No monolayer semiconductor is ever truly alone. A sheet of tungsten diselenide one atom thick has no bulk beneath it to screen the attraction between an electron and a hole, so the electric field lines between them spill out of the material and into whatever sits above and below. Change that surroundings — swap a low-permittivity oxide for a high-permittivity one — and the exciton changes with it: its binding energy, its size, the balance between the neutral exciton and the charged trion that carries an extra carrier along. Every WSe₂ device result is therefore, in part, a result about its substrate. A paper published in Applied Physics Letters on November 2, 2026, makes that dependence systematic for the material as it is actually grown at scale.
Xinxin Yue and colleagues at Nankai University — working in the Ultrafast Electron Microscopy Laboratory and the Ministry of Education Key Laboratory of Weak-Light Nonlinear Photonics — studied chemical-vapour-deposition-grown monolayer WSe₂ under different dielectric environments, using temperature-dependent Raman and photoluminescence spectroscopy. Their headline finding is a clean one: both low temperature and a high-dielectric environment favour the formation of trions, so that the trion’s share of the emitted light grows relative to the neutral exciton. The mechanism they propose runs through the material’s defects: the dielectric environment drives a deep-to-shallow transition of defect levels, changing the effective charged defect population, and that in turn changes how excitons and trions form and recombine — differently for each species.
This is not a device record and it is not a new emission record. It is a calibration result, the kind that decides whether a spectrum measured in one laboratory transfers to a device built in another. And because the material here is CVD-grown rather than exfoliated, it is a calibration result for the growth route that could actually supply devices — the route that brings with it the defects, grain boundaries, and background doping that exfoliated flakes largely avoid.
In a monolayer transition metal dichalcogenide, the photoluminescence spectrum is not a single line. The brightest feature is usually the neutral exciton, X⁰: an electron and a hole bound together, recombining to emit a photon. A few tens of millielectronvolts lower in energy sits the trion, X⁺ or X⁻: the same pair with an extra hole or electron attached, a three-body complex whose emission is a direct reporter of the free-carrier population in the material. The relative weight of the two peaks — how much of the light comes from trions versus neutral excitons — is therefore a measurement of the material’s charge environment, not just of its band structure.
Yue and colleagues tracked exactly that ratio, across temperature and across dielectric surroundings. Their observation, as reported in the paper’s abstract, is that the trion gains ground on both axes: cool the sample down, and trions take a larger share of the emission; raise the dielectric constant of the surroundings, and the same thing happens. At low temperature the explanation is the familiar one — the trion binding energy is larger relative to the available thermal energy, so the three-body complex survives where warmth would tear it apart.
The mechanism: defects that change depth
The proposed mechanism deserves to be stated carefully, because it is the paper’s central interpretive claim and it is not the mechanism a reader might guess first. The naive expectation from dielectric screening is a tug-of-war: a higher-permittivity environment screens the Coulomb interaction more strongly, which reduces both the exciton binding energy and the quasiparticle bandgap, and the two reductions roughly cancel in the observed emission energy. That cancellation is well documented in WSe₂, and it is why the neutral exciton’s peak position is a poor thermometer for the dielectric environment.
Yue and colleagues argue for something more specific than generic screening. In their account, the dielectric environment changes the energetic depth of the defect levels in the WSe₂ — shifting them from deep to shallow — and that changes the effective population of charged defects. Charged defects are the source of the extra carriers that trions need: a trion cannot form without a free electron or hole to bind, and in an undoped monolayer, that carrier typically comes from a defect. If the dielectric environment makes more defects effectively charged, it supplies more trions, and the trion peak grows relative to the neutral exciton.
This is a claim about the interaction between two things the field usually treats separately: the dielectric environment as a screening parameter, and the defect population as a materials-quality parameter. The paper’s point is that they are not separate. The dielectric does not merely screen the exciton; it re-tunes the defects, and the defects re-tune the exciton population.
Two qualifications belong here. First, the quantitative details of the dielectric range tested — which environments, and the magnitude of the spectral shifts in millielectronvolts — are in the full paper rather than the abstract, and this article does not reproduce numbers it could not verify against the primary source. The trend and the mechanism are what the published abstract states. Second, the deep-to-shallow defect transition is the authors’ interpretation of their deconvolved spectra, not a direct image of a defect level moving. It is consistent with the data as reported, and it is a testable claim: it predicts that the trion fraction should track the charged-defect population, which can in principle be measured independently.
Why CVD matters
The choice of CVD-grown material is not incidental. Most foundational work on dielectric screening in WSe₂ used exfoliated flakes — small, pristine, and unrepresentative of anything manufacturable. CVD monolayers are the opposite: large-area and scalable, but carrying the selenium vacancies, grain boundaries, and background doping that come with growth at high temperature from vapour precursors. Their defect population is not a nuisance to be annealed away; it is a constitutive part of the material, and it is the very population whose charging the dielectric environment is claimed to control.
A photodetector, a light-emitting device, or a single-photon source built from CVD WSe₂ will sit on a dielectric — a silicon dioxide layer, a high-k oxide, an encapsulating hexagonal boron nitride flake — and the emission it produces will be the trion-weighted spectrum this paper describes, not the clean neutral-exciton line of an exfoliated flake on a favourite substrate. The paper’s guidance, as its abstract puts it, is for “dielectric engineering of optoelectronic and photonic devices based on transition metal dichalcogenides”: choose the dielectric deliberately, because it is choosing the exciton population for you.
It also reframes a familiar frustration. Groups comparing WSe₂ photoluminescence across laboratories have long found that nominally identical material gives different spectra, and the usual suspects have been growth variation, strain, and contamination. This work adds a systematic variable to that list: if two groups put the same CVD material on different dielectrics, they should expect different trion-to-exciton ratios, by a mechanism that has nothing to do with sample quality. That is a useful kind of result. It turns an apparent irreproducibility into a calibration curve.
