WorksPhysics and Astronomy
Optical characterization of Fermi polarons in monolayer MoSe encapsulated in amorphous silicon nitride
Abstract
Amorphous silicon nitride is the leading low-loss, CMOS-compatible platform for wafer-scale integrated photonics. Encapsulating transition-metal dichalcogenide (TMD) monolayers within this platform offers a promising route towards scalable TMD-photonic circuits. Yet the effect of this amorphous dielectric on the optical response of the monolayer remains poorly understood. Here, we compare monolayer MoSe encapsulated in amorphous silicon nitride with a hexagonal boron nitride-encapsulated reference, using differential reflectivity and photoluminescence. Whereas the reference is dominated by the neutral exciton, the nitride-encapsulated monolayer shows attractive and repulsive polaron resonances separated by ~meV, with emission dominated by the attractive branch. A quantitative comparison of temperature-dependent reflectance with a microscopic finite-temperature Fermi-polaron theory shows that the nitride induces substantial electron doping in the monolayer, with an electron density of –~cm. Using the extracted density and trion binding energy as the only input parameters, the theory reproduces the photoluminescence spectra up to K in excellent quantitative agreement with experiments, thus corroborating the Fermi-polaron description of the observed excitonic response.
Introduction
The integration of transition metal dichalcogenide (TMD) monolayers with photonic circuits is a promising route towards compact and scalable light sources, modulators, and nonlinear elements. Silicon nitride is the natural platform for this goal: it is the workhorse dielectric of low-loss, CMOS-compatible integrated photonics, combining wafer-scale deposition, foundry availability, and tight optical confinement in waveguides and resonators, with broadband transparency spanning the exciton transitions of most TMDs [1, 2]. Routing and multiplexing have been demonstrated on this platform [3, 4]. Monolayer TMDs have accordingly been interfaced with silicon nitride in many geometries of interest: waveguides and resonators, for modulation, nonlinear conversion, and emission control [5, 6, 7, 8, 9, 10]; and patterned membranes and metasurfaces, for strong light-matter coupling and photoluminescence (PL) enhancement [11, 12]. The silicon nitride, as a photonic platform, may also be engineered to host its own intrinsic single-photon emitters [13]. In these directions, hexagonal boron nitride (hBN) offers a less versatile route: the growth and transfer of large-area, single-crystal films have yet to reach the maturity required for seamless integration with photonic-foundry processing [14, 15].
In nearly all of these demonstrations, however, the monolayer sits in direct contact with the nitride, and its optical response has not been fully characterized. This is important because the Coulomb interaction in a two-dimensional crystal is only weakly screened, and thus the properties of the excitonic and charged-exciton complexes are highly sensitive to the dielectric and electrostatic environment of the monolayer [16, 17, 18]. This sensitivity can be a resource, since it allows the response to be engineered, but it can also become a liability, since disorder at the monolayer-substrate interface—in the form of surface roughness, physisorbed residues, and substrate charge traps—inhomogeneously broadens the resonances [19], obscuring information about the intrinsic properties of the monolayer. hBN is the standard remedy: its atomic flatness, chemical inertness, and wide band gap allow encapsulated monolayers to reach PL and absorption linewidths approaching the homogeneous limit, with residual charge disorder essentially eliminated [20, 21, 22]. A recent comparative study of monolayer WS found substantially greater inhomogeneous broadening in silicon nitride environments than in hBN, attributing the difference primarily to substrate-induced charge trapping [23]. Whether silicon nitride can serve as an effective protective and encapsulating layer for TMDs, using hBN as the benchmark, therefore hinges on its electrostatic influence on the monolayer.
In particular, sub-stoichiometric amorphous nitride (SiN) grown by plasma-enhanced chemical vapor deposition has been shown to induce substantial -type doping in adjacent WSe flakes [24], and dielectric-induced charge doping has recently been reported for MoS on hBN, ZrO, and SiO substrates [25]. More recently, studies of MoSe at a GaInP interface [26] have shown that charged excitonic complexes acquire substantial oscillator strength in absorption and exhibit suppressed carrier recoil in PL, consistent with their identification as attractive Fermi-polaron resonances. Other studies have demonstrated substrate-induced doping using flexible conducting redox-polymer substrates and strain-tunable plasmonic metasurfaces for controlling and probing Fermi-polaron resonances [27]. However, the doping induced by SiN encapsulation remains largely unexplored, with previous estimates relying on indirect indicators such as linewidth broadening [23].
In this work, we compare monolayer MoSe encapsulated in hBN and amorphous SiN using differential-reflectivity and PL spectroscopy, and we establish that SiN encapsulation induces substantial electron doping in the monolayer. In particular, we observe that the SiN-encapsulated monolayer exhibits both attractive (AP, trion-like) and repulsive (RP, exciton-like) Fermi-polaron resonances, whereas the optical response of the hBN-encapsulated reference is dominated by the neutral exciton. Through a quantitative comparison with a microscopic finite-temperature Fermi-polaron theory [28, 29], we estimate the electron density and trion binding energy, which are then used to describe the PL spectra at different temperatures with no additional fitting parameters. The theory reproduces the AP emission profile well at low temperatures, where the RP emission is very weak, and yields excellent agreement with both the AP and RP resonances at the highest temperatures. These results provide direct optical evidence for substrate-induced charge-transfer doping in SiN-encapsulated TMDs, and quantify its influence on the excitonic response of monolayer MoSe. Since the optical dielectric constants of the two hBN and SiN encapsulants are similar, the results further show that it is the charged defects of the amorphous SiN network that set the excitonic response of the monolayer.
