---
title: Proximity-Induced Interlayer Excitons
url: https://www.emergentmind.com/topics/proximity-induced-interlayer-excitons
type: topic
---

# Proximity-Induced Interlayer Excitons

Proximity-induced interlayer excitons are Coulomb-bound electron–hole pairs whose constituents reside in different adjacent layers and whose existence, generation, fine structure, or optical activity is created or strongly modified by layer proximity. In atomically thin van der Waals bilayers this proximity is expressed through type-II band alignment, interlayer tunneling, layer hybridization, dielectric screening, moiré modulation, and permanent out-of-plane dipoles; in intrinsic multilayers and layered perovskites it can arise from adjacent inorganic sheets within the same crystallographic unit cell, which supply a genuine layer degree of freedom to the excitonic spectrum [2407.10713][1905.02391][2604.19327]. The term therefore covers both spatially indirect excitons that are weakly allowed or dark in the centrosymmetric limit and layer-hybridized excitons that retain substantial optical dipole while acquiring interlayer character [1903.02157][2108.04248][2509.21956].

## 1. Definitions, taxonomy, and material realizations

The basic distinction is between intralayer and interlayer excitons. Intralayer excitons place electron and hole in the same monolayer or inorganic sheet and typically exhibit large oscillator strength; interlayer excitons place the electron and hole in different layers, acquire a permanent out-of-plane dipole $p = e d$, and usually have reduced electron–hole overlap, smaller oscillator strength, and longer radiative lifetimes. In transition-metal dichalcogenide heterobilayers this distinction is commonly tied to type-II alignment, whereas in homobilayers and few-layer crystals it can emerge from layer-selective confinement or carrier-species-specific hybridization [2407.10713][1804.08412][1905.02391].

A second taxonomy concerns how “interlayer” the exciton actually is. In conventional spatially indirect interlayer excitons, the electron and hole are predominantly localized in different layers; in layer-hybridized excitons, one carrier remains layer-confined while the other is extended across both layers. Commensurate H-type WSe$_2$/MoSe$_2$ and WS$_2$/MoS$_2$ realize the latter situation through valence-band hybridization at $K$, yielding excitons that combine a large optical transition dipole with a large out-of-plane electric dipole [1903.02157]. Naturally 2H-stacked bilayer MoS$_2$ provides another nonstandard case: the observed interlayer exciton at $K$ has strong oscillator strength and is described as a hole state delocalized across the two layers bound to an electron localized in one of the layers [2108.04248].

The same language extends beyond TMDs. In few-layer 2H-WSe$_2$, the band-edge states at $K/K'$ are effectively confined to alternating layers, producing bright direct (quasi-)intralayer excitons and lower-energy interlayer dark excitons; the bright states are organized by the coupled spin, valley, and layer rule $l_z S_z T_z = +1$ [1905.02391]. In the $n=1$ Ruddlesden–Popper perovskite (PEA)$_2$PbI$_4$, the two Pb–I octahedral sheets in one crystallographic unit cell double the intralayer fine structure and enable spatially indirect interlayer excitons within the same unit cell [2604.19327]. Proximity-induced interlayer excitons have also been identified in magnetic WSe$_2$/CrI$_3$ stacks and in bilayer armchair graphene nanoribbons, where stacking, hybridization, and screening again determine whether the lowest excitations are intralayer, interlayer, or hybridized [2509.21956][2507.10887].

## 2. Microscopic routes by which proximity produces interlayer excitons

In many van der Waals heterobilayers, the minimal route is type-II band alignment. The review literature summarizes the corresponding exciton energy as $E_{\mathrm{IX}} \approx E_g^{\mathrm{type-II}} - E_b^{\mathrm{IX}} - p E_z$, with $p = e d$ set by the interlayer spacing [2407.10713]. In MoSe$_2$–WSe$_2$, the type-II offset places the conduction-band minimum in MoSe$_2$ and the valence-band maximum in WSe$_2$, so optical excitation of an intralayer exciton is followed by hole tunneling and formation of an interlayer exciton [1804.08412]. In twisted WSe$_2$/MoS$_2$, femtosecond photoemission momentum microscopy shows that interlayer excitons are dominantly formed on the sub-50 fs timescale via interlayer tunneling at the $K$ valleys, followed by slower formation through dark $\Sigma$-valley excitons and phonon cooling [2112.05011].

