betaFeedback

WorksPhysics and Astronomy

Phonon-Assisted Radiative Lifetimes and Exciton Dynamics from First Principles

Abstract

Exciton-phonon interactions play a fundamental role in phonon-assisted radiative recombination and exciton dynamics in solids. In this work, we present a first-principles framework for computing phonon-assisted radiative lifetimes and exciton dynamics at finite temperatures. Starting from the solution of the Bethe-Salpeter equation, we construct an effective excitonic Hamiltonian that incorporates both exciton-photon and exciton-phonon interactions. Phonon-assisted radiative lifetimes in anisotropic media are evaluated using time-dependent second-order perturbation theory. We further analyze the temperature and phonon-mode dependence of phonon-assisted radiative lifetime and compare our results with available experimental data. We explain the nonmonotonic temperature dependence of the phonon-assisted radiative lifetime by different mechanisms at low and high-temperature regimes. Finally, we perform real-time exciton relaxation at the diagonal approximation of Lindbladian dynamics for time-resolved exciton occupation, providing insights into ultrafast thermalization and scattering pathways. Our ab-initio theory offers a detailed microscopic understanding of phonon-mediated exciton relaxation and recombination processes, and provides in-depth perspectives on phonon-assisted many-body interactions and their influence on optical properties for light-emitting and optoelectronic applications.

Exciton-phonon interactions play a fundamental role in the optical and electronic transport properties of semiconductors and insulators. In particular, they are essential for describing phonon-assisted radiative recombination processes and exciton dynamics in materials with indirect band gaps, where momentum conservation between the electron and hole requires the assistance of phonons during photon emission. Such phonon-mediated optical transitions are especially relevant in layered van der Waals (vdW) materials and other quantum materials, where the interplay between the many-body effects of electrons and lattice vibrations leads to rich unconventional phenomena [1, 2, 3]. In past decades, the study of excitons, which is the quasiparticle of bound electron-hole pairs, has advanced significantly due to the development of first-principles many-body perturbation theory by solving the Bethe-Salpeter equation (BSE) [4, 5, 6, 7]. It has reliably predicted exciton binding energies, exciton dispersion at finite momentum [8], exciton-phonon scattering rates [9, 10], and optical spectra in bulk and low-dimensional materials [4, 11]. Most of these methods have been applied in the context of first-order or direct optical transitions [12, 13, 14], and do not account for second-order phonon-assisted recombination processes, which require both light-matter interaction and exciton-phonon couplings.

Phonon-assisted luminescence, such as indirect photoluminescence (PL) observed in hexagonal boron nitride (hBN) [1, 2], monolayer WSe2_2 [3], or silicon [15], in principles cannot be captured by the first-order optical processes alone. Instead, second-order processes involving both exciton-photon and exciton-phonon interactions must be considered. Despite their importance, a fully ab initio framework for phonon-assisted radiative lifetimes from many-body perturbation theory remains to be developed. Previous studies have developed second-order phonon-assisted optical transitions at the single particle level [16, 17, 18] or have treated exciton-phonon and exciton-photon interactions separately [3, 10, 19], potentially losing coupling effects amid multiple scattering channels [20]. In this work, we address this gap by developing a rigorous many-body formalism to compute phonon-assisted radiative recombination and time-resolved exciton dynamics from first principles. Our methodology is based on the construction of an effective exciton-phonon-phonon Hamiltonian in second quantization, incorporating the excitonic states from the solution of the finite-momentum BSE [4] and the exciton-phonon coupling derived from electron-phonon matrix elements [10, 21, 22, 23]. We evaluate the phonon-assisted radiative rates using time-dependent second-order perturbation theory [24], explicitly accounting for phonon emission and absorption channels. Furthermore, to capture the out-of-equilibrium dynamics of excitons in open quantum system, we derive a quantum master equation in the Lindblad form [25, 26, 27, 28] that describes the evolution of the exciton density matrix under phonon scatterings. In the semi-classical limit, this reduces to the Boltzmann transport equation (BTE) [29, 30, 31], enabling real-time simulation of exciton relaxation and thermalization.

We first derive the effective exciton-photon-phonon Hamiltonian starting from the one-particle light-matter interaction (electron-photon) Hamiltonian and the electron-phonon Hamiltonian. The exciton, as the eigenstate of the BSE Hamiltonian, is a quasiparticle defined in the electron-hole space [32]. The exciton creation operator can be expressed in the form c^QS=∑vckAvckS(Q)d^ck+Q†d^vk\hat{c}^{S}_{\mathbf{Q}} = \sum_{vck} A^{S(\mathbf{Q})}_{vck}\hat{d}^{\dagger}_{c\mathbf{k}+\mathbf{Q}}\hat{d}_{v\mathbf{k}}, where c^S(Q)\hat{c}^{S(\mathbf{Q})} creates an exciton in state SS with momentum Q\mathbf{Q}, d^ck+Q†\hat{d}^{\dagger}_{c\mathbf{k}+\mathbf{Q}} and d^vk\hat{d}_{v\mathbf{k}} are electron creation and annihilation operators, respectively. AvckS(Q)A^{S(\mathbf{Q})}_{vck} is the exciton amplitude. This formalism implies that a single exciton corresponds to an interband transition. In contrast, describing an intraband transition requires the involvement of two excitons. We start with the non-interacting Hamiltonian, which includes the exciton, photon, and phonon fields:

