Choosing the Right Functional Reshapes Predictions for Two Organic Semiconductor Models

A single line of computer code can quietly change the answer to one of computational chemistry’s most practical questions: how strongly does a candidate organic semiconductor absorb light, and where do its electrons sit? A new open-access study published in Discover Electronics on 30 September 2026 puts that question to a direct test. Researchers M. El Masfioui, S. Bahsine, A. Elbiyaali and A. Waqdim compared four widely used exchange-correlation functionals from density functional theory against two very different molecular models of organic electronic materials, and found that the choice of functional shifts predicted energy gaps by more than a full electron volt and absorption wavelengths by tens of nanometres.

The two molecular systems were deliberately chosen as contrasting cases. Poly(N-vinylcarbazole), known as PVK, is a widely used hole-transport polymer valued for its thermal stability and film-forming ability; here it was represented by a single N-vinylcarbazole repeat-unit model, an asymmetric and polar carbazole-based fragment. Its partner was p-terphenyl, a rigid, nearly planar three-ring aromatic molecule that is highly symmetric and nonpolar, frequently used as a reference chromophore. By running both systems through an identical computational protocol, the authors could separate differences caused by the electronic-structure method from differences caused by molecular symmetry and electronic delocalization.

The calculations used Gaussian 09 with the 6-31G(d) basis set throughout, applying four functionals: the generalized gradient approximation functional PBEPBE and the global hybrids B3LYP, B3PW91 and MPW1PW91. Geometry optimizations were verified as true minima by the absence of imaginary frequencies in harmonic vibrational calculations. Ground-state electronic structures came from DFT, while vertical optical excitations were computed with time-dependent DFT. Selected B3LYP calculations were repeated with the larger 6-311G(d) basis set as a limited sensitivity check, though the authors stress this was not a systematic convergence study and that the absence of diffuse functions remains a recognized limitation for response properties.

The headline numbers are striking. Calculated HOMO-LUMO gaps spanned 3.28 to 5.05 electron volts for the PVK monomer model and 2.99 to 4.64 electron volts for p-terphenyl, while the principal absorption wavelengths spanned 284 to 334 nanometres and 301 to 340 nanometres respectively. For p-terphenyl, PBEPBE gave the smallest gap at 2.9864 electron volts, with the three hybrid functionals clustered between 4.1753 and 4.6374 electron volts. For PVK the ordering flipped: B3LYP and B3PW91 predicted gaps of 4.6794 and 4.6216 electron volts, close to an optical-gap reference of roughly 4.7 electron volts, while PBEPBE gave 5.0488 electron volts and MPW1PW91 only 3.2784 electron volts. The authors conclude that functional performance cannot be inferred from functional class alone.

Against approximate literature reference gaps of 3.97 electron volts for p-terphenyl and 4.70 electron volts for PVK, B3LYP emerged as the most balanced of the four methods, with the lowest combined mean absolute error of 0.113 electron volts, followed by B3PW91 at 0.206 electron volts. For absorption wavelengths, benchmarked against approximate literature values of 300 nanometres for PVK and 315 nanometres for p-terphenyl, B3LYP again led with a mean absolute error of 7.94 nanometres, ahead of B3PW91, PBEPBE and MPW1PW91. The authors are careful, however, to label these figures descriptive benchmarks rather than formal validation, because Kohn-Sham orbital gaps, isolated-molecule TD-DFT excitations and condensed-phase experimental measurements are not equivalent observables.

The electric-field response calculations revealed an even clearer contrast rooted in molecular symmetry. The PVK monomer model showed a total dipole moment of 1.12 to 1.54 debye across the protocols, an isotropic polarizability ranging from 166.15 to 189.15 atomic units at the common basis set, and a first hyperpolarizability between 11.19 and 17.35 atomic units. p-Terphenyl, by contrast, had a dipole moment of exactly zero for every method, a polarizability of only 96.24 to 100.68 atomic units, and a hyperpolarizability of at most 7 times 10 to the minus 4 atomic units, numerically zero for most methods. The near-cancellation of tensor components in p-terphenyl reflects its inversion symmetry, while the finite response of PVK stems from its asymmetric charge distribution. Depending on the functional, the PVK model was roughly 1.7 to 1.9 times more polarizable than p-terphenyl.

