Exciton Mirror: Principles and Applications

• Exciton mirror is an optical system where mirror action is governed by excitonic resonances rather than conventional coatings, enabling controlled and tunable reflection.
• These systems leverage coherent radiative coupling and momentum selectivity to achieve high reflection and tunable optical characteristics in monolayer semiconductors.
• Engineered platforms such as external mirrors, metasurfaces, and nanocavities enhance exciton mirror functionality, enabling applications in exciton-optomechanics and quantum light control.

Searching arXiv for recent and foundational papers on exciton mirrors to support the article. An exciton mirror is an optical system in which mirror action is generated or strongly modified by an excitonic resonance rather than by geometric thickness or a conventional metallic or dielectric coating. In its most direct form, a monolayer semiconductor such as MoSe$_2$ acts as an atomically thin resonant reflector: a driven excitonic polarization reradiates coherently, interferes destructively with the transmitted field, and redirects optical power into reflection. The same term is also used for architectures in which an external mirror, metasurface, or cavity reshapes the radiative environment of a two-dimensional exciton, thereby tuning linewidth, lifetime, absorption, emission, and strong-coupling behavior (Back et al., 2017, Zhou et al., 2019, Park et al., 2022).

1. Microscopic principle

The core exciton-mirror mechanism is coherent resonant scattering by a two-dimensional excitonic polarization. In the idealized homogeneous $2$D limit, the in-plane momentum $k$ is conserved, so an incident optical mode couples only to the exciton mode with the same $k$. This momentum selectivity is what lets a flat defect-free monolayer behave, for each in-plane momentum sector, like an effectively one-dimensional input-output channel. In that limit, perfect extinction of the transmitted coherent field is possible for weak resonant $s$- or $p$-polarized light at any incidence angle, provided the linewidth is radiatively dominated (Zeytinoglu et al., 2017).

A convenient input-output form used for monolayer MoSe$_2$ is

$$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$

with $\Gamma$ the exciton radiative decay rate and $\gamma_\mathrm{tot}$ the total Lorentzian linewidth. For one-sided illumination, the resonant extrema are

$2$0

In the ideal limit $2$1, one has $2$2 and $2$3, so a single atomic layer becomes a perfect resonant reflector (Back et al., 2017).

An equivalent formulation emphasizes dephasing explicitly. For a free-standing monolayer dominated by a single exciton resonance,

$2$4

where $2$5 is the radiative decay rate, $2$6 is the total population decay rate, and $2$7 is the pure dephasing rate. This makes the central criterion transparent: high reflection requires radiative decay to dominate over non-radiative decay, pure dephasing, and disorder broadening (Scuri et al., 2017).

2. Atomically thin semiconductor mirrors

The first direct experimental realizations used hBN-encapsulated monolayer MoSe$2$8. In a charge-controlled van der Waals heterostructure on fused silica, a $2$9 MoSe$k$0 monolayer embedded between $k$1 nm and $k$2 nm hBN layers produced $k$3 extinction of an incident field resonant with the exciton transition and a maximum reflection coefficient of $k$4. In representative low-NA measurements, fitting yielded $k$5, an extinction coefficient $k$6, and an extracted maximum monolayer reflection coefficient of $k$7. Across different locations on the flake, extinction values up to about $k$8 were observed, and comparison with the linewidth dependence implied $k$9 (Back et al., 2017).

A parallel study on hBN-encapsulated monolayer MoSe$k$0 on SiO$k$1/Si reported deconvolved monolayer reflectance up to $k$2 at $k$3 K, remaining above $k$4 up to about $k$5 K, with a vacuum radiative linewidth $k$6. At low temperature, the non-radiative decay and pure dephasing were each about an order of magnitude smaller than the radiative linewidth, and the radiative decay rate was more than $k$7 times larger than the non-radiative and pure dephasing rates (Scuri et al., 2017).

These measurements established that atomically thin mirror action is not set by geometric thickness. It is set by a large excitonic oscillator strength together with a radiative decay rate comparable to the total linewidth. The strongest response occurs near normal incidence and small in-plane momentum. In the MoSe$k$8 device on fused silica, increasing the detection NA from $k$9 to $s$0 reduced the maximal extinction from $s$1 to $s$2 and broadened the extracted linewidth from $s$3 meV to $s$4 meV, an effect attributed to electron-hole exchange and the finite-$s$5 exciton dispersion (Back et al., 2017).