H^0=H^ex+H^ph+H^γ=∑S,QℏωS(Q)c^S(Q)†c^S(Q)+∑νqℏωνq(b^νq†b^νq+12)+∑λkˉℏωλkˉ(a^λkˉ†a^λkˉ+12).(1)\begin{aligned} \hat{H}_{0}=\hat{H}_{\mathrm{ex}}+\hat{H}_{\mathrm{ph}}+\hat{H}_{\gamma} \\ &=\sum_{S,\mathbf{Q}}\hbar\omega_{S(\mathbf{Q})}\hat{c}^{\dagger}_{S(\mathbf{Q})}\hat{c}_{S(\mathbf{Q})}+\sum_{\nu\mathbf{q}}\hbar\omega_{\nu\mathbf{q}}\left(\hat{b}^{\dagger}_{\nu\mathbf{q}}\hat{b}_{\nu\mathbf{q}}+\frac{1}{2}\right) \\ &+\sum_{\lambda\bar{\mathbf{k}}}\hbar\omega_{\lambda\bar{\mathbf{k}}}\left(\hat{a}^{\dagger}_{\lambda\bar{\mathbf{k}}}\hat{a}_{\lambda\bar{\mathbf{k}}}+\frac{1}{2}\right). \tag*{(1)} \end{aligned}

where ν\nu, and λ\lambda are indices representing phonon, and photon modes, respectively, and q\mathbf{q}, and kˉ\bar{\mathbf{k}} denote their corresponding wavevectors. The field operators b^†(b^)\hat{b}^{\dagger}(\hat{b}), a^†(a^)\hat{a}^{\dagger}(\hat{a}) represent the annihilation (creation) operators for phonons, and photons, respectively. Using the plane wave representation of the vector potential A\mathbf{A}, extending the electron-photon Hamiltonian H^e−γ=−emA^⋅p^\hat{H}^{e-\gamma}=-\frac{e}{m}\hat{\mathbf{A}}\cdot\hat{\mathbf{p}} to the excitonic basis H^ex−γ=−emexA^⋅p^ex\hat{H}^{\mathrm{ex}-\gamma}=-\frac{e}{m_{\mathrm{ex}}}\hat{\mathbf{A}}\cdot\hat{\mathbf{p}}_{\mathrm{ex}}, the exciton-photon Hamiltonian is written as:

where [r^,H^ex]\left[\hat{\mathbf{r}},\hat{H}_{\mathrm{ex}}\right] represents the commutator between the position operator r^\hat{\mathbf{r}} and the excitonic Hamiltonian H^ex\hat{H}_{\mathrm{ex}}. Considering the electron-phonon Hamiltonian at first order in atomic displacement [21], it can be expressed as:

H^e−ph=∑k,q∑mnνgmnν(k,q)d^m,k+q†d^n,k(b^qν+b^−qν†),\hat{H}^{e-\mathrm{ph}}=\sum_{\mathbf{k},\mathbf{q}}\sum_{mn\nu}g_{mn\nu}(\mathbf{k},\mathbf{q})\hat{d}^{\dagger}_{m,\mathbf{k}+\mathbf{q}}\hat{d}_{n,\mathbf{k}}\left(\hat{b}_{\mathbf{q}\nu}+\hat{b}^{\dagger}_{-\mathbf{q}\nu}\right),

where d^\hat{d} is the electron field operator, and mm, nn represent the band indices. The term gmnν(k,q)=⟨mk+q∣ΔqνKS∣nk⟩g_{mn\nu}(\mathbf{k},\mathbf{q})=\langle m\mathbf{k}+\mathbf{q}|\Delta_{\mathbf{q}\nu}^{\mathrm{KS}}|n\mathbf{k}\rangle denotes the electron-phonon coupling matrix in the mode representation, with ΔqνKS\Delta_{\mathbf{q}\nu}^{\mathrm{KS}} being the derivative of the Kohn-Sham potential from DFPT calculations. The Born–von Karman (BvK) boundary condition is applied to plane waves [21]. Consequently, the summation over wavevectors is restricted to the first Brillouin zone, and the normalization factor Np−1/2N_{p}^{-1/2} is omitted for simplicity. Therefore, the total Hamiltonian H^\hat{H} is defined as H^0+H^int\hat{H}_{0}+\hat{H}_{\mathrm{int}}, where H^int\hat{H}_{\mathrm{int}} represents the interacting Hamiltonian, given by: H^int=H^ex−γ+H^intraex−ph+H^interex−ph\hat{H}_{\mathrm{int}}=\hat{H}^{\mathrm{ex}-\gamma}+\hat{H}^{\mathrm{ex}-\mathrm{ph}}_{\mathrm{intra}}+\hat{H}^{\mathrm{ex}-\mathrm{ph}}_{\mathrm{inter}},