A particularly instructive result came from the limited basis-set check. Simply enlarging the basis from 6-31G(d) to 6-311G(d) in B3LYP calculations changed the PVK polarizability by about 7.4 percent, the dipole moment by about 4.8 percent, and the first hyperpolarizability by a dramatic 51.9 percent, from 11.42 to 17.35 atomic units. This confirms that first hyperpolarizabilities are exceptionally sensitive to basis-set completeness, and the authors treat the larger-basis values strictly as a sensitivity test rather than evidence of convergence. They also decline to compare PVK against standard nonlinear-optical reference molecules such as urea or para-nitroaniline, since no same-level reference calculations were performed, and explicitly avoid classifying PVK as a high-performance nonlinear-optical material.

The study also explored the relationship between the electronic gap and the optical absorption. Across the four common 6-31G(d) results for each system, the correlation between calculated gap and principal wavelength was negative, with a Pearson coefficient of minus 0.919 for PVK and minus 0.993 for p-terphenyl. With only four data points per system, the authors confine these coefficients to descriptive summaries and warn against reading them as a universal predictive equation. Notably, the functional ordering of wavelengths differed between the two molecules: PBEPBE gave the longest PVK wavelength but the shortest p-terphenyl wavelength, while MPW1PW91 did the reverse, reinforcing the conclusion that functional accuracy is both property-dependent and system-dependent.

Extending the molecular results toward bulk properties, the team converted isolated-molecule polarizabilities into estimated dielectric constants and refractive indices using the Clausius-Mossotti relation with fixed effective molecular volumes of 301.6 and 311.8 cubic angstroms for the PVK model and p-terphenyl. For PVK, the estimated relative dielectric constant ranged from 2.56 to 3.16 and the refractive index from 1.60 to 1.78; B3LYP with 6-31G(d) gave values within 0.3 and 1.2 percent of literature bulk figures of 2.90 and 1.69. For p-terphenyl the estimates were much narrower, at 1.71 to 1.75 and 1.31 to 1.32. The authors emphasize that these are molecular-to-bulk indicators based on assumed volumes, not predictions of macroscopic dielectric properties, and that the apparent agreement of one functional should not be read as rigorous validation.

The study’s limitations are stated candidly. PVK was modelled as a single monomer, so chain-length-dependent conjugation, conformational disorder, polymer packing and solid-state polarization were all outside scope. No range-separated functional such as CAM-B3LYP or omega-B97X-D was included, no diffuse functions were used, no quantitative charge-transfer analysis was performed, and only approximate literature peak positions, not matched experimental spectral curves, were available for comparison. The TD-DFT calculations described isolated molecules without solvent, crystal environment or excited-state geometry relaxation. The authors frame their results as comparative molecular trends rather than quantitatively converged bulk predictions, and identify range-separated functionals, diffuse basis sets and oligomeric or periodic PVK models as essential next steps before any device-level claims about OLEDs, photovoltaics or nonlinear optics could be justified. Within those bounds, the message for anyone screening organic optoelectronic materials computationally is unambiguous: the functional you pick, and the molecule you apply it to, jointly determine what your simulation will tell you.

Subject of Research: Functional dependence of calculated electronic and optical properties of PVK monomer and p-terphenyl models in DFT and TD-DFT

Article Title: Effects of exchange correlation functionals on electronic and optical properties of PVK and p-terphenyl models

Article References: Masfioui, M. E., Bahsine, S., Elbiyaali, A., & Waqdim, A. (2026). Effects of exchange correlation functionals on electronic and optical properties of PVK and p-terphenyl models. Discover Electronics, 3(1), Article 139. https://doi.org/10.1007/s44291-026-00293-w

Image Credits: AI Generated

DOI: 10.1007/s44291-026-00293-w

Keywords: DFT, TD-DFT, organic semiconductors, PVK, p-terphenyl, exchange-correlation functionals, B3LYP, HOMO-LUMO gap, polarizability, hyperpolarizability, UV-visible absorption, computational chemistry