Electrical tunability is integral to the concept. In charge-controlled MoSe$s$6, increasing electron density transfers oscillator strength from the neutral exciton to the attractive exciton-polaron and broadens the neutral-exciton or repulsive-polaron branch. The minimum transmission at that branch rises from about $s$7 to $s$8 as $s$9 is changed, corresponding to much weaker mirror action. At a fixed photon energy of $p$0 eV, the extinction drops from about $p$1 to $p$2 as $p$3 is swept from $p$4 V to $p$5 V (Back et al., 2017). In the SiO$p$6/Si implementation, gate voltage similarly switches the mirror on and off by suppressing the neutral exciton and transferring spectral weight into the weaker charged-exciton response (Scuri et al., 2017).

The linear optical regime can be exceptionally robust. In the fused-silica device, resonant single-mode excitation from $p$7 W/cm$p$8 to $p$9 W/cm$_2$0 produced no change in extinction, consistent with operation deep in the linear regime (Back et al., 2017).

3. External mirrors as exciton-control elements

A second major meaning of exciton mirror refers to systems in which an external mirror controls excitonic radiative coupling. In suspended hBN-encapsulated monolayer MoSe$_2$1 above a planar gold mirror, the membrane-mirror spacing $_2$2 is changed electromechanically. The downward-emitted exciton field reflects from the mirror, returns with a controllable phase, and interferes with the directly emitted field. Destructive interference and prolonged lifetime occur when the TMD-mirror distance is near an even integer multiple of $_2$3, while constructive interference and shortened lifetime occur near an odd integer multiple of $_2$4, with shifts due to skin depth and dielectric environment (Zhou et al., 2019).

The suspended geometry matters because it yields spatially homogeneous, nearly lifetime-broadened excitons. In one suspended device, the PL linewidth is $_2$5 meV and the neutral-exciton peak position is homogeneous across the suspended region. In the mirror device, electromechanical actuation tunes the reflectance-fit total linewidth from about $_2$6 meV to $_2$7 meV, while the environmental radiative linewidth changes from about $_2$8 meV down to $_2$9 meV, with a nonradiative contribution $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$0. Under resonant AC drive in a second device, the mechanical resonance at $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$1 with $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$2 yields $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$3 PL modulation, $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$4 modulation of the absolute reflectance at the exciton resonance, an inferred membrane deflection of $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$5 nm, and radiative-linewidth modulation of about $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$6 (Zhou et al., 2019).

The same strong excitonic optical response enables cavity-free exciton-optomechanics in suspended monolayer MoSe$$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$7. A suspended drumhead resonator shows an effective exciton linewidth $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$8 meV and on-resonance reflection contrast of $$E_\mathrm{out}^\mathrm{r}(\omega) = \frac{\Gamma/2}{\Gamma/2 - i(\omega - \omega_\mathrm{exc})} \left[(1-\eta)E_\mathrm{in}^\mathrm{r}(\omega)-\eta E_\mathrm{in}^\mathrm{l}(\omega)\right], \qquad \eta=\Gamma/\gamma_\mathrm{tot},$$9, consistent with nearly perfect monolayer mirrors. Near-resonant optical pumping then produces optical damping, anti-damping, and optical spring effects through photothermal backaction, with gate tunability arising from electrostatic symmetry breaking and gate-induced exciton shifts (Xie et al., 2021).

For interlayer excitons, whose transition dipoles are strongly out of plane, a conventional planar mirror is less effective. A plasmonic photonic crystal mirror composed of Au nanodisks on Au, separated from the emitter plane by dielectric spacers, was proposed to suppress the radiative decay of interlayer excitons in WSe$\Gamma$0/MoSe$\Gamma$1. In an optimized design, the average normalized radiative decay rate is reduced to $\Gamma$2, corresponding to a radiative lifetime enhancement of about $\Gamma$3 relative to free space and $\Gamma$4 relative to a homogeneous hBN medium (Park et al., 2021).

4. Engineered mirror platforms

Beyond planar mirrors, the exciton-mirror concept has been generalized to metasurfaces, nanocavities, mirror-backed nanoantennas, and organic excitonic films. These platforms do not all use the term identically, but they share a common principle: the optical boundary condition at the exciton is engineered so that radiative decay, linewidth, or strong-coupling phenomenology can be tailored without relying on bulk cavity thickness alone.