H^intraex−ph=∑SS′∑qνGSS′ν(Q,q)c^S′(Q+q)†c^S(Q)(b^qν+b^−qν†);\hat{H}^{\mathrm{ex}-\mathrm{ph}}_{\mathrm{intra}}=\sum_{SS'}\sum_{\mathbf{q}\nu}\mathcal{G}_{SS'\nu}(\mathbf{Q},\mathbf{q})\hat{c}^{\dagger}_{S'(\mathbf{Q}+\mathbf{q})}\hat{c}_{S(\mathbf{Q})}\left(\hat{b}_{\mathbf{q}\nu}+\hat{b}^{\dagger}_{-\mathbf{q}\nu}\right);
H^interex−ph=∑SQνGSν(Q)c^S(Q)(b^−Qν+b^Qν†),(2)\hat{H}^{\mathrm{ex}-\mathrm{ph}}_{\mathrm{inter}}=\sum_{S\mathbf{Q}\nu}\mathcal{G}_{S\nu}(\mathbf{Q})\hat{c}_{S(\mathbf{Q})}\left(\hat{b}_{-\mathbf{Q}\nu}+\hat{b}^{\dagger}_{\mathbf{Q}\nu}\right), \tag*{(2)}

the intraband and interband exciton-phonon interaction Hamiltonians are derived by projecting the electron-phonon coupling from the one-particle basis onto the excitonic basis, where G\mathcal{G} is the interband and intraband exciton-phonon matrix element:

GS′Sν(Q,q)=∑vv′cc′kAv′c′kS′(Q+q)∗AvckS(Q)gc′cν(k+Q,q)\mathcal{G}_{S'S\nu}(\mathbf{Q},\mathbf{q})=\sum_{vv'cc'\mathbf{k}} A^{S'(\mathbf{Q}+\mathbf{q})*}_{v'c'\mathbf{k}}A^{S(\mathbf{Q})}_{vc\mathbf{k}}g_{c'c\nu}(\mathbf{k}+\mathbf{Q},\mathbf{q})
−∑cvv′kAv′ckS′(Q+q)∗Avck+qS(Q)gv′vν(k,q);-\sum_{cvv'\mathbf{k}} A^{S'(\mathbf{Q}+\mathbf{q})*}_{v'c\mathbf{k}}A^{S(\mathbf{Q})}_{vc\mathbf{k}+\mathbf{q}}g_{v'v\nu}(\mathbf{k},\mathbf{q});
GSν(Q)=∑vckAvckS(Q)gvcν(k+Q,−Q).(3)\mathcal{G}_{S\nu}(\mathbf{Q})=\sum_{vc\mathbf{k}}A^{S(\mathbf{Q})}_{vc\mathbf{k}}g_{vc\nu}(\mathbf{k}+\mathbf{Q},-\mathbf{Q}). \tag*{(3)}

The phonon-assisted radiative transition probability is expressed in terms of the time-dependent perturbative coefficients [24] ci→f(n)(t)c^{(n)}_{i\to f}(t) as: Pi→f(t)=∣ci→f(1)(t)+ci→f(2)(t)+⋯ ∣2P_{i\to f}(t)=\left|c^{(1)}_{i\to f}(t)+c^{(2)}_{i\to f}(t)+\cdots\right|^{2}, for the specific i→fi\to f transition under consideration, the first-order coefficient ci→f(1)(t)c_{i\to f}^{(1)}(t) vanishes. Contributions from higher-order terms beyond the second order are not included in this analysis. The transition rate includes contributions from both interband and intraband processes:

γS(Q)=2πℏ4∑λ,kˉ,q,ν,±(nν,∓q+12±12)δ(ωλ,kˉ±ων,∓q−ωS(Q))lim⁡η→0+∣∑S′[GS′ν(−q)MS′(−q)S(Q),λkˉωS′(−q)+ωλ,kˉ−ωS(Q)−iη+GS′Sν(Q,q)MS′(Q+q),λkˉωS′(Q+q)±ων,∓q−ωS(Q)−iη]δQ+q,−kˉ∣2,(4)\gamma_S(\mathbf{Q}) = \frac{2\pi}{\hbar^4} \sum_{\lambda,\bar{k},q,\nu,\pm} \left(n_{\nu,\mp q}+\frac{1}{2}\pm\frac{1}{2}\right)\delta(\omega_{\lambda,\bar{k}}\pm\omega_{\nu,\mp q}-\omega_S(\mathbf{Q})) \lim_{\eta\to0^+}\left|\sum_{S'}\left[\frac{\mathcal{G}_{S'\nu}(-q)\mathcal{M}_{S'(-q)S(\mathbf{Q}),\lambda\bar{k}}}{\omega_{S'}(-q)+\omega_{\lambda,\bar{k}}-\omega_S(\mathbf{Q})-i\eta}+\frac{\mathcal{G}_{S'S\nu}(\mathbf{Q},q)\mathcal{M}_{S'(\mathbf{Q}+q),\lambda\bar{k}}}{\omega_{S'}(\mathbf{Q}+q)\pm\omega_{\nu,\mp q}-\omega_S(\mathbf{Q})-i\eta}\right]\delta_{\mathbf{Q}+q,-\bar{k}}\right|^2, \tag*{(4)}

where the detailed derivation of second-order time-dependent perturbation theory is provided in the SI Sec. S1-S3. The expression of interband and intraband exciton-photon matrix elements M\mathcal{M}, within the dipole approximation, is provided as:

