via the π-spacer [8][9][10][11] (Fig. 1). The π-spacer units improve the conjugation and enhance the stability of the resulting photosensitizers [12]. Moreover, they offer a high degree of electron delocalization, allowing substantial electronic communication between individual π-spacer units. This will enable them to mediate charge/energy over long distances [13]. Triaryl amine, starburst, carbazoles, indoles, phenoxides, phenothiazine, and coumarins are commonly used as donor groups [11][12][13][14][15][16][17][18], while widely used π-spacer units are vinylene and thiophene groups [16,18]. Commonly used acceptor groups are cyanoacetic acid, rhodamine-3-acetic acid, barbituric acid, hydroxyl, phosphoric acid, cyanoacrylic acid, and carboxylic acid [16,19,20].
Smooth electron movements between D and the π-spacer groups can be achieved for small dihedral angles between the former and the latter. The π-electrons can then be effectively delocalized onto A if it is approximately co-planar to the π-spacer unit. This then causes excitation onto A, which is subsequently injected into the conduction band edge of TiO 2 . Once injected, π-electrons are transferred onto a thin, transparent conducting oxide (TCO) electrode, usually tin dioxide (SnO 2 ), which acts as an anode. A current can then be obtained by connecting an external circuit between the anode and a counter electrode. The latter is usually indium dioxide (InO 2 ). The electrons can then be regenerated back into the dye via an electrolyte (usually the I − /I − 3 redox couple) [18]. However, it is essential that A and D should be non-planar; otherwise, dye aggregation can occur due to intermolecular interactions, which results in dimers that suppress electron injection into the conduction band edge of TiO 2 . The introduction of an anti-aggregating agent such as chenodeoxycholic acid (CDCA), which can be used with organic dyes, can reduce dye aggregation by enhancing electron transport, which leads to larger power conversion efficiencies [21]. For a high absorption coefficient, D and the π-spacer must also be non-planar [22,23]. Efficiencies are reduced when a light-generated current encounters a hole that recombines and emits a photon. Recombination can also occur when charge carriers encounter a defect in the crystal structure of the solar cell. Both processes cancel out their input into the electric current [3]. The ideal dye should be luminescent, absorb in the visible and near-infrared (NIR) regions; its lowest unoccupied molecular orbital (LUMO) must lie above the conduction band edge of TiO 2 , while its highest occupied molecular orbital (HOMO) must lie below the HOMO of the redox electrolyte. In addition, it should have a high interfacial dipole moment and the periphery of the dye should be hydrophobic (otherwise, interferences with the regeneration of electrons from the electrolyte will occur) [18,24,25].
There is still an ongoing search for dye materials with higher efficiency which can be incorporated into DSSCs [18]. Hence, this study focuses on a theoretical evaluation of 3,5-bis(2-hydroxyphenyl)-1,2,4-triazole derivatives as potential metal-free organic dye agents using DFT and TD-DFT methods (Fig. 2, where hydrogen atoms were omitted for clarity). Our interest in this compound lies in its aromatic groups that contain active sites to build in various D-π-A moieties. Therefore, we decorated this compound further with starburst donor groups, thiophene π-spacer groups, and cyanoacrylic acid acceptor groups (Fig. 3) before carrying out a theoretical evaluation of their suitability as dyes in DSSCs. Although triazoles have shown that they can be used as conjugative π-spacers in intramolecular electron transfer processes [26], to the best of our knowledge, 1,2,4-triazoles have only been used as electron acceptor units in organic light-emitting diodes [27][28][29][30]. Since several π-spacer units containing electron-deficient moieties have been reported, some of which produced efficiencies of 8.70% [5], it was worthwhile to assess electron-deficient 1,2,4-triazole as a π-spacer unit. Herein, we report the structures of various 3,5-bis(2-hydroxyphenyl)-1,2,4-triazole derivatives (Fig. 4) and their potential abilities as metal-free organic dyes in DSSCs.

Computational details
All calculations were carried out with Gaussian 09 rev E01 software [31]. The geometries of all molecules were optimized in the gas phase for both DFT and TD-DFT calculations. All structures were optimized at the M06-2x/6-31G(d,p) level of theory. The optimization process was carried out in conjunction with frequency calculations to verify that structures obtained were of minimum energy. Using their optimized structures as input, we carried out TD-DFT analyses on the organic dyes labelled as A1-A8 (Fig. 3) in the gas phase, and their frontier molecular orbital (FMO) natures were analyzed using multiwfn software [32] by dividing their structures into D-π-A fragments ( Fig. 4) to study their ICT at the same level of theory. In addition to this, molecular orbital structures and energies of their HOMO and LUMO were also obtained from an analysis of the optimized structures.