Platform	Mirror function	Representative reported effect
Plasmonic meta-mirror under monolayer MoSe$\Gamma$5	Phase-engineered vacuum-field interference	Radiative decay tuned by two orders of magnitude
Nanoparticle-on-a-mirror cavity	Ultrasmall gap cavity for bright and dark excitons	$\Gamma$6 near the pseudomode for $\Gamma$7 nm
WS$\Gamma$8/Au nanoantenna with monolayer WSe$\Gamma$9	Mirror reshapes Mie mode toward the monolayer	Rabi splitting $\gamma_\mathrm{tot}$0 meV in dark-field scattering
12 nm J-aggregated film in open microcavity	Excitonic layer itself becomes mirror-like	Excitonic mirror adds a $\gamma_\mathrm{tot}$1 phase and links cavity modes of different order

A plasmonic meta-mirror beneath hBN-encapsulated monolayer MoSe$\gamma_\mathrm{tot}$2 uses an Au bottom mirror, SiO$\gamma_\mathrm{tot}$3 spacer, Au nanodisk array, and HSQ planarization layer to control reflection phase at the exciton emission energy while keeping pump conditions nearly fixed. For neutral excitons, the average PL intensity ratio on meta-mirrors A:B:C is $\gamma_\mathrm{tot}$4, the median linewidths are $\gamma_\mathrm{tot}$5, $\gamma_\mathrm{tot}$6, and $\gamma_\mathrm{tot}$7 meV, and the extracted radiative decay rates are $\gamma_\mathrm{tot}$8, $\gamma_\mathrm{tot}$9, and $2$00 meV. An anisotropic nanorod meta-mirror further yields an H/V reflection phase difference of $2$01 at the emission frequency, linewidths $2$02 meV and $2$03 meV for H and V polarization, and extracted radiative decay rates of $2$04 and $2$05 meV (Park et al., 2022).

In the nanoparticle-on-a-mirror cavity, the metallic mirror and nanoparticle form a nanogap resonator with bright low-order gap modes and a high-density pseudomode near $2$06 eV. This geometry is especially effective for dark quadrupolar excitons because it provides extreme field gradients and strong coupling to high-order dark modes. For representative parameters, the paper finds $2$07 near the pseudomode for a $2$08 nm gap, and near $2$09 the spectral densities of dipolar and quadrupolar excitons become comparable despite their free-space radiative mismatch (Cuartero-González et al., 2019).

Mirror-backed dielectric nanoantennas provide a localized strong-coupling route. A $2$10 nm thick WS$2$11 nanoantenna on a $2$12 nm Au film with a $2$13 nm Ti adhesion layer and a monolayer WSe$2$14 wrapper exhibits a mirror-modified electric-dipole Mie resonance. For the WS$2$15/Au geometry, the simulated field reaches $2$16, compared with $2$17 for the same antenna on SiO$2$18. Strong coupling with the monolayer WSe$2$19 exciton then gives $2$20 in dark-field scattering, $2$21 in reflectance contrast, and $2$22 in simulation (Wang et al., 5 Jun 2025).

An organic variant embeds a $2$23 nm J-aggregated squaraine film in an open microcavity. Near the exciton resonance at $2$24 eV, the real part of the dielectric function becomes negative in the range $2$25–$2$26 eV, so the film behaves quasi-metallic even though it occupies below $2$27 of the cavity volume. In this regime the excitonic layer changes the lower mirror boundary condition from dielectric-like to metallic-like, adds a $2$28 phase, and links cavity modes of different longitudinal order. The shortest cavities approach a Rabi splitting of about $2$29 of the exciton energy, which the authors describe as the onset of ultrastrong coupling (Bennenhei et al., 24 Jun 2026).

5. Nonlinear and quantum-optical regimes

The exciton mirror is intrinsically a resonant many-body optical system, so nonlinear and quantum-optical behavior emerges once exciton-exciton interactions, Rydberg excitons, or cavity hybridization are included. In the ideal radiatively broadened theory of a flat defect-free TMD monolayer, repulsive exciton-exciton interactions can increase reflection with increasing power for blue-detuned excitation, because the interaction-induced blueshift pulls the exciton into resonance. At the same time, pair scattering into $2$30 modes depletes the coherent reflected beam and generates an outgoing two-mode squeezed state, so interactions also limit the maximum coherent reflection (Zeytinoglu et al., 2017).

Experiments on monolayer MoSe$2$31 indeed revealed distinct nonlinear regimes. Under continuous-wave excitation, the reflectance shows sudden jumps, strong hysteresis, and optical bistability for red-detuned light, attributed to exciton-induced lattice heating and the resulting thermal redshift. Time-resolved measurements indicate reflectivity jumps on nanosecond timescales, with about $2$32 ns for the downward jump and $2$33 ns for recovery. Under $2$34-ps pulsed excitation, the nonlinear response instead shows a blueshift and broadening attributed to exciton-exciton interactions, with fitted density-dependent coefficients $2$35, $2$36, and $2$37 (Scuri et al., 2017).