MS′(Q′)S(Q),λkˉ=iℏe22ℏωλkˉϵ0V(ωS(Q)−ωS′(Q′))eλkˉ∗⋅⟨S′(Q′)∣r^∣S(Q)⟩;\mathcal{M}_{S'(\mathbf{Q}')S(\mathbf{Q}),\lambda\bar{k}}=i\hbar\sqrt{\frac{e^2}{2\hbar\omega_{\lambda\bar{k}}\epsilon_0 V}}(\omega_S(\mathbf{Q})-\omega_{S'}(\mathbf{Q}'))\mathbf{e}_{\lambda\bar{k}}^*\cdot\langle S'(\mathbf{Q}')|\hat{\mathbf{r}}|S(\mathbf{Q})\rangle;
MS(Q),λk=iℏe22ℏωλkϵ0VωS(Q)eλk∗⋅⟨G∣r^∣S(Q)⟩.(5)\mathcal{M}_{S(\mathbf{Q}),\lambda k}=i\hbar\sqrt{\frac{e^2}{2\hbar\omega_{\lambda k}\epsilon_0 V}}\omega_S(\mathbf{Q})\mathbf{e}_{\lambda k}^*\cdot\langle G|\hat{\mathbf{r}}|S(\mathbf{Q})\rangle. \tag*{(5)}

In the following discussion, we focus on the phonon-assisted radiative process, specifically, intraband exciton-phonon coupling, which is described by the second term in Eq. (4) and depicted in Figure 1(a). The interband exciton-photon coupling, which is described by the first term in Eq. (4) and depicted in Figure 1(b), corresponding to transitions between the valence and conduction bands induced by phonon excitation, is typically negligible in semiconductors, because of the large energy gap between these bands compared to phonon energies.

Schematic illustration of phonon-assisted radiative processes

Figure 1. Schematic illustration of phonon-assisted radiative processes. (a) Intraband phonon scattering (ph) combined with interband photon emission (hνh\nu), and (b) interband phonon scattering combined with intraband photon emission.

Considering the propagation of light in uniform nonmagnetic anisotropic media [34, 35], the relative dielectric tensor ϵ^\hat{\epsilon} can be diagonalized along the principal optical axes. In our later discussion, we focus on a special case when ϵxx=ϵyy=ϵ∥\epsilon_{xx}=\epsilon_{yy}=\epsilon_{\parallel} and ϵzz=ϵ⊥\epsilon_{zz}=\epsilon_{\perp}, which is sufficient to describe out-of-plane anisotropic medium. We convert the sum of the photon wavevector kˉ\bar{k} into an integral V(2π)3∫d3kˉ\frac{V}{(2\pi)^3}\int d^3\bar{k} in Eq. (4), then use the dispersion relation ωλkˉ=c∣kˉ∣/nλ\omega_{\lambda\bar{k}}=c|\bar{k}|/n_\lambda, where nλn_\lambda is the refractive index and λ=±1\lambda=\pm1 represents the photon mode. We finally obtain the state-resolved phonon-assisted radiative rate, γS(Q)\gamma_S(\mathbf{Q}) as follows (the details can be found in the SI Sec. S4):

γS(Q)=43α∑νq±(ωS(Q)∓ων,∓q)∣ϵ^1/2ΛSν±(Q,q)∣2nν,q±,(6)\gamma_S(\mathbf{Q})=\frac{4}{3}\alpha\sum_{\nu q\pm}\left(\omega_S(\mathbf{Q})\mp\omega_{\nu,\mp q}\right)\left|\hat{\epsilon}^{1/2}\boldsymbol{\Lambda}_{S\nu}^{\pm}(\mathbf{Q},q)\right|^2n_{\nu,q}^{\pm}, \tag*{(6)}

where the dimensionless fine-structure constant α\alpha is expressed as α=e24πϵ0ℏc\alpha=\frac{e^2}{4\pi\epsilon_0\hbar c}, the dimensionless effective amplitude Λ\boldsymbol{\Lambda} in the long wave-length limit is expressed as:

ΛSν±(Q,q)=1ℏc∑S′ωS′,0μS′′(0)GS′Sν±(Q,q)ωS′,0−ωS,Q±ων,∓q−iηδQ,−q,\boldsymbol{\Lambda}_{S\nu}^{\pm}(\mathbf{Q},q)=\frac{1}{\hbar c}\sum_{S'}\omega_{S',0}\frac{\boldsymbol{\mu}_{S'}^{\prime}(0)\mathcal{G}_{S'S\nu}^{\pm}(\mathbf{Q},q)}{\omega_{S',0}-\omega_{S,\mathbf{Q}}\pm\omega_{\nu,\mp q}-i\eta}\delta_{\mathbf{Q},-q},

the μ\boldsymbol{\mu} represents the exciton transition dipole moment:

μS′(Q′)=⟨G∣r^∣S′(Q′)⟩.\boldsymbol{\mu}_{S'}(\mathbf{Q}')=\langle G|\hat{\mathbf{r}}|S'(\mathbf{Q}')\rangle.