A second route is proximity-controlled exciton conversion rather than direct optical creation. In Janus-based heterobilayers such as MoS$_2$@WSSe, the intrinsic out-of-plane electric field of the Janus layer tunes the type-II offsets and hence the energy separation $\Delta E$ between bright intralayer and dark interlayer excitons. When $\Delta E$ is resonant with strong in-plane optical phonons, first-order exciton–phonon scattering can efficiently convert bright intralayer excitons into lower-lying interlayer excitons without an external gate [2212.05615]. For MoS$_2$@WSSe in AB stacking, the reported $\Delta E \approx 0.05$ eV matches the transverse optical-like $E$ modes near $40$ meV, and the projected exciton–phonon couplings reach $0.5$–$2$ meV [2212.05615].

A third route is interlayer hybridization. In commensurate H-type WSe$_2$/MoSe$_2$ and WS$_2$/MoS$_2$, momentum alignment and symmetry-allowed interlayer hopping hybridize the valence-band edges while the relevant conduction state remains layer-confined. The resulting two-level valence Hamiltonian,
$$
H_v=\begin{pmatrix}
\varepsilon_u & t_v\\
t_v & \varepsilon_l
\end{pmatrix},
$$
produces hybridized hole states whose excitons inherit both optical brightness and an electric dipole. Experimentally extracted parameters are $t_v = 43 \pm 9$ meV and $\Delta = 12 \pm 2$ meV in H-type WSe$_2$/MoSe$_2$, and $t_v = 36 \pm 4$ meV and $\Delta = 17 \pm 5$ meV in H-type WS$_2$/MoS$_2$ [1903.02157].

Intrinsic layered materials supply yet another route. In few-layer WSe$_2$, symmetry suppresses conduction-band hopping at $K/K'$ while strong spin–orbit splitting keeps the valence states effectively layer-localized, so adjacent-layer Coulomb binding yields lower-energy interlayer dark excitons without requiring a heterointerface [1905.02391]. In (PEA)$_2$PbI$_4$, structural inequivalence of the two adjacent inorganic layers generates a layer-dependent offset of band edges, furnishing interlayer transitions whose electron–hole probability density lies on adjacent layers [2604.19327].

## 3. Fine structure, selection rules, and optical activity

The fine structure of proximity-induced interlayer excitons is often inseparable from that of nearby intralayer states. In (PEA)$_2$PbI$_4$, polarization-resolved low-temperature photoluminescence at $8$ K resolves four bright intralayer transitions at $X_1 = 2.3402$ eV, $Y_1 = 2.3420$ eV, $X_2 = 2.3423$ eV, and $Y_2 = 2.3438$ eV, each with $\pm 0.2$ meV uncertainty. These form two closely spaced $X/Y$-like pairs associated with the two inorganic layers, with intralayer $Y-X$ anisotropy of $\approx 1.5$–$1.8$ meV and an $L1/L2$ centroid offset of $\approx 2.0$ meV. A weaker doublet at $2.3794 \pm 0.0002$ eV and $2.3832 \pm 0.0002$ eV, about $45$ meV above the bright intralayer states and split by $\approx 3.8$ meV, is assigned to interlayer excitons [2604.19327]. Static $G_0W_0$+BSE places the interlayer manifold higher, at $\approx 58$ meV above the bright intralayer states, and predicts small oscillator strengths; distortion-induced mixing with bright intralayer excitons is proposed to account for the observed brightening and nearly orthogonal in-plane polarizations [2604.19327].

Few-layer WSe$_2$ supplies a complementary fine-structure limit in which interlayer states are intrinsically dark. In bilayer WSe$_2$, the lower-energy $A1s(1)$ is interlayer and dark, whereas the higher-energy $A1s(2)$ is intralayer and bright; the calculated emission energies differ by about $5$ meV, and the bright state has about two orders of magnitude larger radiative rate when $\Gamma_{\mathrm{NR}}=0$ [1905.02391]. More generally, an $n$-layer film contains two doubly degenerate bright $A1s$ excitons, distributed over even or odd layers, while the remaining $2n-2$ $A1s$ bands are interlayer and dark [1905.02391].

