GW+BSE+DFPT: Unified First-Principles Framework

• The paper presents a unified framework combining GW for quasiparticle corrections, BSE for excitonic effects, and DFPT for vibrational and electron–phonon properties.
• The methodology employs a modular workflow with systematic convergence in DFT, DFPT, GW, and BSE stages to accurately determine bandgaps, dielectric response, and polaronic effects.
• The approach is crucial for predicting optoelectronic and transport properties in perovskites, chalcogenides, and low-dimensional nanomaterials, providing actionable insights into exciton binding and carrier mobilities.

The GW+BSE+DFPT approach is a unified, first-principles computational framework for determining electronic, optical, vibrational, and carrier-phonon coupling properties of materials. By combining the GW approximation for quasiparticle corrections, the Bethe–Salpeter equation (BSE) for neutral excitations, and density functional perturbation theory (DFPT) for lattice dynamics and electron–phonon (el-ph) coupling, this approach provides a quantitatively robust description of quasiparticle gaps, exciton binding energies, optical spectra, dielectric screening, and polaronic effects. It has become a standard workflow in computational condensed matter and materials science for quantitatively predicting optoelectronic and transport properties across a broad range of solid-state systems, including perovskites, chalcogenides, and low-dimensional nanomaterials (Adhikari et al., 2024, Chakravorty et al., 20 Feb 2025, Basera et al., 2020, Grillo et al., 10 Oct 2025).

1. Workflow Structure and Implementation

The GW+BSE+DFPT workflow proceeds through the following modular stages:

1. Ground-State DFT: Structure optimization and generation of the electronic ground state, typically using the PBE-GGA exchange-correlation functional and PAW or norm-conserving pseudopotentials. Parameters such as plane-wave cutoffs, k-mesh density, and total energy/force convergence thresholds are systematically tested (e.g., 400 eV and 4×4×4 k-mesh for perovskites (Adhikari et al., 2024); 8×8×6 k-mesh for chalcogenides (Chakravorty et al., 20 Feb 2025)).
2. DFPT Lattice Response: Calculation of the phonon spectrum, Born effective charges, and ionic contributions to the static dielectric constant via DFPT. Supercells or large unit cells are employed to capture LO phonon modes and accurate force constants, yielding longitudinal-optical (LO) phonon frequencies and metrics such as εion and ε∞ (Chakravorty et al., 20 Feb 2025, Adhikari et al., 2024).
3. Single-Shot GW (G₀W₀): Computation of quasiparticle energy corrections by evaluating the one-particle self-energy Σ(ω) on top of the DFT electronic structure, using the plasmon-pole model, large numbers of unoccupied states (e.g., NBANDS=540 or 640), and a consistent plane-wave/dielectric cutoff. The quasiparticle energy shift is applied via:

$$E^{QP}_{n,k} = E^{DFT}_{n,k} + \langle \psi_{n,k}| \Sigma(E^{QP}_{n,k}) - V_{xc} | \psi_{n,k} \rangle$$

(Adhikari et al., 2024, Chakravorty et al., 20 Feb 2025, Basera et al., 2020).

1. Bethe–Salpeter Equation (BSE): Construction and diagonalization of the two-particle (electron–hole) Hamiltonian using GW quasiparticle energies and statically screened electron–hole Coulomb interaction, typically in the Tamm-Dancoff approximation. Key outputs include the absorption spectrum, excitonic eigenvalues, and binding energies:

$E_{B} = E^{GW}_{gap} - E_{1}^{opt}$

(Adhikari et al., 2024).

2. Additional Analysis: Extraction of effective masses (via parabolic fits), evaluation of dielectric constants (from GW-RPA and DFPT), calculation of polaronic coupling parameters (Fröhlich α), and charge carrier mobilities using models such as Hellwarth’s polaron mobility formula.