The context: what screening was already known to do
The Nankai result lands in a literature that has been circling this question for a decade, and the earlier measurements give it scale. The cleanest demonstration remains the 2016 high-magnetic-field study by Stier, Wilson, Clark, Xu, and Crooker, who tuned the dielectric environment around exfoliated WSe₂ monolayers and measured the exciton’s size directly through its diamagnetic shift in fields up to 65 tesla. As the average dielectric constant of the surroundings rose from 1.55 to 3.30 times the vacuum permittivity, the exciton’s radius grew from about 1.2 to 1.6 nanometres and its binding energy fell from roughly 480 to 220 millielectronvolts — a factor-of-two change from a change of environment alone.
More recent work has pushed the same question onto CVD material and device-relevant dielectrics. A 2025 study in ACS Applied Optical Materials examined CVD-grown WSe₂ and WS₂ on silicon dioxide and hafnium dioxide — the low-k and high-k dielectrics of CMOS processing — and found that p-type WSe₂ shows higher photoluminescence intensity and red-shifted trion emission on the high-k HfO₂, while n-type WS₂ moves the other way. The sign of the response depends on the doping type, which is exactly what one expects if the dielectric acts partly through the carrier population rather than purely through screening. The Nankai paper’s defect mechanism is consistent with that pattern, and it gives it a microscopic story: the dielectric changes which defects are charged, and the charged defects set the carrier population that the trions report on.
What it is not
Three cautions, stated plainly because the framing invites overreading. First, this is not a demonstration of a better device. Nothing in the paper reports an efficiency, a responsivity, or a device figure of merit, and the guidance it offers is about how to think about dielectric choice, not a recipe for a record-breaking emitter. The value is in the calibration, not in a performance number.
Second, this is not a complete quantitative map of dielectric effects in WSe₂. The paper reports a trend — low temperature and high dielectric favour trions — and a mechanism — the deep-to-shallow defect transition. The specific dielectric environments tested, the size of the spectral shifts, and the temperature range are details that live in the full text. Any account that quotes those numbers without having read the paper would be inventing them, and this article does not.
Third, this is one growth route and one material. CVD WSe₂ is defect-rich by the standards of exfoliated flakes, and the defect-mediated mechanism the paper proposes may be correspondingly more important in CVD material than in cleaner samples. Whether the same deep-to-shallow transition governs the trion fraction in exfoliated WSe₂, or in other dichalcogenides such as MoSe₂ and WS₂, is an open question rather than a corollary. The mechanism should be tested on materials with different defect populations before it is treated as general.
What to watch
The most immediate follow-up is the obvious one: a dielectric series with published numbers — the same CVD material on a ladder of dielectrics from low-k to high-k, with the trion-to-exciton ratio and the defect population measured together, so that the claimed link between the dielectric, the defect depth, and the trion fraction can be tested quantitatively rather than qualitatively.
A second follow-up is to separate the dielectric’s two roles. The screening effect and the defect effect are entangled in any experiment that changes the substrate, because a new substrate changes both at once. Encapsulation experiments — the same flake, the same substrate, with and without a high-k capping layer — and gated devices, in which the carrier population is controlled electrically rather than by defects, could pull the two apart.
A third is to ask what this means for the single-photon emitters that have made WSe₂ a quantum-optics material. Those emitters are widely believed to be defect-bound excitons, and if the dielectric environment changes the depth of the defect levels that host them, it may also change their emission energy, brightness, and stability — which would make dielectric engineering a design variable for quantum emitters, not just for classical optoelectronics. That connection is speculative, and this paper does not make it. But it is the direction in which the defect mechanism points, and it is worth watching.
The deeper reason this result matters is the one stated at the outset. A monolayer semiconductor has no interior. Its electrons and holes live at the surface, in the open, and everything around them is part of their physics. The Nankai team has shown that the dielectric is not a passive support but an active participant — one that reaches through the defect population and decides which excitonic complex carries the light. The substrate is part of the exciton. Choose it accordingly.
Sources and further reading
Yue, X.; Chen, L.; Huo, Y.; Liu, J.; Chen, X.; Gao, C.; Feng, M.; Deng, S.; Liu, F.; Fu, X. “Influence of dielectric environment on the optical properties of the CVD-grown monolayer WSe₂.” Applied Physics Letters 129, 182101 (2026), published November 2, 2026. DOI: 10.1063/5.0344968. Ultrafast Electron Microscopy Laboratory, MOE Key Laboratory of Weak-Light Nonlinear Photonics, School of Physics, Nankai University.
For the benchmark measurement of dielectric screening in WSe₂, see Stier, A. V.; Wilson, N. P.; Clark, G.; Xu, X.; Crooker, S. A. “Probing the influence of dielectric environment on excitons in monolayer WSe₂: Insight from high magnetic fields.” Nano Letters 16, 7054–7060 (2016). The exciton radius grows from about 1.2 to 1.6 nm and its binding energy falls from roughly 480 to 220 meV as the surrounding dielectric constant rises from 1.55 to 3.30 times the vacuum permittivity. For the device-relevant dielectric comparison on CVD material, see the 2025 study of CVD-grown WSe₂ and WS₂ on SiO₂ and HfO₂ in ACS Applied Optical Materials 3, 1330–1338 (2025), DOI: 10.1021/acsaom.5c00105. For the earlier substrate comparison on exfoliated and CVD WSe₂, see Lippert, S.; et al. “Influence of the substrate material on the optical properties of tungsten diselenide monolayers.” 2D Materials 4, 025045 (2017). This article is analysis of published research, not investment advice.