Multiscale Quasiparticle Electronic Structure and Excitonic Properties of CdSe Nanoclusters

Technical Summary: Multiscale Quasiparticle Electronic Structure and Excitonic Properties of CdSe Nanoclusters

Problem Statement
Accurately describing excited states in semiconducting nanoclusters, particularly cadmium selenide (CdSe), presents a significant theoretical challenge. While wave-function-based quantum chemistry methods are accurate for small systems, their computational cost becomes prohibitive as cluster size increases. Conversely, standard Density Functional Theory (DFT) and Time-Dependent DFT (TDDFT) often fail to capture the complex interplay between quasiparticle self-energies and excitonic interactions under severe quantum confinement. Historically, many-body perturbation theory (MBPT) studies using the GW/Bethe-Salpeter equation (BSE) formalism on CdSe nanoclusters reported systematic blue-shifts of 1–2 eV compared to experimental data. It remained unclear whether these discrepancies arose from deficiencies in the electron–hole interaction kernels or from insufficient numerical convergence in the single-particle GW calculations. Furthermore, there is a need for scalable models capable of predicting optical properties for realistic nanostructures containing thousands of atoms, which are currently inaccessible to fully ab initio workflows.

Methodology
The authors employed a multiscale approach combining first-principles many-body perturbation theory with a parameterized atomistic tight-binding (TB) framework.

• First-Principles Calculations: Ground-state properties were calculated using DFT (PBE functional) within the Quantum ESPRESSO package. Many-body corrections were applied using the WEST code. Specifically, single-shot $G_0W_0$ calculations were performed using the projective dielectric eigen-decomposition (PDEP) technique to ensure high numerical convergence (approx. 10 meV) without explicit summation over empty states. Optical excitations were determined by solving the Bethe-Salpeter equation (BSE) within the Tamm-Dancoff approximation.
• Systems Studied: The study focused on a representative size series of stoichiometric $Cd_nSe_n$ nanoclusters ($n = 3, 6, 13, 33$), ranging from molecular rings to wurtzite-like nanocrystals.
• Scalable Model: An 8-band atomistic TB model was parameterized from bulk DFT calculations and validated against the GW/BSE results. A rigid "scissor" operator was applied to align the bulk band gap with the experimental value (1.76 eV).
• Analysis Tools: The inverse participation ratio (IPR) was used to analyze state localization, and transition dipole moments (TDM) were calculated to determine optical selection rules.

Key Results

• Resolution of Historical Discrepancies: The study demonstrates that the previously reported 1–2 eV blue-shifts in MBPT predictions are artifacts of single-particle GW convergence rather than deficiencies in the electron–hole kernels. With rigorous convergence, the calculated exciton binding energies ($E_b$) and quasiparticle gaps ($\Delta_{QP}$) align well with experimental trends and previous theoretical benchmarks.
• Breakdown of Cancellation: In the smallest clusters (e.g., $Cd_3Se_3$), there is a near-perfect cancellation between the large quasiparticle correction ($\Delta_{QP} \approx 4.21$ eV) and the large exciton binding energy ($E_b \approx 4.07$ eV) due to the collapse of dielectric screening. As cluster volume increases, this compensation breaks down. The exciton binding energy attenuates faster than the quasiparticle correction due to the onset of bulk-like dielectric screening, leading to a systematic divergence between mean-field DFT gaps and optical onsets in larger nanostructures.
• Origin of Optical Suppression: Spatial analysis via IPR reveals that the optical suppression of fundamental pre-peaks in smaller clusters stems from a severe spatial mismatch: valence orbitals (HOMO/HOMO-1) are highly localized on the surface, while conduction states (LUMO) are delocalized in the core. This mismatch quenches the transition dipole moments, rendering the lowest energy transitions optically "dark." As clusters grow, symmetry breaking and increased orbital mixing restore spatial overlap, leading to broader, bulk-like absorption profiles.
• Validation of the TB Model: The scissor-corrected, DFT-parameterized TB model accurately reproduces the confinement-induced scaling of the quasiparticle gap and the optical onset. While the independent-particle TB model captures the general spectral features, it requires a rigid shift by the exciton binding energy to align with GW/BSE results, highlighting the necessity of including electron–hole correlations for higher-energy excitations.

Significance and Claims
The paper claims to provide a quantitative multiscale roadmap for embedding effective many-body effects into computationally efficient models. By validating that the historical discrepancies in the literature were due to convergence issues rather than theoretical flaws, the work establishes the reliability of the GW/BSE approach for zero-dimensional systems. Furthermore, it demonstrates that a properly parameterized atomistic TB model, when combined with effective many-body kernels, can reliably predict optical properties for realistic semiconducting nanostructures containing up to thousands of atoms. This approach offers a practical alternative to fully ab initio methods for studying linear and nonlinear optical phenomena in large-scale quantum-confined systems, bridging the gap between small-molecule accuracy and macroscopic scalability.