The phonon occupation factor for phonon emission (+)/absorption (-) process is nν,q±=nν,∓q+12±12n_{\nu,q}^{\pm}=n_{\nu,\mp q}+\frac{1}{2}\pm\frac{1}{2}. The dimensionless effective tensor ϵ^\hat{\boldsymbol{\epsilon}} is expressed as:

ϵ^=(ϵ∥−1/2ϵ⊥+3ϵ∥1/24000ϵ∥−1/2ϵ⊥+3ϵ∥1/24000ϵ∥1/2).(7)\hat{\boldsymbol{\epsilon}}=\begin{pmatrix} \frac{\epsilon_{\parallel}^{-1/2}\epsilon_{\perp}+3\epsilon_{\parallel}^{1/2}}{4} & 0 & 0\\ 0 & \frac{\epsilon_{\parallel}^{-1/2}\epsilon_{\perp}+3\epsilon_{\parallel}^{1/2}}{4} & 0\\ 0 & 0 & \epsilon_{\parallel}^{1/2} \end{pmatrix}. \tag*{(7)}

The finite-temperature phonon-assisted radiative rate γ\gamma is calculated as a thermal average of γS(Q)\gamma_S(\mathbf{Q}) in the exciton dilute limit: γ=⟨γS(Q)⟩=1Z∑SγSe−βℏωS(Q)\gamma=\langle\gamma_S(\mathbf{Q})\rangle=\frac{1}{Z}\sum_S\gamma_S e^{-\beta\hbar\omega_S(\mathbf{Q})}, where the partition function is given by Z=∑Se−βℏωS(Q)Z=\sum_S e^{-\beta\hbar\omega_S(\mathbf{Q})} [36]. The finite-temperature phonon-assisted radiative lifetime τ\tau is defined as the inverse of the phonon-assisted radiative rate.

To study the exciton dynamics and thermalization, we derived the quantum master equation in the Lindblad form [28] for exciton-phonon dynamics (the details can be found in the SI Sec. S5):

dρ12dt=12∑345((1+ρ)14P4253ρ53−(1+ρ)34P3415∗ρ52)+h.c.,(8)\frac{d\rho_{12}}{dt}=\frac{1}{2}\sum_{345}\left((1+\rho)_{14}\mathcal{P}_{4253}\rho_{53}-(1+\rho)_{34}\mathcal{P}_{3415}^*\rho_{52}\right)+\mathrm{h.c.}, \tag*{(8)}

where the scattering terms are defined as: P1234=\mathcal{P}_{1234}=

∑qν±A13qν±A24qν±,∗withA13qν±=D13qν±nν,∓q+12±12.\sum_{\mathbf{q}\nu\pm} A_{13}^{q\nu\pm} A_{24}^{q\nu\pm,*}\quad\text{with}\quad A_{13}^{q\nu\pm}=\mathcal{D}_{13}^{q\nu\pm}\sqrt{n_{\nu,\mp\mathbf{q}}+\frac{1}{2}\pm\frac{1}{2}}.

The exciton-phonon coupling term D12qν±\mathcal{D}_{12}^{q\nu\pm} is expressed as:

D12qν±=2πℏG12qνδG1/2(ω1−ω2±ω∓qν),\mathcal{D}_{12}^{q\nu\pm}=\sqrt{\frac{2\pi}{\hbar}}\mathcal{G}_{12}^{q\nu}\delta_{G}^{1/2}(\omega_{1}-\omega_{2}\pm\omega_{\mp\mathbf{q}\nu}),

where δG\delta_G represents an energy-conserving Dirac delta function broadened to a Gaussian. Here, the exciton-phonon matrix element GS′Sν(Q,q)\mathcal{G}_{S'S}^{\nu}(\mathbf{Q},\mathbf{q}) is denoted for simplicity as G12qν\mathcal{G}_{12}^{q\nu}, where the indices are defined as 1:=S′(Q+q)1:=S'(\mathbf{Q}+\mathbf{q}) and 2:=S(Q)2:=S(\mathbf{Q}). ρ\rho represents the exciton density matrix. Notably, the primary distinction between the Lindblad equation for the exciton dynamics and that for the electron dynamics[27, 28, 37] lies in the term 1−ρ1-\rho being replaced by 1+ρ1+\rho. This difference arises due to the distinct (anti)commutation relations governing fermionic and bosonic operators. The related detailed derivation can be found in the SI Sec. S5. In this study, we focus on exciton occupation changes due to exciton-phonon scatterings, where the phase coherence information between different exciton states (corresponding to the off-diagonal elements of exciton density matrices) is not important, which will leave for our future studies. Therefore, by applying the diagonal (semiclassical) approximation to exciton density matrices and Lindbladian operators, we obtain the (semiclassical) quantum master equation:

ddtN2=∑1Γ21N1(1+N2)−∑1Γ12N2(1+N1),(9)\frac{d}{dt}\mathcal{N}_{2}=\sum_{1}\Gamma_{21}\mathcal{N}_{1}(1+\mathcal{N}_{2})-\sum_{1}\Gamma_{12}\mathcal{N}_{2}(1+\mathcal{N}_{1}), \tag*{(9)}

where the scattering rate Γ12\Gamma_{12} describes the transition rate between exciton states 1, 2. This rate is expressed as:

Γ12=P1122=∑qν±A12qν±A12qν±,∗=2πℏ2∑qν±∣G12qν∣2(nν,∓q+12±12)δG(ω1−ω2±ω∓qν),(10)\begin{aligned} \Gamma_{12}=\mathcal{P}_{1122}=\sum_{\mathbf{q}\nu\pm} \mathcal{A}_{12}^{q\nu\pm}\mathcal{A}_{12}^{q\nu\pm,*} \\ &=\frac{2\pi}{\hbar^{2}}\sum_{\mathbf{q}\nu\pm}\left|\mathcal{G}_{12}^{q\nu}\right|^{2}\left(n_{\nu,\mp\mathbf{q}}+\frac{1}{2}\pm\frac{1}{2}\right)\delta_{G}(\omega_{1}-\omega_{2}\pm\omega_{\mp\mathbf{q}\nu}), \tag*{(10)} \end{aligned}

which aligns with the transition rate derived from the Fermi’s golden rule.

