The two polymorphs of TiO2, rutile and anatase, have been investigated at the ab initio level using different Hamiltonians with all-electron Gaussian and projector augmented plane wave basis sets. Their equilibrium lattice parameters, relative stabilities, binding energies, and band structures have been evaluated. The calculations have been performed at the Hartree-Fock, density functional theory (DFT), and hybrid (B3LYP and PBE0) levels. As regards DFT, the local density and generalized gradient (PBE) approximations have been used. Our results show an excellent agreement with the experimental band structures and binding energies for the B3LYP and PBE0 functionals, while the best structural descriptions are obtained at the PBE0 level. However, no matter which Hamiltonian and method are used, anatase is found more stable than rutile, in contrast with recent experimental reports, although the relative stabilities of the two phases are very close to each other. Nevertheless, based on the overall results, the hybrid PBE0 functional appears as a good compromise to obtain an accurate description of both structural and electronic properties of solids.

Five low-index stoichiometric TiO2 rutile and anatase surfaces, i.e., rutile (110), (100), and (001) as well as anatase (101) and (100), have been investigated using different Hamiltonians with all-electron Gaussian basis sets, within a periodic approach. Full-relaxations of the aforementioned surfaces have been essentially carried out at the Hartree-Fock (HF) level, but selected surfaces were treated also using pure and hybrid Density Functional Theory (DFT) models. Mulliken charges, band structures, and total and projected-densities of states have been computed both at the HF and the hybrid DFT (B3LYP and PBE0) levels. As regards DFT, the local density (LDA) and generalized gradient approximations (GGA) have been used. No matter which Hamiltonian is considered, as long as sufficiently thick slabs are taken into account, computed atomic relaxations show an overall excellent agreement with the most recent experimental reports. This is especially true when using hybrid functionals which enable the clarification of some conflicting results. Moreover, both at the LDA and HF levels, we were able to classify the surface relative energies in the following sequence:  anatase (101) < rutile (110) < anatase (100) < rutile (100) ≪ rutile (001). Instead, when using PBE, B3LYP, or PBE0, the two most stable surfaces are reversed.

Since dye-sensitized solar cells (DSSCs) appeared as a promising inexpensive alternative to the traditional silicon-based solar cells, DSSCs have attracted a considerable amount of experimental and theoretical interest. In contrast with silicon-based solar cells, DSSCs use different components for the light-harvesting and transport functions, which allow researchers to fine-tune each material and, under ideal conditions, to optimize their overall performance in assembled devices. Because of the variety of elementary components present in these cells and their multiple possible combinations, this task presents experimental challenges. The photoconversion efficiencies obtained up to this point are still low, despite the significant experimental efforts spent in their optimization. The development of a low-cost and efficient computational protocol that could qualitatively (or even quantitatively) identify the promising semiconductors, dyes, and electrolytes, as well as their assembly, could save substantial experimental time and resources. In this Account, we describe our computational approach that allows us to understand and predict the different elementary mechanisms involved in DSSC working principles. We use this computational framework to propose an in silico route for the ab initio design of these materials. Our approach relies on a unique density functional theory (DFT) based model, which allows for an accurate and balanced treatment of electronic and spectroscopic properties in different phases (such as gas, solution, or interfaces) and avoids or minimizes spurious computational effects. Using this tool, we reproduced and predicted the properties of the isolated components of the DSSC assemblies. We accessed the microscopic measurable characteristics of the cells such as the short circuit current (J(sc)) or the open circuit voltage (V(oc)), which define the overall photoconversion efficiency of the cell. The absence of empirical or material-related parameters in our approach should allow for its wide application to the optimization of existing devices or the design of new ones.

A theoretical investigation of eosin-Y (EY) loaded ZnO thin films, the basic components of a dye-sensitized solar cell (DSSC), is presented. The EY/ZnO wurtzite (10-10) system has been fully described within a periodic approach using density functional theory (DFT) and a hybrid exchange-correlation functional. Reduced systems were also analyzed to simulate an electron transfer from the dye to the substrate. Injection times from dye to the semiconductor were calculated using the Newns-Anderson approach. Finally, the UV-visible spectra of EY/ZnO films were simulated using a time-dependent DFT approach and compared to that of the EY molecule computed in solution. The results obtained highlight that EY strongly adsorbs on the ZnO substrate contributing significantly to the electronic structure of the adsorbed system. The UV-visible spectral signature of the isolated EY molecule is still found when adsorbed on ZnO but the analysis of Gamma-point crystalline orbitals reveals that a direct HOMO-->LUMO excitation cannot lead to a direct electron injection into the semiconductor, the first unoccupied orbital with contributions from the ZnO substrate being the LUMO + 1. As a consequence, a two photon injection mechanism is proposed explaining the low efficiency of the EY/ZnO solar cells. On this basis, possible strategies for enhancing the cell efficiency are presented and discussed.

No abstract

Density functional theory (DFT) and time-dependent DFT are useful computational approaches frequently used in the dye-sensitized solar cell (DSSC) community in order to analyze experimental results and to clarify the elementary processes involved in the working principles of these devices. Indeed, despite these significant contributions, these methods can provide insights that go well beyond a purely descriptive aim, especially when suitable computational approaches and methodologies for interpreting and validating the computational outcomes are developed. In the present contribution, the possibility of using recently developed computational approaches to design and interpret the macroscopic behavior of DSSCs is exemplified by the study of the performances of three new TiO2-based DSSCs making use of organic dyes, all belonging to the expanded pyridinium family.

No abstract

We have investigated the role of electrolyte composition, in terms of solvent and additive, on the open-circuit voltage (V(oc)) of ZnO-based dye-sensitized solar cells (DSSCs) using a combined experimental and theoretical approach. Calculations based on density functional theory (DFT) have been performed in order to describe the geometries and adsorption energies of various adsorbed solvents (nitromethane, acetonitrile and dimethylformamide) and p-tert-butylpyridine (TBP) (modeled by methylpyridine) on the ZnO (100) surface using a periodic approach. The densities of states (DOS) have been calculated and the energy position of the conduction band edge (CBE) has been evaluated for the different molecules adsorbed. The effect of the electrolyte composition on the standard redox potential of the iodide/triiodide redox couple has been experimentally determined. These two data values (CBE and standard redox potential) allowed us to determine the dependence of V(oc) on the electrolyte composition. The variations determined using this method were in good agreement with the measured V(oc) for cells made of electrodeposited ZnO films sensitized using D149 (indoline) dye. As in the case of TiO(2)-based cells, a correlation of V(oc) with the donor number of the adsorbed species was found. The present study clearly points out that both the CBE energy and the redox potential variation are important for explaining the experimentally observed changes in the V(oc) of DSSCs.