A quantum-optical formulation treats the monolayer as a two-dimensional quantum mirror. For a bare collective exciton resonance, the reflected amplitude is

$2$38

with nearly perfect reflection when non-radiative broadening is small. When finite-range interactions are introduced through Rydberg excitons, two scenarios arise. In the direct-coupling case, a fully blockaded illuminated spot behaves like a saturable quantum emitter and yields $2$39 for the reflected light. In the ladder-EIT case, the monolayer is transparent in the linear regime but a single transmitted photon creates a Rydberg excitation that blocks EIT for the next photon, switching the monolayer from high transmission back to high reflectance. The group delay is $2$40 for the bare mirror and $2$41 in the EIT configuration, and strong antibunching remains robust even for $2$42 and asymptotically for $2$43 in the weak-drive regime (Walther et al., 2021).

Embedding an atomically thin mirror in a $2$44D cavity yields additional regimes. A resonant cavity-exciton system can produce a dark resonance with width $2$45, while at large cavity detuning the lower-polariton bright resonance acquires a much narrower linewidth $2$46. Because this linewidth can be much smaller than the bare radiative and disorder scales, resonant excitation of the bright resonance can yield strong photon antibunching even when the exciton-exciton interaction strength is much smaller than the bare radiative decay rate (Zeytinoğlu et al., 2018).

A recent Heisenberg-Langevin treatment extends the nonlinear exciton-mirror problem to include retardation and long-range electron-hole exchange. In that theory, the expected optical bistability of the pumped $2$47 exciton mode is prone to modulational instability toward non-radiative surface polariton modes. Above threshold, the pumped $2$48D exciton gas acts as an optical parametric generator of twin polariton beams; below threshold, the mirror acquires phase-conjugating properties (Andreev, 10 Jun 2026).

6. Limitations, terminology, and scope

Unity reflection is limited by the same mechanisms that degrade any coherent resonant scatterer: non-radiative decay, pure dephasing, inhomogeneous broadening, disorder, scattering into uncollected or high-$2$49 modes, dielectric-environment asymmetry, finite spot size, and imperfect mode overlap. In monolayer MoSe$2$50, these effects appear experimentally as substrate-induced Fano line shapes, reduced radiative fractions, spatially varying linewidths from strain, and loss from the measured channels. The raw absolute reflection can also be dominated by interference with the surrounding stack, so intrinsic monolayer reflectivity must often be extracted by fitting rather than read directly from peak values (Back et al., 2017, Scuri et al., 2017).

The term itself is therefore best treated as a family resemblance concept rather than a single device geometry. In the narrowest sense, it denotes an atomically thin resonant reflector formed by an excitonic transition in a $2$51D semiconductor. In a wider sense, it includes mirror-controlled excitonic membranes, meta-mirror platforms that pattern radiative decay landscapes, and excitonic layers inside nanocavities or microcavities that themselves behave as mirror-like optical elements. A related but distinct usage appears in reconstructed near-$2$52 twisted MoSe$2$53/MoSe$2$54, where mirror-related AB and BA domains host interlayer excitons with opposite out-of-plane dipole moments and therefore opposite Stark responses; this is a symmetry statement about excitonic dipoles rather than a high-reflectance optical mirror (Sung et al., 2020).

Related extensions include chiral and magneto-optical reflective interfaces. In a coherently coupled exciton-plasmon system based on monolayer WSe$2$55 and Au nanodisks, the bare exciton at $2$56 with linewidth $2$57 imprints a narrow, magnetically tunable Fano feature onto the broad plasmonic differential-reflection spectrum, creating a spectral window with polarization-dependent reflectivity in an otherwise broadband opaque medium (Vadia et al., 2023). A different extension uses mirror-mediated imaging of the photonic component of exciton-polariton condensates to create coherent long-range coupling between spatially separated condensates, showing that the “mirror” concept also reaches beyond bare excitons into driven-dissipative polariton networks (Liang et al., 26 Jun 2025).

Taken together, these works define the exciton mirror as a resonant optical boundary condition engineered by excitons. The unifying requirement is not a particular stack, material, or cavity, but a regime in which excitonic coherence, radiative coupling, and optical interference are strong enough that a single excitonic transition materially determines reflection, transmission, linewidth, or vacuum-field coupling.