Layer-hybridized excitons modify these selection rules by redistributing the oscillator strength between branches. In the H-type bilayers analyzed by Tang and co-workers, the optical dipoles scale as $|\mu_u^{\pm}|^2 \approx |\mu_0|^2 \{\cos^2\theta,\sin^2\theta\}$ and the electric dipoles as $p_u^{\pm} \approx e d \sin^2\theta$ [1903.02157]. Accordingly, H-type WSe$_2$/MoSe$_2$ shows a WSe$_2$-derived doublet split by $92$ meV with near-$1{:}1$ spectral weight, while H-type WS$_2$/MoS$_2$ shows a corresponding doublet split by $74 \pm 6$ meV; in R-type or twisted bilayers the doublets disappear because the interlayer hopping at $K$ is symmetry-forbidden or momentum-suppressed [1903.02157].

Coupling between interlayer and intralayer excitons can itself become a spectroscopic observable. In naturally 2H-stacked bilayer MoS$_2$, reflectivity-derived optical susceptibility reveals an avoided crossing between an interlayer exciton branch and the intralayer B exciton with positive coupling $\kappa_{\mathrm{IE-B}} = +35.8 \pm 3.6$ meV, and a weaker avoided crossing with the intralayer A exciton with negative coupling $\kappa_{\mathrm{IE-A}} = -3.5 \pm 0.4$ meV [2108.04248]. In magnetic WSe$_2$/CrI$_3$ and CrI$_3$/WSe$_2$/CrI$_3$, a robust interlayer magnetic exciton appears near $1.84$ eV between the WSe$_2$ A and B excitons, with oscillator strength nearly twenty times larger than the intralayer CrI$_3$ Frenkel excitons and dominant out-of-plane dipole character [2509.21956].

## 4. Spatiotemporal formation, localization, and transport

The temporal sequence of interlayer exciton formation has been established most clearly in TMD heterostructures. In MoSe$_2$–WSe$_2$, a $100$ fs Gaussian pump resonant with the MoSe$_2$ $1s$ intralayer exciton creates coherent intralayer polarization that decays within hundreds of femtoseconds via radiative emission and exciton–phonon scattering; interlayer hole tunneling then converts the population to interlayer excitons on a sub-picosecond timescale, with near-complete transfer by $\sim 2$ ps [1804.08412]. The interlayer photoluminescence at $1.39$ eV surpasses intralayer emission by around $4.5$ ps, and by $\sim 15$ ps the equilibrium intensity ratio $\mathrm{IeX/IaX} \approx 100$ at $77$ K [1804.08412]. In WSe$_2$/MoS$_2$, the trARPES signature is an interlayer-exciton peak at $\approx 1.2$ eV above the WSe$_2$ valence-band maximum, a characteristic three-peak momentum fingerprint at the mini-Brillouin-zone $\kappa/\kappa'$ points, and a wavefunction with approximately $60\%$ of its probability within one moiré unit cell [2112.05011].

The spatial structure of interlayer excitons is not universally localized. Real-space calculations for reconstructed H-type MoSe$_2$/WSe$_2$ show that at a twist deviation of $1.3^\circ$ from $60^\circ$, where the moiré period is $a_M^{1.3^\circ}=14.3$ nm and the reconstructed potential minima are about $-12$ meV, the ground-state interlayer-exciton center-of-mass wavefunction extends over multiple moiré cells and reaches the computational domain limit [2411.19616]. At $1.1^\circ$ there is a mixture of extended and localized states, whereas at $0.9^\circ$ the lowest states become localized within a single moiré unit cell and some excited states form quasi-1D quantum-wire-like modes [2411.19616]. Cryogenic photoluminescence supports this crossover: near $60^\circ$, the steady-state line at $E_{\mathrm{IX}} = 1.398$ eV is Lorentzian with FWHM $\approx 4$ meV and dominates at long delays, while earlier-time lines have FWHM values of $14$, $17$, and $21$ meV [2411.19616].

Transport at finite density is likewise proximity-sensitive. A microscopic theory for MoSe$_2$–WSe$_2$ derives the interaction energy per exciton as $\Delta E(n_x) = (g_{d-d}+g_{x-x})n_x + \Sigma_{\mathrm{CH}}$, where the repulsive dipolar term competes with attractive fermionic exchange and a Coulomb-hole shift [2207.03942]. Above $n_x \gtrsim 10^{12}\,\mathrm{cm}^{-2}$, the resulting drift term dominates over ordinary diffusion and drives highly nonlinear propagation; for $n_x = 5 \times 10^{12}\,\mathrm{cm}^{-2}$, the model yields $D_{\mathrm{eff}} \simeq 1.5\,\mathrm{cm}^2\,\mathrm{s}^{-1}$ around $t \simeq 1.5$ ns without spacer, and up to $\simeq 6\,\mathrm{cm}^2\,\mathrm{s}^{-1}$ at early times with hBN spacers [2207.03942]. Counter to the naive expectation that weaker environmental screening should speed transport, the theory finds that free-standing samples propagate more slowly than hBN-encapsulated ones because stronger screening reduces the magnitude of the attractive Coulomb-hole term, leaving more dipolar repulsion uncompensated [2207.03942].