Methods of calculation
The incident photon conversion efficiency (IPCE), charge collection efficiency (η c ), electron injection efficiency (Φ inj ), and light-harvesting efficiency (LHE) were shown to be related to Eqs. (1) and (2) [33]: where f is the oscillator strength corresponding to the maximum absorption wavelength (λ max ) in the visible and NIR region.
The electron injection free energy (ΔG inj ) can be obtained through Eq. (3) [33]: where E dye * ox and E CB are the dye's excited state oxidation potential and the ground-state reduction potential of TiO 2 , respectively. The literature value of E CB is − 4.21 eV [34], while E dye * ox was calculated using Eq. (4) [33]: where E dye ox is the dye's ground-state oxidation potential given by E dye ox = −HOMO [33] and ΔE is the absorption energy corresponding to λ max in the visible or NIR region.
Since energy from the excited dye (dye * ) is released through diffusion into the TiO 2 conduction band, a diffusion coefficient D π can be defined and is calculated using Stokes' equation [33,34]: T is temperature, η is the viscosity of the medium, which is assumed to be He at 300 K (20.0 × 10 −6 Pas) in this case since the investigations were performed in the gas phase, N A is Avogadro's number, and r dye is the molecular radii of the dye. The latter can be obtained from the following equation [33,34]: where a is Onsager cavity radii, M is the molecular weight of the dye, and ρ is the density of He gas (9.00 × 10 −2 kg.m −3 ) at STP [33,34].
If we assume that the electron injection efficiency (Φ inj ) of the dye is equal to the fluorescence emission factor (Φ f ), then we can define the latter as: where Iε em and Iε abs are the integrated emission and absorption coefficients respectively, corresponding to the areas under the emission and absorption spectra. The former can be obtained through extrapolation of the absorption spectra [33].
The charge collection efficiency (n c ) can then be obtained through the following equation [33]: Fig. 3 Donor, π-spacer, and acceptor groups that were selected for this study where δ p is the potential difference between the LUMO of the dye and the conduction band gap of TiO 2 [33].
A relationship between the open-circuit voltage (V OC ) and LUMO of the dye also exists and is given in Eq. (9) [23]: An important parameter to consider is the driving force for charge recombination Δ G 0 rec which can be obtained through Eq. (10) [23]: where E CBe is the redox potential of the electrolyte I − /I − 3 (-4.6 eV) and E HOMO the redox potential of the dye in its ground state [23].

Optimized geometry
The correctness of the optimized structure of a molecule will determine the accuracy of a predicted property. Optimization of each structure required a series of optimization steps before reaching its minimum energy configuration ( Figure S1), conventionally known as the optimized structure. Frequency calculations were performed to validate the optimized structures as minima or transition state structures ( Figure S2). Theoretically, any structure with one imaginary frequency (i.e., negative vibrational frequency) is designated as a transition state. In contrast, those with two or   Hence, it is generally accepted that a minimum energy structure should have no negative frequency. However, negative vibrational frequencies within the range of 10-100 cm −1 can be ignored for floppy molecules [35]. Since the molecules under investigation show negligible stretches for negative vibrational frequencies, they can all be considered stable structures for the study. Figure 5 shows the parent compound from which all the triazole derivatives under investigation were constructed. Figure 6 shows a graphical representation of the absorption spectra for these compounds investigated for potential metalfree organic dyes. These spectra were obtained from optimized ground-state structures via TD-DFT calculations. To be able to rationalize the transition strengths, the absorption wavelengths (λ max ) were presented against both the absorptivity coefficients (ε) and oscillator strengths (f) in Table 1. An accurate picture of the absorption spectra is revealed by   observing the transition probability per molecule from f, as against ε. The latter depends on the molecule's molecular weight [12], while the former serve as the probability for absorbing a photon [36]. The large f in A1 and A2 indicates their strong transition intensities compared to the other molecules. Thus, the probabilities of absorbing photons during these transitions are higher than the other compounds under study. This induces large ε. The molecules' λ max and ε also reflect on the absorbed incident photon's energy and the absorption's intensity, except that ε considers the molecular weight, while f does not [12]. This could be a possible reason

