A Markovian approach to $N$-photon correlations beyond the quantum regression theorem

Multi-photon correlations from quantum emitters coupled to vibrational environments lie beyond the reach of standard tools such as the quantum regression theorem (QRT). Here, we introduce a Markovian framework for computing frequency-resolved $N$-photon correlation functions that overcomes this limitation. Applying our approach to a driven semiconductor quantum dot provides a tractable description of phonon effects on fluorescence beyond the single-photon spectrum. Our method accurately captures the emergence of the phonon sideband, missed by conventional QRT treatments, and reveals rich phonon-induced structure in the filtered two-photon spectrum. Strikingly, we find that photons emitted via the phonon sideband inherit second-order coherence properties of the Mollow triplet.

💡 Research Summary

The paper addresses a fundamental limitation of the quantum regression theorem (QRT) when applied to quantum emitters interacting with structured vibrational (phonon) environments. The QRT assumes a flat environmental spectral density and neglects the finite frequency resolution of realistic detectors, which makes it unsuitable for describing phonon sidebands (PSBs) and higher‑order photon correlations in solid‑state systems. To overcome these shortcomings, the authors combine two powerful ideas: (i) the sensor method, which introduces auxiliary two‑level systems (sensors) weakly coupled to the emitter and acting as Lorentzian frequency filters, and (ii) a Markovian master‑equation treatment that traces out the phonon bath in the eigenbasis of the combined emitter‑sensor Hamiltonian. By doing so, the sensors directly sense energy exchanges between the emitter and phonons, allowing the structured environment to be incorporated without violating the Markovian framework.

The formalism starts from a standard Lindblad master equation for the emitter, adds N sensors with Hamiltonians H_m = ω_m ς_m† ς_m and weak couplings ε_m(σ ς_m† + σ† ς_m), and writes a joint master equation for the emitter‑sensor density matrix. The phonon bath is modeled as a collection of harmonic oscillators with a super‑Ohmic spectral density J_ph(ν)=α ν³ exp(−ν²/ν_c²). Using the Born‑Markov approximation in the interaction picture of the full emitter‑sensor Hamiltonian, the authors derive a phonon dissipator K(ρ) that depends on a rate operator Z constructed from the eigenstates |ψ_α⟩ of the combined system. This dissipator acts on the full emitter‑sensor space, ensuring that phonon‑induced transitions are correctly reflected in the sensor dynamics.

To demonstrate the method, the authors consider a resonantly driven semiconductor quantum dot (QD) modeled as a two‑level system with radiative decay γ and Rabi driving Ω. Electron‑phonon coupling is taken as A=σ†σ. They compute the single‑photon physical spectrum S^{(1)}(ω) using one sensor (Γ≈10⁻⁴ ps⁻¹) and compare three approaches: (a) the standard QRT, (b) the new sensor‑based Markovian method, and (c) the numerically exact TEMPO algorithm. The sensor method reproduces the PSB and the phonon‑renormalized Mollow sideband positions in quantitative agreement with TEMPO, while the QRT completely misses the sideband. This shows that even within a weak‑coupling Markovian picture, PSBs can be captured once the QRT is bypassed.

The authors then extend the analysis to two‑photon correlations by introducing two sensors with equal bandwidth Γ=2γ, enabling the calculation of the normalized second‑order spectrum g^{(2)}_{Γ₁,Γ₂}(ω₁,ω₂). The resulting two‑photon spectra reveal rich structure: without phonons, the familiar Mollow‑triplet correlations appear (central‑central uncorrelated, side‑side antibunched, central‑side antibunched, etc.). When phonons are included, a new set of three parallel diagonal features emerges within the PSB region, spaced exactly by the phonon‑renormalized Rabi frequency Ω_r. These features mirror the correlation pattern of the original Mollow triplet, indicating that photons emitted via the PSB inherit the same second‑order coherence properties as those from the zero‑phonon line. Moreover, the background antibunching seen without phonons becomes largely uncorrelated in the presence of the PSB, reflecting the additional emission pathways opened by phonon‑assisted processes.

Overall, the paper provides a conceptually simple yet powerful framework that (i) avoids the high‑dimensional integrals normally required for frequency‑resolved N‑photon correlations, (ii) faithfully incorporates structured vibrational environments within a Markovian description, and (iii) yields results that match numerically exact methods while remaining computationally inexpensive (e.g., a 16 × 16 matrix diagonalization per frequency for a two‑level emitter). The approach opens the door to systematic studies of higher‑order photon statistics in solid‑state quantum optics, phonon‑mediated quantum information protocols, and experimental spectroscopy where detector bandwidth and environmental structure play crucial roles.