## 5. Theoretical descriptions

Most microscopic descriptions start from a Bethe–Salpeter treatment of correlated electron–hole pairs. In the perovskite case, the exciton eigenproblem is written as
$$
H_{\mathrm{exc}} |S\rangle = \Omega_S |S\rangle,\qquad
H_{\mathrm{exc}} = (E_c-E_v)\delta - K,
$$
or equivalently $\Omega_S = (E_c-E_v) - E^{S}_{\mathrm{bind}}$, with oscillator strength
$$
f_S \propto \left|\sum_{vck} A^S_{vck}\langle c k |\, \mathbf{e}\!\cdot\!\mathbf{v}\,| v k\rangle\right|^2.
$$
Within this framework, layer-doubled intralayer manifolds and weak interlayer manifolds emerge naturally once the two-layer unit cell and symmetry lowering are included [2604.19327]. In Janus heterobilayers, the same BSE amplitudes are projected onto phonon modes through
$$
\mathcal{G}_{\alpha\beta}^{\mu}
= \sum_{vck}\left[\sum_{v'} (A_{\beta}^{cv'k})^* g_{vv'k}^{\mu} A_{\alpha}^{cvk}
-\sum_{c'} (A_{\beta}^{c'vk})^* g_{c'ck}^{\mu} A_{\alpha}^{cvk}\right],
$$
so that the interconversion rate follows Fermi’s Golden Rule and is maximal when $\Delta E \approx \hbar\omega_{\mathbf{q}\nu}$ [2212.05615].

For TMD heterostructures, an explicit two-body Dirac formulation has also been developed. Starting from monolayer massive Dirac Hamiltonians, one constructs a four-band interlayer-exciton Hamiltonian and couples it to a proximity-modified interaction
$$
V(\rho;h)=\frac{q_1 q_2}{4\pi\varepsilon_0}\int_0^\infty dq\,
\frac{J_0(q\rho)}{\varepsilon\!\left(q,-\frac{h}{2},\frac{h}{2}\right)},
$$
whose dielectric kernel includes the top, bottom, and barrier media, as well as polarization effects in the transition-metal and chalcogen layers [1809.01065]. This formalism predicts interlayer binding energies on SiO$_2$ of roughly $100$–$120$ meV for several TMD pairs at $h=0.6$ nm, larger average in-plane electron–hole separation than in intralayer excitons, and a linear Stark shift $\Delta E_X = -p_z E_\perp$ for the interlayer peak [1809.01065].

Effective Hamiltonians remain useful when the spectroscopy is dominated by a few interacting branches. In bilayer MoS$_2$, a coupled-oscillator susceptibility model,
$$
H = \sum_i E_i(F_z)|i\rangle\langle i| + \sum_{i\neq j}\kappa(|i\rangle\langle j|+|j\rangle\langle i|),
$$
combined with intensity asymmetries at avoided crossings, determines both magnitude and sign of the inter-exciton couplings [2108.04248]. At the many-body level, transport and condensation problems often reduce to density-dependent bosonic descriptions. For interlayer excitonic insulators, an explicitly solvable mean-field model uses an interlayer interaction $V_q = 2\pi e^2 F_q/(\varepsilon q)$ and predicts that including proximity-enhanced overlap in $F_q$ permits an interlayer excitonic-insulator phase up to $d \approx 2$ nm for $\varepsilon \approx 6.9$, whereas neglecting the overlap restricts the phase to $d < 0.5$ nm [2207.03501].