Electronic and spectral properties
show different trends, which was observed by Sanusi et al. [33] as well. The low dipole moment (μ) in A6 arises from its more symmetrical nature than the other compounds under study. This low dipole moment will cause an electron to easily relax to the ground state upon excitation, rendering charge recombinations. As μ can also be used as a guide for the dissolution of compounds in polar solvents, this compound will also be the least soluble. The interaction between solvent molecules and these compounds in solutions will cause their λ max to shift to higher wavelengths [36]. The nature of solvent media, the incident photon energy, and its pulse width are some of the factors that contribute to the emission of photons as fluorescence by molecules from the first excited state (S 1 ) to the ground state (S 0 ) [37]. In our case, we can ignore these factors in interpreting the results since our study was carried out in silico. Thus, our model only considers the calculated excited state properties that depend on the studied molecules' geometrical configurations. Various excited state processes such as internal conversion, vibrational relaxation, intersystem crossing, and fluorescence emission play a role in deactivating the molecule's excited state energy. Possibilities exist for both highly absorbing and poorly absorbing molecules to fluoresce greatly if deactivation by fluorescence is favored. The other processes quench fluorescence whether the molecule is highly or poorly absorbing [37].
TD-DFT calculated emission spectra for λ max and ε em are depicted in Table 2. A1 and A2 absorb just below the spectral range. Other than A5, no significantly large Stokes' shifts in wavelengths occurred. The negligible shifts in λ max for some compounds can be ascribed to small oscillator strengths observed for the second and third excited states when using multiwfn software to plot their emission spectra. Although the emitted λ max shifted to the visible region in A5, it also displayed a very poor emission spectrum. No emission spectrum was obtained for A3.
Significant differences between absorption and emission spectra are observed in Fig. 7 c, d, f, and g. As A3 does not display an emission spectrum, no comparison between absorption and emission spectra could be drawn for it. Figure 8 shows the relationship between ΔG 0 rec and the energy of the HOMO for the dyes. A negative value for ΔG 0 rec in A7 indicates that when electrons are regenerated, they do not reach the HOMO of the dye. This can be circumvented by replacing the I − /I − 3 electrolyte with Co III/II electrolyte, since the latter has a redox potential of − 5.0 eV [4]. Low Δ G 0 rec for A4 and A5 indicates that their electrons are more easily regenerated than the other compounds. The trend in Fig. 8 suggests that the potential dyes with higher HOMO levels are less inclined to experience charge recombination. Apart from A8, low levels of co-planarity are observed between the donor and π-spacer groups, suggesting that the transfer of electrons within these fragments would not proceed smoothly. The smoothest transfer between π-spacer groups and acceptor groups would occur in A3. It is improbable that dye aggregation from intermolecular interactions would occur for all dye molecules examined due to the large dihedral angles observed between their donor and acceptor groups, with A5 being the outlier. Negative intra-fragment charge transfer (IFCT) for D-π-A in A3 and A6 means that their acceptor groups receive no excess charge from the donor and π-spacer groups. This observation is also verified by the low μ observed in the case of A6. The high IFCT in A4 indicates a significant transfer of charge to its acceptor fragment. Table 3 shows sizeable HOMO-LUMO energy gaps within all the compounds except for A7. In the case of A1, both HOMO and LUMO are localized over the triazole ring, one of the phenyl rings, and the acceptor group. When adding the donor moiety to produce A2, the π electrons are localized over the donor and π-spacer groups and delocalized over the acceptor group. As A3 contains α and β HOMO and LUMO, in the absence of strong donor groups, its π-electrons are localized over the triazole ring. In contrast, its π-electrons are then delocalized over only one of its three acceptor groups. None of the π-electrons is localized over the two donor groups in A4, although they are delocalized over the acceptor group. Localization of π-electrons occurred in only one of the two donor groups in A5 and delocalized to the acceptor group. A similar phenomenon occurred in A6. The π-electrons were localized over the π-spacer group rather than the donor group in A7, although they were delocalized to the acceptor group.