Next, we present the results of temperature dependent phonon-assisted radiative lifetime in Figure 2(a). The black curve shows the total radiative lifetime, which includes both phonon emission (the red curve) and phonon absorption contributions (the blue curve). At temperature below 150K, the total radiative lifetime exhibits a nonmonotonic behavior, which can be understood qualitatively by a two-valley model. We approximate the exciton energy surface as a two-level model consisting of valleys 1 and 2, where valley 1 dominates the radiative process. Neglecting exciton dispersion, we assume the radiative rate for valley 1 is given by γ1(T)=γ~1(nph+1)\gamma_{1}(T)=\tilde{\gamma}_{1}(n_{\mathrm{ph}}+1), where nph=(eβℏω~−1)−1n_{\mathrm{ph}}=(e^{\beta\hbar\tilde{\omega}}-1)^{-1} is the Bose-Einstein phonon occupation factor for a single effective phonon mode of frequency ω~\tilde{\omega}. At low temperatures, the phonon emission process dominates. The total low temperature thermal-

Temperature-dependent phonon-assisted radiative lifetime and phonon contributions

Figure 2. Temperature dependence of phonon-assisted radiative lifetime in bulk hBN. (a) Total radiative lifetime (black), with contributions from phonon emission (red) and phonon absorption (blue) shown separately. The magenta star marks the experimental radiative lifetime of 27 ns at room temperature, reported by S. Roux et al. [2]. The gray solid line represents a fit using a two-valley model, capturing the non-monotonic behavior at low temperatures. The gray dashed line corresponds to Eq. 7 using constant exciton-phonon coupling G\mathcal{G}. (b) Labeling of phonon branches in the phonon band structure of bulk hBN. (c) Phonon mode contributions to the radiative lifetime from phonon emission processes. (d) Phonon mode contributions to the radiative lifetime from phonon absorption processes.

averaged radiative rate is then given by:

γ(T)∼γ~1(11+e−βΔE)(1eβℏω~−1+1),(11)\gamma(T) \sim\widetilde{\gamma}_1 \left(\frac{1}{1+e^{-\beta\Delta E}}\right) \left(\frac{1}{e^{\beta\hbar\widetilde{\omega}}-1}+1\right), \tag*{(11)}

where the first bracket term accounts for the thermal occupation of the dominate valley and decreases with increasing temperature, while the second bracket term captures the temperature dependence of phonon emission occupation factor and increases with temperature. Their competition can lead to the observed nonmonotonic behavior in the total and phonon emission radiative lifetimes at low temperature. We fitted this two-valley model to phonon-assisted radiative rate γ(T)\gamma(T) from first-principles calculations. A detailed discussion of this two-valley model and effective parameters fit in Eq. (11) is provided in the SI Sec. S6. The resulting curve, shown as the gray solid line in Figure 2(a), agrees with the first-principles results for the total and phonon-emission radiative lifetimes in the low-temperature regime with the non-monotonic behavior. At temperature above 150K, the radiative lifetime exhibits a monotonic decrease with increasing temperature. This behavior is mainly driven by the sharp increase in the phonon population as the thermal energy kBTk_B T becomes comparable to the phonon energy. To analyze this effect, we set the exciton-phonon matrix elements in Eq. (6) to a constant value that matches the calculated radiative lifetime at room temperature. The resulting curve, shown as the gray dashed line in Figure 2(a), reproduces the decreasing trend observed in the full first-principles calculation. This comparison indicates that, unlike the low-temperature regime, where differences in valley-dependent scattering rates dominate, the high-temperature behavior is governed by the thermal-induced increase of the phonon population and is less dependent on the detailed exciton-phonon coupling structure.

We analyze the phonon-mode contributions to the phonon-assisted radiative lifetime, separating the effects of phonon emission and absorption, as shown in Figure 2(b-Figure 2d). For the phonon emission process in Figure 2(c), high-energy optical modes (green) dominate at low temperatures due to their strong exciton-phonon coupling strength. The phonon-mode resolved exciton-phonon matrix elements are provided in the SI Sec. S7. Their contributions are relatively less sensitive to temperature due to much higher phonon energies compared to thermal energy kBTk_B T. As the temperature increases, low-energy optical and acoustic modes (red and blue) become increasingly important, as their phonon occupations rise rapidly when kBTk_B T becomes comparable to their phonon energies. In contrast, the phonon absorption process in Figure 2(d) shows negligible contributions from all phonon modes at low temperatures due to the lack of thermal population of phonons. As the temperature increases, low-energy optical and acoustic modes become dominant for phonon absorption processes, and correspondingly the radiative rate, as their phonon occupations rise significantly.

