### Refine

#### Institute

CONSPECTUS: Density functional theory (DFT) and its time-dependent extension (TD-DFT) are powerful tools enabling the theoretical prediction of the ground- and excited-state properties of organic electronic materials with reasonable accuracy at affordable computational costs. Due to their excellent accuracy-to-numerical-costs ratio, semilocal and global hybrid functionals such as B3LYP have become the workhorse for geometry optimizations and the prediction of vibrational spectra in modern theoretical organic chemistry. Despite the overwhelming success of these out-of-the-box functionals for such applications, the computational treatment of electronic and structural properties that are of particular interest in organic electronic materials sometimes reveals severe and qualitative failures of such functionals. Important examples include the overestimation of conjugation, torsional barriers, and electronic coupling as well as the underestimation of bond-length alternations or excited-state energies in low-band-gap polymers.
In this Account, we highlight how these failures can be traced back to the delocalization error inherent to semilocal and global hybrid functionals, which leads to the spurious delocalization of electron densities and an overestimation of conjugation. The delocalization error for systems and functionals of interest can be quantified by allowing for fractional occupation of the highest occupied molecular orbital. It can be minimized by using long-range corrected hybrid functionals and a nonempirical tuning procedure for the range-separation parameter.
We then review the benefits and drawbacks of using tuned long-range corrected hybrid functionals for the description of the ground and excited states of pi-conjugated systems. In particular, we show that this approach provides for robust and efficient means of characterizing the electronic couplings in organic mixed-valence systems, for the calculation of accurate torsional barriers at the polymer limit, and for the reliable prediction of the optical absorption spectrum of low-band-gap polymers. We also explain why the use of standard, out-of-the-box range-separation parameters is not recommended for the DFT and/or TD-DFT description of the ground and excited states of extended, pi-conjugated systems. Finally, we highlight a severe drawback of tuned range-separated hybrid functionals by discussing the example of the calculation of bond-length alternation in polyacetylene, which leads us to point out the challenges for future developments in this field.

Long-range corrected hybrid functionals that employ a nonempirically tuned range-separation parameter have been demonstrated to yield accurate ionization potentials and fundamental gaps for a wide range of finite systems. Here, we address the question of whether this high level of accuracy is limited to the highest occupied/lowest unoccupied energy levels to which the range-separation parameter is tuned or whether it is retained for the entire valence spectrum. We examine several pi-conjugated molecules and find that orbitals of a different character and symmetry require significantly different range-separation parameters and fractions of exact exchange. This imbalanced treatment of orbitals of a different nature biases the resulting eigenvalue spectra. Thus, the existing schemes for the tuning of range-separated hybrid functionals, while providing for good agreement between the highest occupied energy level and the first ionization potential, do not achieve accuracy comparable to reliable G(0)W(0) computations for the entire quasiparticle spectrum.

We investigate the torsion potentials in two prototypical pi-conjugated polymers, polyacetylene and polydiacetylene, as a function of chain length using different flavors of density functional theory. Our study provides a quantitative analysis of the delocalization error in standard semilocal and hybrid density functionals and demonstrates how it can influence structural and thermodynamic properties. The delocalization error is quantified by evaluating the many-electron self-interaction error (MESIE) for fractional electron numbers, which allows us to establish a direct connection between the MESIE and the error in the torsion barriers. The use of non-empirically tuned long-range corrected hybrid functionals results in a very significant reduction of the MESIE and leads to an improved description of torsion barrier heights. In addition, we demonstrate how our analysis allows the determination of the effective conjugation length in polyacetylene and polydiacetylene chains.

The electronic coupling between redox sites in mixed-valence systems has attracted the interest of the chemistry community for a long time. Many computational studies have focused on trying to determine its magnitude as a function of the nature of the redox sites and of the bridge(s) between them. However, in most instances, the quantum-chemical methodologies that have been employed suffer from intrinsic errors that lead to either an overlocalized or an overdelocalized character of the electronic structure. These deficiencies prevent an accurate depiction of the degree of charge (de)localization in the system and, as a result, of the extent of electronic coupling. Here we use nonempirically tuned long-range corrected density functional theory and show that it provides a robust, efficient approach to characterize organic mixed-valence systems. We first demonstrate the performance of this approach via a study of representative Robin-Day class-II (localized) and class-III (delocalized) complexes. We then examine a borderline class-II/class-III complex, which had proven difficult to describe accurately with standard density functional theory and Hartree-Fock methods.

Predicting accurate bond-length alternations (BLAs) in long conjugated molecular chains has been a major challenge for electronic-structure theory for many decades. While Hartree-Fock (HF) overestimates BLA significantly, second-order perturbation theory and commonly used density functional theory (DFT) approaches typically underestimate it. Here, we discuss how this failure is related to the many-electron self-interaction error (MSIE), which is inherent to both HF and DFT approaches. We use tuned long-range corrected hybrids to minimize the MSIE for a series of polyenes. The key result is that the minimization of the MSIE alone does not yield accurate BLAs. On the other hand, if the range-separation parameter is tuned to yield accurate BLAs, we obtain a significant MSIE that grows with chain length. Our findings demonstrate that reducing the MSIE is one but not the only important aspect necessary to obtain accurate BLAs from density functional theory.