Photophysicochemical properties
Although several studies suggested energy gaps of below 4 eV between HOMO and LUMO as a criterion for an electron to be excited in a molecule [18,20,25,33], a recent study by Bulat et al. revealed that the most energetic electrons are not necessarily located in the HOMO of a molecule as the contour of the molecule's electronic density and an electron's position within atomic space also plays a role [38]. The implication is that although we use the HOMO-LUMO energy gap and the spatial distribution characteristics of these orbitals here as one of the criteria to determine the suitability of compounds A1-A8 as dyes in DSSCs, experimental testing of their performance might reveal some discrepancies from the trends observed. This also explains the observation that some π-electrons within the donor fragments are not localized over HOMO orbitals, and other π-electrons within acceptor fragments delocalize over LUMO orbitals as observed in Fig. 9.

Photovoltaic properties
The estimated molecular radii (r dye ) and the diffusion coefficients (D π ) in Table 4 are consistent with literature values [33], verifying valid assumptions and accurate methods of calculations. An inversely proportional relationship between D π and the size of the dye (Fig. 10) reveals its expected dependence on the molecular dye radii [33]. The low D π for A4 indicates that electrons diffuse quite slowly towards the conduction band of TiO 2 , which provides a plausible indicative of the smoothest transfer of electrons between the π-spacer and the acceptor fragment. The large D-A dihedral angle between the donor and acceptor fragment in A6 is an indication that it is less prone to aggregation than the other compounds. To generate electronic absorption and emission spectra, time-dependent density functional theory (TD-DFT) calculations were performed. From these calculations, the highest occupied molecular orbital (HOMO), the lowest unoccupied molecular orbital (LUMO) concerning the TiO 2 conduction band (δ p ), pielectron diffusion coefficients (D π ), intra-fragment charge transfer (IFCT), and molar absorptivity (ε) and emission (ε em ) coefficients at maximum absorption and emission wavelengths were determined. As IFCT is an indicator of the number of electrons being transferred to the acceptor fragment of the compound, the high IFCT in A4 indicates that more electrons are transferred to the acceptor fragment within this compound than the other compounds. The high D π in A1 shows the most rapid diffusion of electrons to the conduction band edge of TiO 2 . Through these parameters, the fluorescence emission (Φ f ), free energy injection (ΔG inj ), charge recombinations (Δ G 0 rec ) and charge collection (η c ) efficiencies were obtained. A high Φ f was displayed in A1. The high ΔG inj in A2 means more energy is available for injection. A high Δ G 0 rec in A4 indicates a high success rate of electrons being regenerated in this compound, while a negative value in A7 indicates explanation for its poor ε em . . The ΔG inj values favor the electron injection process into the E CB edge, which is also suggested by the δ p values. From Table 4, a relationship between ΔG inj and δ p can be established where the sign between the latter and the former alternates. At the interface between the acceptor fragment and the conduction band edge of TiO 2 , the charge collection efficiency (η c ) can determine the probability of an electron's availability before injection. It is expected that more available electrons should produce a higher incident photo conversion efficiency (IPCE), which will subsequently cause a greater efficiency when being used as a DSSC. The low n c for A4 further support our observations for ε em and D π . Since A3 has no emission spectrum, no IPCE is observed for it.

Conclusions
Highly conjugated novel metal-free 1,2,4-triazole compounds were evaluated for applications as metal-free organic dyes in dye-sensitized solar cells (DSSCs). A density functional theory (DFT) method was used to study their photophysicochemical and photovoltaic properties. The smallest D-π dihedral angle in A8 indicates that the smoothest transfer of electrons between the donor and π-spacer fragment should occur within this compound. In contrast, the smallest π-A dihedral angle in A3 is × that electrons will not be successfully regenerated unless Co III/II electrolyte is used instead of I − /I − 3 electrolyte due to lower redox potential for the former. The high n c in A2 means that it has the most readily available photoelectrons for injection. The dyes' incident photon conversion efficiency (IPCE) was then obtained by multiplying LHE, Φ f , and η c values with each other. As IPCE is the critical parameter to determine the efficiency of a dye, the IPCE value of A2 indicates that it is the most suitable metal-free organic dye. However, it absorbs slightly below the visible region, followed by A6 and A1, while A3 showed no dye characteristics. Although this method is helpful to predict photovoltaic activities of molecules as a guide for designing sensitive dyes without having to do the laborious trial and error method in laboratory synthesis, it cannot predict the efficiencies of the dyes as no DFT calculations were performed on a selected TiO 2 cluster. Thus, the study was restricted to evaluate the properties of selected compounds for metal-free organic dyes only.