Dark-State Coherent Control of Excitons in TMD Monolayers: A Reduced Intravalley Lindblad Framework for WSe$_2$

Technical Summary: Dark-State Coherent Control of Excitons in TMD Monolayers

Problem Statement
Transition-metal dichalcogenide (TMD) monolayers, such as WSe$_2$, host tightly bound excitons with strong spin-valley locking, making them promising platforms for optoelectronics and valleytronics. While coherent superpositions of excitonic states have been observed, actively engineering dark-state coherence (as in Electromagnetically Induced Transparency, EIT, or Stimulated Raman Adiabatic Passage, STIRAP) in these solid-state systems faces significant challenges absent in dilute atomic gases. The excitonic levels are embedded in a solid, coupled simultaneously to phonon reservoirs, the quantized radiation field, and a dilute exciton gas. These dissipative channels can destroy the quantum interference required for dark-state protocols. Furthermore, the exciton spectrum in TMDs features a strongly non-hydrogenic Rydberg series where bright (s-like) and dark (p-like) states are interleaved, complicating the selection of viable three-level manifolds. To date, a microscopic open-system treatment of EIT and STIRAP protocols for TMD excitons, including the rigorous selection of the excitonic manifold itself, has not been reported.

Methodology
The authors develop a topology-resolved, reduced Lindblad framework for driven three-level excitonic manifolds at a fixed center-of-mass momentum ($K_0$). The methodology proceeds in three stages:

1. Systematic Manifold Screening: A screening procedure selects candidate three-level manifolds from the exciton spectrum based on interband brightness, intra-excitonic dipole strength, and state compactness. This approach classifies manifolds by their drive architecture (interband optical vs. intra-excitonic THz/mid-IR) rather than strict energy ordering. Two topologies are evaluated: the ladder ($\Xi$) configuration and the $V$ configuration.
2. Microscopic Open-System Model: The selected manifolds are embedded in a multi-reservoir Lindblad framework derived under the Born–Markov and full-secular approximations. The tagged exciton is coupled to three environments: acoustic phonons, the quantized radiation field, and a dilute intravalley exciton gas.
3. Fixed-$K_0$ Reduction: A critical structural constraint is the restriction of the tagged exciton to a fixed center-of-mass momentum $K_0$. This kinematic restriction has profound consequences:
   • Phonons: The acoustic deformation-potential coupling vanishes identically because momentum conservation forbids momentum transfer to the lattice at fixed $K_0$. Thus, phonons generate no dissipator or Lamb shift in this reduced model.
   • Exciton Gas: The gas enters only as a coherent exchange mean field, renormalizing exciton energies and control detunings linearly with density, without introducing dissipation.
   • Radiation: The photon reservoir reduces to a one-dimensional continuum. Interband transitions retain their radiative width, while intra-excitonic emission is suppressed by the system area ($A^{-1}$). The resulting dissipator is purely radiative.

Key Results
The framework is applied to hBN-encapsulated WSe$_2$ at a baseline temperature of 4 K.

• Manifold Selection: The screening identifies the $\Xi$ manifold as $(|0\rangle, |1s\rangle, |2p\rangle)$, where an optical leg drives the ground-to-1s transition and a THz leg drives the 1s-to-2p transition. The $V$ manifold is selected as $(|0\rangle, |1s\rangle, |2s\rangle)$, with two optical legs.
• $\Xi$ Manifold (Dark-State Control):
  • STIRAP-like Conversion: Using a counterintuitive pulse sequence (THz pulse preceding the weak optical pulse), the system converts the excitation created by the optical pulse into the optically dark $2p$ exciton via dark-state following. The lossy intermediate $1s$ state remains nearly empty.
  • Efficiency and Loss: The conversion efficiency is limited primarily by emission from the intermediate $1s$ state rather than non-adiabatic coupling. The process is controlled by a single adiabaticity parameter of the THz leg. The maximum conversion efficiency reaches $\sim 48\%$ under optimal conditions, with the unconverted excitation lost to vertex emission.
  • Density Calibration: The exchange mean field induces a blue shift in the exciton levels. At moderate densities ($n_X \gtrsim 1.3 \times 10^{10}$ cm$^{-2}$), this shift displaces the two-photon resonance beyond the tolerance of the conversion. Consequently, the control fields must be calibrated to the actual exciton density. The center of the cascade transparency window provides an in situ method for this calibration.
  • EIT vs. Autler-Townes: The paper distinguishes between EIT and Autler-Townes splitting. True EIT (interference-dominated) is observed only below a specific control strength ($\hbar\Omega_{23} \lesssim 1.5$ meV); above this threshold, the spectrum transitions to an Autler-Townes doublet.
• $V$ Manifold (Reference):
  • In the $V$ configuration, both excited states ($1s$ and $2s$) are optically bright and radiatively decaying. The coherence between them decays on the radiative timescale ($T_2 \approx 0.37$ ps).
  • No transparency window forms between the spectrally isolated arms. Instead, the coherence manifests as a damped quantum beat. This manifold serves as an unprotected reference, highlighting the necessity of a dark state for robust coherence.

Significance and Claims
The paper claims to provide a minimal, material-specific baseline for dark-state coherent control of excitons in two-dimensional semiconductors in the radiative limit.

• Reproducibility: The manifold selection is derived systematically from the exciton spectrum and drive architecture, making the choice reproducible and transferable to other materials.
• Upper Bounds: The reported conversion efficiencies and coherence times are upper bounds. The reduced model omits pure dephasing channels (phonon scattering, intervalley processes) and finite-momentum recoil effects that would be present in a full treatment.
• Experimental Requirements: The work establishes specific requirements for realizing these protocols in real samples:
  • The dephasing of the dark $2p$ state must be sufficiently low (specifically, the measured linewidth must be dominated by radiative or other mechanisms rather than pure dephasing) to support the predicted coherence times.
  • The exchange mean field must be accounted for via density calibration.
  • Gated, charge-neutral samples are identified as the necessary environment to minimize residual carrier broadening of the dark state.

The authors conclude that while the current reduction is limited to the intravalley regime and low temperatures, it offers a rigorous starting point. Future extensions would need to incorporate finite momentum transfer (restoring phonon scattering and collisional broadening) and non-Markovian effects to fully describe real TMD monolayers.