A Protocol for Fast Prediction of Electronic and Optical Properties of Donor–Acceptor Polymers Using Density Functional Theory and the Tight-Binding Method

Murat Mesta, Jin Hyun Chang, Suranjan Shil, Kristian S. Thygesen, J. M. García‐LastraThe Journal of Physical Chemistry A 2019 · DOI

TL;DR

This paper introduces a new, computationally efficient method to predict the electronic and optical properties of donor-acceptor (D-A) polymers, crucial for organic solar cell development. By combining density functional theory (DFT) with a tight-binding (TB) model, they can accurately estimate properties for polymers of any length, significantly speeding up the search for optimal materials.

What Problem Does This Paper Address?

Organic solar cells (OSCs) hold great promise due to their flexibility and low cost, but their power conversion efficiency still lags behind traditional inorganic photovoltaics. Improving OSC performance hinges on optimizing the electronic and optical properties of the light-absorbing materials, often D-A polymers. These polymers are built from repeating donor and acceptor units, and their electronic structure (specifically the HOMO and LUMO energy levels) dictates their light absorption and charge transport capabilities.

The challenge lies in the sheer number of possible D-A polymer structures (estimated to be over 10^60). Experimentally testing each one is infeasible, and traditional DFT calculations for long polymer chains are computationally very expensive, hindering large-scale screening efforts. This paper aims to bridge this gap by developing a method that can quickly and accurately predict these crucial properties for a vast number of D-A polymer candidates.

Methods

The core of the proposed method is a tight-binding (TB) model that leverages the weak electronic coupling between repeating monomer units in D-A polymers. This weak coupling allows for an accurate description of the polymer's electronic structure using only information from a monomer and a dimer.

Here's a breakdown of the key methodological components:

  • Density Functional Theory (DFT) Calculations: Standard DFT calculations are used to obtain the fundamental electronic properties (HOMO and LUMO energies, ionization potential (IP), and electron affinity (EA)) for the basic building blocks: monomers and dimers.
  • B3LYP functional with TZVP basis set: Used for ground-state geometry optimizations and to obtain accurate energy levels, serving as a benchmark.
  • BLYP functional with a similar basis set: Used for a larger set of calculations due to its lower computational cost, enabling the estimation of B3LYP-level accuracy.
  • Tight-Binding (TB) Model: This model uses the DFT-derived energies of monomers and dimers as input. It assumes a simple hopping mechanism between adjacent monomer units. The key idea is that the energy levels of longer oligomers (and thus infinite polymers) can be extrapolated from the monomer and dimer data.
  • Relationship between BLYP and B3LYP: The authors establish a strong linear correlation between the orbital energy shifts (β values) calculated with BLYP and B3LYP functionals. This allows them to estimate B3LYP-level accuracy for dimer properties using less expensive BLYP calculations, significantly reducing computational cost.
  • Quasiparticle Gap Calculation: Two methods are employed:
  • Kohn-Sham (KS) gap: The direct difference between LUMO and HOMO energies from DFT.
  • ΔSCF gap: Calculated from IP and EA values obtained via ΔSCF calculations, which better accounts for electron-hole interactions.
  • Optical Gap Calculation: This is determined using the ΔSCF method, specifically accounting for vertical excitation from HOMO to LUMO using the ground-state geometry. This method explicitly includes electron-hole interaction.
  • Singlet vs. Triplet Excitations: Both singlet and triplet excited states are considered for optical gap calculations. A systematic deviation of ~0.3 eV is found between triplet and singlet excitation energies, allowing for estimation of singlet energies from triplet calculations.

Datasets

The study utilizes a dataset of 104 distinct conjugated donor-acceptor (D-A) polymers. These polymers are constructed by combining 8 different donor moieties and 13 different acceptor moieties. The paper does not introduce a new benchmark dataset but rather uses this collection of polymers to validate their proposed computational protocol.

Key Insights and Ideas

  • Weak Coupling is Key: The alternating donor-acceptor structure in these polymers leads to weak electronic coupling between monomer units. This is the fundamental property that makes the tight-binding model so effective, as it simplifies the problem by focusing on nearest-neighbor interactions.
  • Size Independence of Optical Gap: Due to the weak coupling, the optical gap of these D-A polymers is largely independent of chain length. This means the optical gap of a very long or infinite polymer can be accurately predicted from the properties of a single monomer.
  • Bridging Accuracy and Speed: The paper demonstrates a practical way to achieve high accuracy (comparable to more expensive B3LYP and ΔSCF methods) while drastically reducing computational cost. This is achieved by using cheaper BLYP calculations to estimate B3LYP results and by employing the TB model for extrapolating to longer chains.
  • ΔSCF vs. KS Gap: The ΔSCF quasiparticle gap (derived from IP and EA) is found to be a more reliable indicator of the true electronic gap compared to the simpler KS gap, especially when considering the condition that the quasiparticle gap should always be larger than the optical gap.
  • Exciton Binding Energy Matters: The optical gap is consistently lower than the quasiparticle gap due to exciton binding energy. This effect is significant in D-A polymers (0.3-1.0 eV) and needs to be accounted for when comparing these values.