We finally study exciton dynamics based on Eq. (9), the computational details can be found in SI Sec. S6-S7. Our analysis of exciton occupation at different time steps in Figure 3(a-Figure 3d) reveals the exciton redistribution (thermalization) toward finite-momentum energy valleys close to Q. Initially, the exciton population is prepared at the first bright optical exciton state as shown in Figure 3(a). We set the exciton occupation number to 1, consistent with the exciton dilute limit. As the system evolves, exciton-phonon scattering leads to relaxation and redistribution towards lower-energy states. Our simulations indicate that the relaxation process towards the energy minimum follows an exponential decay law, see SI Sec. S6, with a fitted valley relaxation time of approximately 114 fs at room temperature. This timescale is consistent with experimental PL linewidth measurements at room temperature [1], which report a linewidth Δ\Delta of around 30 meV (corresponding to an effective relaxation time τ≈60\tau\approx60 fs, based on the optical theorem relation τ−1=2ℏΔ\tau^{-1} = \frac{2}{\hbar}\Delta). Note that our exciton Lindbladian dynamics formalism is well suited for extending beyond the semiclassical Boltzmann equation by incorporating exciton density matrix and full Lindbladian operators. This framework provides a direct pathway for studying exciton dephasing and exciton spin dynamics, capturing quantum coherence effects [38, 39] that are absent in the semiclassical limits.

Real-time exciton occupation of bulk hBN

Figure 3. Real-time exciton occupation of bulk hBN at room temperature simulated using the real-time semi-classical quantum master equation Eq. (10) at different time steps: (a) 0 fs (initial step), (b) 750 fs, (c) 1500 fs, and (d) 2000 fs (final step). Exciton occupation is projected onto the exciton band structure and plotted as a colormap along Γ→Q→K\Gamma\to Q \to K.

In summary, we developed a comprehensive ab initio framework for evaluating phonon-assisted radiative processes and exciton dynamics in semiconductors, based on time-dependent second-order perturbation theory and many-body perturbation theory, starting with an effective exciton-photon-phonon Hamiltonian. By applying this formalism to a prototypical indirect materials bulk hBN, we computed the phonon-assisted radiative lifetime in a wide range of temperatures and found good agreement with the available experimental data. The resulting temperature dependence of the radiative lifetime is non-monotonic at low temperatures and monotonically decreases at higher temperatures above 150 K. This trend contrasts with the conventional T3/2T^{3/2} increase observed for direct exciton radiative lifetime [40, 41, 42]. We explain the non-monotonic behavior at low temperatures with a simplified two-valley model, where a large difference in scattering rates between different valleys plays a critical role. At higher temperatures, the decrease in radiative lifetime is shown to be governed by the thermal occupation factors. To capture the out-of-equilibrium dynamics of excitons, we derived and implemented a real-time quantum master equation in the Lindblad formalism. In the semiclassical limit, this reduces to the Boltzmann transport equation (BTE) with exciton-phonon scattering dynamics. Our simulations reveal ultrafast relaxation of excitons into finite-momentum valleys with information of relaxation and thermalization dynamics. This framework lays the groundwork for future extensions to study exciton coherence, dephasing, and exciton spin dynamics in realistic materials.

acknowledgement

We thank very helpful discussions with Feliciano Giustino, Davide Sangalli, and Claudio Attaccalite. We acknowledge support from the Air Force Office of Scientific Research under AFOSR Award No. FA9550-21-1-0087. This research used resources of the Scientific Data and Computing center, a component of the Computational Science Initiative, at Brookhaven National Laboratory under Contract No. DE-SC0012704, the National Energy Research Scientific Computing Center (NERSC) a U.S. Department of Energy Office of Science User Facility operated under Contract No. DE-AC02-05CH11231. This work used the TACC Stampede3 system at the University of Texas at Austin through allocation PHY240212 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services and Support (AC-CESS) program [43], which is supported by US National Science Foundation grants No. 2138259, No. 2138286, No. 2138307, No. 2137603, and No. 2138296.