## 6. Control modalities, device concepts, and unresolved issues

Because interlayer excitons carry a permanent dipole, electric fields are a primary control axis. In MoS$_2$/WSe$_2$ on LiNbO$_3$, Rayleigh-mode surface acoustic waves generate in-plane and out-of-plane piezoelectric fields with $E_x \approx 2.07 E_z$, enabling contactless and selective manipulation of intralayer and interlayer species [2511.03301]. The interlayer excitons IX$_{K\Gamma}$ and IX$_{K-K}$, emitting near $1300$ nm and $1500$ nm with a robust splitting of $\approx 120$ meV across twist angles from $\approx 3^\circ$ to $\approx 50^\circ$, respond linearly through $\Delta E \approx -p E_z$ with extracted dipoles $p \approx 0.72\,e\!\cdot\!\mathrm{nm}$ and $0.86\,e\!\cdot\!\mathrm{nm}$, whereas intralayer excitons exhibit the quadratic Stark effect $\Delta E = -\tfrac{1}{2}\alpha E^2$ [2511.03301]. The same acoustoelectric fields strongly quench IX photoluminescence, with $Q_{\mathrm{PL}} \approx 0.1$ at $4$ K and $18$ dBm for IX$_{K-K}$, evolving to near-complete quenching at $70$ K [2511.03301].

Nanoscale electrostatic trapping uses the same dipolar physics in a more localized form. A nanopatterned graphene gate above MoSe$_2$/WSe$_2$ creates a sharp reduction of the out-of-plane electric field beneath a $\sim 15$–$20$ nm effective hole, producing a trap potential $V(r) = -pE(r)$ for the interlayer exciton [2103.08838]. The free IX tunes with slope $dE/dE = 0.59$ eV/(V/nm), implying $d \approx 0.59$ nm and $p \approx 0.59\,e\!\cdot\!\mathrm{nm}$, while the trapped IX tunes with $84 \pm 2\%$ of that slope because the local field is reduced [2103.08838]. At maximum trapping bias the trap depth is $42$–$48$ meV with lateral FWHM $\approx 22$ nm; the corresponding harmonic approximation gives $\hbar\omega \approx 5$ meV, a ground-state energy of $\approx 2.5$ meV, and a ground-state spatial width of $\approx 6.1$ nm [2103.08838]. The trapped state also shows low-power saturation at $P_{\mathrm{sat}} \approx 330$ nW and an increased lifetime $\tau_{\mathrm{trap}} = 15 \pm 1$ ns relative to $\tau_{\mathrm{free}} = 8.6 \pm 0.1$ ns [2103.08838].

Stacking and dimensionality broaden the design space further. In bilayer armchair graphene nanoribbons, vertical stacking produces both type-I and type-II alignments; in 6$\beta$ homobilayers the bright intralayer exciton lies at $\approx 1.30$ eV and an interlayer exciton at $\approx 1.62$ eV retains up to $\sim 13\%$ of the maximum absorption, while in 6-10AGNR heterobilayers the lowest exciton is interlayer at $\approx 1.18$ eV with oscillator strength $\approx 0.25\%$ of $X_1$ and a room-temperature radiative lifetime of $9.4\,\mu$s [2507.10887]. More speculative many-body control follows from proximity-engineered single-particle dispersions: if either the electron or hole acquires a Mexican-hat dispersion, the interlayer-exciton dispersion evolves from parabolic to quartic and eventually to a ring minimum at finite momentum, opening routes to liquid–solid transitions and finite-momentum condensates under external-field tuning [1602.00325].

Several recurring misconceptions are corrected by the present literature. In (PEA)$_2$PbI$_4$, the low-energy multiplet is consistently explained by exchange, symmetry lowering, octahedral distortions, and layer doubling, without invoking Rashba splitting or exciton–polaron physics as the primary origin; Rashba-like features can appear as a consequence of the same distortions that brighten the interlayer manifold [2604.19327]. In reconstructed MoSe$_2$/WSe$_2$, atomic reconstruction does not inevitably localize interlayer excitons; for sufficiently shallow reconstructed potentials, the ground state is laterally extended over multiple moiré cells [2411.19616]. In WSe$_2$/CrI$_3$, by contrast, the interlayer magnetic excitons are robust across bilayer and trilayer stackings, whereas the proximity-induced modifications of the WSe$_2$ A and B excitons depend sensitively on fine details of interface alignment and hybridization [2509.21956]. Taken together, these results place proximity-induced interlayer excitons at the intersection of interfacial band engineering, lattice symmetry, many-body screening, and nonequilibrium dynamics, rather than under a single universal mechanism.

Source: https://www.emergentmind.com/topics/proximity-induced-interlayer-excitons