An overview table for G₀W₀+BSE+DFPT settings in vacancy-ordered perovskites (Adhikari et al., 2024):

Stage	K-point mesh	Bands	Cutoff
DFT	4×4×4	–	400 eV
DFPT	2×2×2 supercell	–	400 eV
GW	4×4×4	540	400 eV
BSE	4×4×4	6+6	–

2. Key Formalism and Physical Quantities

• GW Approximation: The GW method addresses the limitations of DFT's Kohn-Sham eigenvalues by computing the self-energy Σ(ω) as

$$\Sigma(\mathbf{r}, \mathbf{r}'; \omega) = i \int \frac{d\omega'}{2\pi} G(\mathbf{r}, \mathbf{r}'; \omega+\omega') W(\mathbf{r}, \mathbf{r}'; \omega')$$

where $W = \epsilon^{-1} v$, with ε built in the RPA (Chakravorty et al., 20 Feb 2025).

• Bethe–Salpeter Equation (BSE): The neutral excitation spectrum is obtained by solving

$$\sum_{v'c'k'} [H_{vck,v'c'k'}] A^S_{v'c'k'} = \Omega_S A^S_{vck}$$

where $H = H^{diag} + K^{eh}$ includes quasiparticle transitions and the screened e–h interaction (Adhikari et al., 2024, Chakravorty et al., 20 Feb 2025).

• DFPT: Linear response theory is used to obtain phonon frequencies via dynamical matrix diagonalization, Born effective charges for LO-phonon coupling analysis, and the ionic part of the dielectric tensor:

$$\varepsilon_{ion,\alpha\beta} = \frac{4\pi}{\Omega} \sum_m \frac{Z_{m,\alpha} Z_{m,\beta}}{\omega_m^2}$$

(Chakravorty et al., 20 Feb 2025).

• Exciton Binding Energy: Both BSE and analytic Wannier–Mott models are used:

$$E_{B} = \frac{\mu}{\varepsilon_{\rm eff}^2} R_{\infty}$$

where μ is the reduced mass and ε_eff typically interpolates between high-frequency electronic and static dielectric constants, as resolved by DFPT (Basera et al., 2020).

3. Applications and Physical Insights

The GW+BSE+DFPT methodology yields direct access to key optoelectronic and transport parameters:

• Bandgaps: GW-corrected electronic gaps (e.g., 3.63–5.14 eV for Rb₂BCl₆ VODPs (Adhikari et al., 2024); 0.646–2.001 eV for AGeX₃ (Chakravorty et al., 20 Feb 2025)) ensure agreement with experimental PES/IPES.
• Excitonic Properties: Accurate exciton binding energies from BSE (e.g., 0.16–0.98 eV in VODPs; 0.03–73.63 meV in chalcogenides) and derived radii, elucidating the fundamental limit for photoexcited carrier dissociation (Adhikari et al., 2024, Chakravorty et al., 20 Feb 2025, Basera et al., 2020).
• Dielectric Screening: Determination of both electronic (ε_∞) and ionic (ε_ion) dielectric response enables separation of high-frequency and static screening, crucial for evaluating exciton dissociation and polaronic renormalizations (Chakravorty et al., 20 Feb 2025).
• Carrier Masses and Mobilities: Effective masses (e.g., m*_e ≈ 0.75–2.0 m₀; m*_h ≈ 2.97–3.17 m₀ for VODPs) and mobilities (electron μ_p ≈ 0.31–5.9 cm² V⁻¹ s⁻¹) computed using polaron models constrained by DFPT LO phonon data (Adhikari et al., 2024).
• Polaronic Effects: Fröhlich coupling constants (α = 4.02–10.05), polaron radii, and energies allow quantitative prediction of transport limitations due to strong electron–LO phonon coupling—especially prominent in polar semiconductors (Adhikari et al., 2024, Chakravorty et al., 20 Feb 2025).
• Emerging Material Classes: The approach has been successfully applied to 1D nanowire semiconductors, revealing extremely strong exciton binding (1–3 eV) and Wannier–Mott character even with large electron–hole separation, due to weak screening in reduced dimensions (Grillo et al., 10 Oct 2025).