References

References

  1. [1]T. Q. P. Vuong, G. Cassabois, P. Valvin, S. Liu, J. H. Edgar, and B. Gil, Phys. Rev. B 95, 201202 (2017).
  2. [2]S. Roux, C. Arnold, F. Paleari, L. Sponza, E. Janzen, J. H. Edgar, B. Toury, C. Journet, V. Garnier, P. Steyer, T. Taniguchi, K. Watanabe, F. m. c. Ducastelle, A. Loiseau, and J. Barjon, Phys. Rev. B 104, L161203 (2021).
  3. [3]S. Brem, A. Ekman, D. Christiansen, F. Katsch, M. Selig, C. Robert, X. Marie, B. Urbaszek, A. Knorr, and E. Malic, Nano letters 20, 2849 (2020).
  4. [4]D. Sangalli, A. Ferretti, H. Miranda, C. Attaccalite, I. Marri, E. Cannuccia, P. Melo, M. Marsili, F. Paleari, A. Marrazzo, et al., Journal of Physics: Condensed Matter 31, 325902 (2019).
  5. [5]M. Govoni and G. Galli, Journal of Chemical Theory and Computation 11, 2680 (2015).
  6. [6]J. Deslippe, G. Samsonidze, D. A. Strubbe, M. Jain, M. L. Cohen, and S. G. Louie, Computer Physics Communications 183, 1269 (2012).
  7. [7]Y. Ping, D. Rocca, and G. Galli, Chemical Society Reviews 42, 2437 (2013).
  8. [8]D. Y. Qiu, T. Q. Cao, and S. G. Louie, Phys. Rev. Lett. 115, 176801 (2015).
  9. [9]G. Antonius and S. G. Louie, Phys. Rev. B 105, 085111 (2022).
  10. [10]H.-Y. Chen, D. Sangalli, and M. Bernardi, Phys. Rev. Lett. 125, 107401 (2020).
  11. [11]D. Y. Qiu, F. H. da Jornada, and S. G. Louie, Phys. Rev. Lett. 111, 216805 (2013).
  12. [12]F. Wu, T. J. Smart, J. Xu, and Y. Ping, Physical Review B 100, 081407 (2019).
  13. [13]M. Palummo, M. Bernardi, and J. C. Grossman, Nano Letters 15, 2794 (2015).
  14. [14]H.-Y. Chen, V. A. Jhalani, M. Palummo, and M. Bernardi, Phys. Rev. B 100, 075135 (2019).
  15. [15]G. G. Macfarlane, T. P. McLean, J. E. Quarrington, and V. Roberts, Phys. Rev. 111, 1245 (1958).
  16. [16]J. Noffsinger, E. Kioupakis, C. G. Van de Walle, S. G. Louie, and M. L. Cohen, Phys. Rev. Lett. 108, 167402 (2012).
  17. [17]C. E. Patrick and F. Giustino, Journal of Physics: Condensed Matter 26, 365503 (2014).
  18. [18]S. Tiwari, E. Kioupakis, J. Menendez, and F. Giustino, Phys. Rev. B 109, 195127 (2024).
  19. [19]P. Lechiffart, F. Paleari, D. Sangalli, and C. Attaccalite, Phys. Rev. Mater. 7, 024006 (2023).
  20. [20]Y.-h. Chan, J. B. Haber, M. H. Naik, J. B. Neaton, D. Y. Qiu, F. H. da Jornada, and S. G. Louie, Nano Letters 23, 3971 (2023).
  21. [21]F. Giustino, Rev. Mod. Phys. 89, 015003 (2017).
  22. [22]G. Antonius and S. G. Louie, Phys. Rev. B 105, 085111 (2022).
  23. [23]F. Paleari and A. Marini, Phys. Rev. B 106, 125403 (2022).
  24. [24]J. J. Sakurai and J. Napolitano, Modern quantum mechanics (Cambridge University Press, 2020).
  25. [25]R. Rosati, R. C. Iotti, F. Dolcini, and F. Rossi, Phys. Rev. B 90, 125140 (2014).
  26. [26]R. Rosati, F. Dolcini, and F. Rossi, Phys. Rev. B 92, 235423 (2015).
  27. [27]J. Xu, A. Habib, S. Kumar, F. Wu, R. Sundararaman, and Y. Ping, Nature communications 11, 2780 (2020).
  28. [28]J. Xu and Y. Ping, Journal of Chemical Theory and Computation 20, 492 (2024), pMID: 38157422, https://doi.org/10.1021/acs.jctc.3c00598.
  29. [29]R. Rosati, K. Wagner, S. Brem, R. Perea-Causin, E. Wietek, J. Zipfel, J. D. Ziegler, M. Selig, T. Taniguchi, K. Watanabe, et al., ACS Photonics 7, 2756 (2020).
  30. [30]M. Katzer, M. Selig, L. Sigl, M. T. Troue, J. Figueiredo, J. Kiemle, F. Sigger, U. Wurstbauer, A. W. Holleitner, and A. Knorr, Phys. Rev. B 108, L121102 (2023).
  31. [31]H.-Y. Chen, D. Sangalli, and M. Bernardi, Phys. Rev. Res. 4, 043203 (2022).
  32. [32]G. Onida, L. Reining, and A. Rubio, Rev. Mod. Phys. 74, 601 (2002).
  33. [34]J. D. Jackson, Classical electrodynamics (1999).
  34. [35]D.-K. Yang and S.-T. Wu, Fundamentals of liquid crystal devices (John Wiley & Sons, 2014).
  35. [36]F. Wu, D. Rocca, and Y. Ping, Journal of Materials Chemistry C 7, 12891 (2019).
  36. [37]J. Xu, A. Habib, R. Sundararaman, and Y. Ping, Phys. Rev. B 104, 184418 (2021).
  37. [38]J. Xu, K. Li, U. N. Huynh, M. Fadel, J. Huang, R. Sundararaman, V. Vardeny, and Y. Ping, Nat. Commun. 15, 188 (2024).
  38. [39]J. Quinton, M. Fadel, J. Xu, A. Habib, M. Chandra, Y. Ping, and R. Sundararaman, Phys. Rev. B 111, 115113 (2025).
  39. [40]G. Lasher and F. Stern, Phys. Rev. 133, A553 (1964).
  40. [41]H.-Y. Chen, V. A. Jhalani, M. Palummo, and M. Bernardi, Phys. Rev. B 100, 075135 (2019).
  41. [42]J. Im, A. Moritz, F. Steuber, V. Härle, F. Scholz, and A. Hangleiter, Applied physics letters 70, 631 (1997).
  42. [43]T. J. Boerner, S. Deems, T. R. Furlani, S. L. Knuth, and J. Towns, in Practice and Experience in Advanced Research Computing 2023: Computing for the Common Good, PEARC ’23 (Association for Computing Machinery, New York, NY, USA, 2023) p. 173–176.

Paper details

vlnc:W_hcu3cft2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents