Scalable Cyclic Olefin Copolymer Encapsulation for High Optical Quality of TMD Monolayers

Abstract: Monolayer transition metal dichalcogenides (TMDs) combine a direct bandgap, strongly bound excitons, and pronounced second-order optical nonlinearity, which makes them promising materials for ultrathin optoelectronic and nanophotonic devices. However, their optical performance is often degraded by environmental exposure and substrate-induced charge trapping, motivating the development of scalable encapsulation strategies. Here, we investigate spin-coated cyclic olefin copolymer (COC) as a scalable encapsulant for TMDs. Room-temperature and cryogenic optical spectroscopy reveal enhanced photoluminescence and second-harmonic generation, accompanied by excitonic linewidth narrowing and an increased exciton-to-trion ratio. In addition, COC encapsulation induces an excitonic peak splitting and an overall spectral blueshift. First-principles calculations attribute these spectral modifications to local symmetry breaking at the chalcogen interface and macroscopic compressive strain, respectively. These findings establish spin-coated COC as an effective, scalable encapsulation strategy and a potential platform for post-growth excitonic and band-structure engineering.

• The paper presents a scalable and cost-effective method using cyclic olefin copolymer (COC) encapsulation to enhance the optical quality of CVD-grown monolayer MoS2, achieving a sixfold cryogenic PL enhancement and a 71:29 exciton-to-trion ratio at 4 K.
• COC encapsulation significantly improves the dielectric interface, reduces substrate doping, and suppresses charge transfer, leading to enhanced photoluminescence and reduced inhomogeneous broadening.
• The encapsulation method introduces a new emission feature ($XU′$), attributed to a weakly allowed excitonic transition induced by the polymer interface, while also inducing compressive strain and breaking interfacial symmetry.

Monolayer transition metal dichalcogenides (TMDs) such as MoS₂ offer direct bandgaps, strongly bound excitons, and broken inversion symmetry, but their optical performance is routinely degraded by substrate-induced charge trapping, interfacial disorder, oxidation, and photodegradation. While hexagonal boron nitride (hBN) encapsulation is the established route to intrinsic optical quality, it relies on mechanically exfoliated crystals and labor-intensive dry stacking, limiting scalability. The work by Das et al. (2608.20973) demonstrates that spin-coated cyclic olefin copolymer (COC) provides a scalable, low-cost alternative that simultaneously enhances the linear and nonlinear optical response of CVD-grown monolayer MoS₂ and induces reproducible excitonic spectral modifications attributable to symmetry breaking and compressive strain.

Sample architecture and structural characterization

The samples consist of CVD-grown monolayer MoS₂ transferred onto fused silica substrates, either bare or pre-coated with a ~10 nm COC film (TOPAS 6013-S04), followed by a second spin-coated COC layer to form a symmetric COC/MoS₂/COC heterostructure. Atomic force microscopy shows that the COC layer is smoother than bare glass ($0.31 \pm 0.01$ nm versus $0.48 \pm 0.03$ nm rms roughness), establishing an improved dielectric interface. Room-temperature Raman spectroscopy reveals blueshifts of both the $E'$ and $A'_1$ modes after encapsulation, a signature of compressive strain in the monolayer.

Photoluminescence enhancement and charge neutrality

Room-temperature PL intensity increases by at least a factor of two upon COC encapsulation, accompanied by an A-exciton blueshift from 1.849 eV to 1.894 eV, consistent with the Raman-inferred compressive strain, with possible additional contributions from reduced substrate doping. The enhancement is attributed to suppressed charge transfer and fewer interface trap states.

The cryogenic results are more striking. At 4 K, averaged over 15 spectra from three spatially separated crystals, the PL intensity rises from $(5.4 \pm 1.2)\times10^3$ counts for the bare monolayer to $(29.8 \pm 3.0)\times10^3$ counts after encapsulation — an almost sixfold enhancement — with a neutral-exciton blueshift of $29.5 \pm 0.1$ meV. Spectral deconvolution into Lorentzian components quantifies the change in the exciton-to-trion ratio:

Condition	Bare MoS₂	COC-encapsulated
Exciton:trion ratio at 300 K	32:68	40:60
Exciton:trion ratio at 4 K	52:48	71:29

The shift toward neutral excitons indicates suppression of trion formation and hence a more charge-neutral environment. The neutral-exciton linewidth narrows from approximately 40 meV to 30 meV at 4 K, evidencing reduced inhomogeneous broadening from a more uniform dielectric environment. Reproducibility across two independent cooling cycles confirms the structural integrity of the encapsulated stack.

Polarization-resolved PL under resonant 633 nm excitation yields degrees of circular polarization (DOCP) of up to 74% (bare) and 72% (encapsulated) at 4 K. Because DOCP depends on the ratio of intervalley scattering to exciton decay rates, the persistence of high valley polarization despite the narrowed linewidth implies that COC encapsulation also suppresses intervalley relaxation — a nontrivial result given that improved optical quality alone would be expected to reduce DOCP.

The anomalous X^U state

COC encapsulation produces a narrow additional emission feature near 1.96 eV, denoted $X^U$, observed exclusively in the encapsulated sample. Differential white-light reflectivity shows a corresponding A-exciton splitting persisting up to roughly 210 K before merging into a single resonance; resonant SHG resolves a related feature even at room temperature, about 20 meV above the A-exciton, indicating that the splitting survives thermal broadening.

Several measurements constrain the nature of $X^U$. It exhibits a finite DOCP of up to 40% under blue-detuned excitation, retaining spin-valley selection rules. Its integrated intensity saturates sub-linearly with excitation power, unlike the linear response of the neutral exciton, suggesting a localized or weakly allowed transition rather than a bright free-exciton line. Wavelength-selective maps show that $X^U$, like $0.48 \pm 0.03$0, is distributed across the entire flake rather than confined to defect hotspots where the low-energy defect emission concentrates. The authors conclude that $0.48 \pm 0.03$1 is most plausibly a weakly allowed excitonic transition induced by the polymer interface, while explicitly conceding that its microscopic origin cannot yet be identified unambiguously.

Second-harmonic generation

SHG excitation spectroscopy shows increased intensity, a ~40 meV peak blueshift, and narrower resonance FWHM for the encapsulated sample relative to bare MoS₂ on glass, consistent with enhanced oscillator strength and exciton coherence. Polarization-dependent SHG retains the six-fold pattern expected for the $0.48 \pm 0.03$2 point group but deviates moderately from the ideal fit, indicating strain-induced local breaking of three-fold rotational symmetry — corroborating the Raman evidence for compressive strain.

First-principles interpretation

DFT calculations within GGA-PBE (with fully relativistic pseudopotentials) reproduce the dual spectral modifications through two cooperative mechanisms. A rigid lateral displacement of a single sulfur layer by 0.18 Å breaks the horizontal mirror symmetry $0.48 \pm 0.03$3, lifting the near-degeneracy of the lowest conduction-band states at K and thereby splitting the A-exciton transitions — the proposed origin of $0.48 \pm 0.03$4. Separately, 0.5% biaxial compressive strain acts symmetrically on both sulfur sublayers but rigidly upshifts the conduction band minimum, accounting for the macroscopic blueshift. The authors acknowledge that semi-local DFT underestimates absolute band gaps; the analysis is therefore framed as identifying qualitative mechanisms rather than predicting quasiparticle energies.

Limitations and open questions

Several caveats bear directly on the central claims. The microscopic identity of $0.48 \pm 0.03$5 remains unresolved; the DFT model uses an idealized rigid chalcogen displacement as an exemplary proxy for amorphous-polymer-induced shear, so the actual interfacial geometry is not determined. The room-temperature absorption resonance likely contains unresolved overlapping transitions, and the assignment of the SHG feature to $0.48 \pm 0.03$6 relies on accounting for bandgap renormalization. The spatial PL maps show localized emission hotspots attributed to defect-bound excitons, indicating residual inhomogeneity introduced by the encapsulation process itself. Finally, whether the strain magnitude and symmetry-breaking strength can be controlled via COC thickness or processing conditions — a prerequisite for deliberate post-growth band-structure engineering — is left open.

Conclusion

This work establishes spin-coated COC as a scalable encapsulant that delivers many of the optical-quality benefits of hBN — sixfold cryogenic PL enhancement, linewidth narrowing from ~40 meV to ~30 meV, a 71:29 exciton-to-trion ratio at 4 K, preserved valley polarization above 70%, and enhanced SHG — through a transfer-compatible, wafer-scale deposition process. Beyond passivation, the encapsulation deterministically modifies the excitonic spectrum via interfacial symmetry breaking and compressive strain, suggesting polymer coatings as a route to post-growth excitonic engineering. Resolving the precise atomic-scale origin of the $0.48 \pm 0.03$7 state and demonstrating controllability of the induced strain remain open problems.