4. Technical Best Practices and Methodological Challenges

Systematic convergence testing is required at each stage:

• Empty bands and k-points: GW and BSE require large numbers of empty states and fine k-meshes to converge polarization and kernel contributions, controlled by explicit convergence criteria (e.g., bandgap change <0.05 eV when increasing empty bands (Chakravorty et al., 20 Feb 2025)).
• Frequency Treatment: The plasmon-pole model is typically used for computational efficiency but must be validated against explicit frequency grids where possible (Chakravorty et al., 20 Feb 2025).
• Screening Consistency: Dielectric constants used for screening in GW, BSE, and polaron models must be cross-validated between GW-RPA, DFPT, and optical BSE calculations (Chakravorty et al., 20 Feb 2025, Basera et al., 2020).
• Kernel Construction: In BSE, the number of included bands in the kernel and the treatment of electron–hole direct/exchange terms (screened vs bare, static vs dynamic) significantly impact accuracy, especially for exciton binding calculations in low-dimensional or weakly screened systems (Adhikari et al., 2024, Grillo et al., 10 Oct 2025).
• Limitations: The current implementation of single-shot G₀W₀ may miss higher-order self-consistency effects; the BSE step is computationally intensive, limiting the number of bands and k-points in large unit cells. Dynamic screening and temperature effects are typically neglected in the static BSE+DFPT approach, so high-temperature exciton renormalization and non-adiabatic effects may be underestimated (Adhikari et al., 2024).

5. Extensions and Analytical Force Formalism

Recent methodological developments include analytic GW+BSE+DFPT force formalism for excited-state geometry optimization (Grande et al., 7 Feb 2025). Here, the total excited-state energy is

$E_{tot}(R) = E_0(R) + x\,\Omega^A(R)$

and the excited-state force incorporates GW and BSE corrections. The approach enables efficient, analytic computation of the force

$$F^{exc}_{I\alpha} = -\frac{\partial \Omega^A}{\partial R_{I\alpha}}$$

where $E_{B} = E^{GW}_{gap} - E_{1}^{opt}$0 is the lowest BSE solution at ionic coordinate R. DFPT-derived el–ph couplings are renormalized by GW-level bandstructure corrections, and analytic derivatives allow geometry relaxation on excited-state surfaces, including the study of self-trapped excitons (e.g., in LiF). This reduces the computational effort of obtaining excited-state phonon modes compared to O(3N) finite-difference approaches (Grande et al., 7 Feb 2025).

6. Computational Advances and Hybrid Functional Integration

Recent work using adaptively compressed exchange (ACE) accelerates GW+BSE calculations with hybrid functional starting points, dramatically reducing computational cost without degrading quasiparticle or excitation accuracy (Yu et al., 2024). By constructing a low-rank surrogate for the exact exchange operator and employing iterative DFPT/DMPT solvers that avoid explicit sums over virtual orbitals, GW+BSE can be scaled to larger systems with hybrid accuracy and cost approaching that of semi-local DFT. This is particularly impactful for systems where accurate dielectric screening and bandgap alignment require non-empirical hybrid functional tuning (Yu et al., 2024).

7. Advantages, Limitations, and Outlook

Advantages:

• Enables predictive, parameter-free calculation of bandgaps, optical absorption, excitonic spectra, and transport coefficients; provides insight into structure–property relationships in complex semiconductors and low-dimensional materials (Adhikari et al., 2024, Chakravorty et al., 20 Feb 2025, Grillo et al., 10 Oct 2025).
• Integrated workflow ensures consistency across electronic, vibrational, and many-body components.
• Provides analytic access to excited-state forces, polaronic parameters, and temperature-dependent behavior with controlled approximations (Grande et al., 7 Feb 2025).

Limitations:

• Computational cost remains significant for fine BSE sampling and full-frequency GW or large supercells.
• Static screening and Tamm–Dancoff approximations neglect dynamic correlations and temperature effects; a plausible implication is that hot-carrier and transient phenomena may be incompletely captured.
• For strongly correlated systems or those with strong non-adiabatic el–ph coupling, higher-level self-consistent GW or beyond-DFPT corrections may be required.

Continued algorithmic development (e.g., ACE, density-matrix perturbation) and integration with excited-state force formalism suggest further expansion of GW+BSE+DFPT’s reach into excited-state dynamics, nano-structure design, and complex heterostructure modeling (Yu et al., 2024, Grande et al., 7 Feb 2025).