Key Results

  • Accurate Prediction of Properties: The TB model, combined with DFT calculations, accurately predicts HOMO/LUMO energies, IP, and EA for oligomers of varying lengths, including infinite polymers. The mean absolute error for HOMO/LUMO energies in trimers and tetramers was less than 50 meV.
  • Computational Speedup: The proposed protocol significantly reduces computational cost. By leveraging BLYP calculations to estimate B3LYP results and using the TB model, the cost can be reduced by an order of magnitude, making high-throughput screening feasible.
  • Correlation between Functionals: A strong linear correlation (R=0.97) was found between β values calculated using BLYP and B3LYP functionals, with B3LYP values being approximately 1.2 times larger. This allows for the estimation of B3LYP-level accuracy from BLYP calculations.
  • Singlet-Triplet Gap: A consistent blue-shift of ~0.3 eV was observed between triplet and singlet excitation energies, enabling the estimation of singlet optical gaps from triplet calculations.
  • Optical vs. Quasiparticle Gap: The optical gap is consistently smaller than the ΔSCF quasiparticle gap, with exciton binding energies ranging from 0.20 to 1.08 eV for the polymers studied.

Limitations and Future Work

  • Rigid Polymer Assumption: The TB model assumes a rigid polymer structure, meaning the geometry does not change significantly with chain length. The authors acknowledge that side chains (neglected in this study) could indirectly affect polymer-polymer distances and orientations, influencing properties.
  • SIE in GGA Calculations: The paper notes that Self-Interaction Error (SIE) in GGA calculations can lead to an underestimation of the quasiparticle gap. While hybrid functionals like B3LYP alleviate this, it remains a factor.
  • Predicting Optical Gap from KS Gap: The study found that reliably predicting the ΔSCF optical gap from the KS gap is not straightforward due to systematic errors and the exciton binding energy.
  • Future Work: The authors propose that this protocol can be used to accelerate future large-scale screening studies of D-A polymers. They suggest using the TB model for energy level alignment and the 0.3 eV bias for triplet excitation energies to predict singlet excitation energies. They also mention that machine learning techniques can be further trained using the data generated by this protocol.

Summary updated

Abstract

The ability of donor-acceptor (D-A) type polymers to control the positions of the highest occupied (HOMO) and lowest unoccupied (LUMO) molecular orbitals makes them a popular choice for organic solar cell applications. The alternating D-A pattern in a monomer leads to a weak electronic coupling between the constituent monomers within the polymer chain. Exploiting the weak electronic coupling characteristics, we developed a method to efficiently calculate (1) the electronic properties and (2) the optical gap of such polymer chains. The electronic properties (HOMO and LUMO energies, ionization potential, electron affinity, and quasiparticle gap of an oligomer of any length up to an infinitely long polymer) of the D-A polymers are predicted by combining density functional theory calculation results and a tight-binding model. The weak electronic coupling implies that the optical gap of the polymer is size-independent, and thus, it can be calculated using a monomer. We validated the methods using a set of 104 polymers by checking the consistency where the electronic gap of a polymer is larger than the optical gap. Furthermore, we establish relationships between the results obtained from more accurate, yet slower methods (i.e., B3LYP functional, singlet-ΔSCF) with those obtained from the faster counterparts (i.e., BLYP functional, triplet-ΔSCF). Leveraging the found relationships, we propose a way in which the electronic and optical properties of the polymers can be calculated efficiently while retaining high accuracy. The use of the tight-binding model combined with the approach to estimate more accurate results based on less expensive simulations is crucial in the applications where a large volume of computations needs to be carried out efficiently with sufficiently high accuracy, such as high-throughput computational screening or training a machine-learning model.
BibTeX
@article{mesta2019protocol,
  author = {Murat Mesta and Jin Hyun Chang and Suranjan Shil and Kristian S. Thygesen and J. M. García‐Lastra},
  title = {{A Protocol for Fast Prediction of Electronic and Optical Properties of Donor–Acceptor Polymers Using Density Functional Theory and the Tight-Binding Method}},
  journal = {The Journal of Physical Chemistry A},
  year = {2019},
  doi = {10.1021/acs.jpca.9b02391},
}
Figure