Results
hBN versus SiN encapsulation
Figure Figure 1 compares the two reference samples: MoSe monolayer encapsulated in hBN (Figure 1a) and in SiN (Figure 1b). The absorptive response, measured as differential reflectivity at K, is shown in Figure 1c. The hBN-encapsulated monolayer displays a single, well-isolated resonance at 1.6425 eV, which we assign to the neutral exciton ; no charged-exciton feature is resolved above the noise, indicating a low density of resident carriers. The SiN-encapsulated monolayer instead exhibits two resonances: a stronger one at 1.657 eV and a weaker one 31 meV below it, which we identify as the RP and AP polaron branches, respectively. The sizeable AP oscillator strength provides spectroscopic evidence of a nonzero resident electron density, since the AP acquires oscillator strength only in the presence of carriers [30, 31].
Figure 1. hBN- versus SiN-encapsulated monolayer MoSe at K. a hBN/1L MoSe/hBN reference sample (top and bottom hBN flakes outlined in green and orange, respectively, and monolayer in dashed white). b SiN/1L MoSe/SiN sample (top and bottom membranes illustrated by cyan and yellow solid lines respectively, monolayer regions dashed). In both b and c, the white circle within the monolayer region indicates the location at which the reflectance and PL measurements were performed, while the inset depicts the cross section of the sample. c Differential reflectivity, , of the SiN (red) and hBN (blue) samples, where an offset between the traces is added for the sake of clarity. d PL spectra of the same samples, each normalized to its maximum, on a logarithmic scale. and denote neutral and trion excitons in the sample with hBN capping, respectively; AP and RP denote attractive and repulsive polarons in the SiN-encapsulated monolayer MoSe sample.
The corresponding PL spectra are shown in Figure 1d. For the hBN-encapsulated monolayer, PL is dominated by with the trion appearing 25 meV below it at roughly a quarter of the peak intensity, the standard signature of a weakly and unintentionally doped MoSe monolayer[32, 33]. The emission in the SiN-capped monolayer is instead dominated by the AP: a single line at 1.6235 eV, with the RP barely visible in log scale at 1.655 eV. Furthermore, the AP-RP splitting is larger than the exciton–trion splitting in the hBN-capped sample, consistent with a larger electron density[30, 31]. The AP dominance of the PL emission is not restricted to the measured spot: spatially resolved PL maps show the AP line as the dominant emission across several disconnected monolayer islands of the same stack (Supplementary Note 2 and Supplementary Fig. 2).

Figure 2. Temperature dependence of reflectance spectra of the SiN-encapsulated monolayer and comparison with the Fermi-polaron theory. a spectra for different temperatures, from K to 90 K, where a trace offset is introduced for the sake of clarity. Two dashed lines indicate the AP/RP resonances as a guide to the eye. b Normalized oscillator strengths (green symbols) and (orange symbols) extracted from the reflectance spectra; solid lines are the calculated for (blue) and 0.9 (red) cm. c AP-RP energy splitting : experiment (symbols) and theory for the same two densities, with the same colour code as in b; shaded areas span –28 meV and solid lines correspond to the central value, meV. Error bars in b and c are obtained from a Monte Carlo analysis of the spectral fits (Methods).
Carrier density from temperature-dependent reflectance
To quantify the doping, we investigate the temperature evolution of the reflectance of the SiN-encapsulated monolayer from 8 to 90 K (Fig. 2a). Both resonances are resolved at all temperatures and redshift in parallel by meV, following the temperature dependence of the band gap of MoSe[34] (Supplementary Figure 4). From each spectrum we extract the AP–RP splitting, , and the oscillator strengths , defined as the integrated spectral weights of the two resonances normalized such that ; error bars are obtained from a Monte Carlo analysis (Methods). We find that the splitting changes little, from 30.9 0.6 meV at 8 K to 32.4 0.4 meV at 90 K (Fig. 2c), while the measured AP oscillator strength fluctuates between and 0.26 without a strong systematic dependence on temperature (Fig. 2b).
Figure 4. Supplementary Fig. 4. Temperature dependence of and PL of the SiN-encapsulated sample. Top row, a–c: analysis; bottom row, d–f: PL analysis. Orange circles, repulsive polaron (RP); blue squares, attractive polaron (AP). a Energies of the reflectivity minima. b Normalized area of the two resonances (sum to unity). c FWHM of the reflectivity resonances. d PL peak energies. e Integrated PL intensity of the AP line (log scale). f FWHM of the AP PL line. Error bars are shown where larger than the markers.
We apply the same procedure to reflectance spectra calculated with the Fermi-polaron model (Methods), which depend on two unknown parameters, the electron density and the trion binding energy . No calculation of is available for MoSe encapsulated in SiN. Since the optical dielectric constant of the nitride is close to that of hBN (Supplementary Note 5), we explore the interval –28 meV, around the – separation of our hBN reference. Fitting the temperature dependence of the experimental oscillator strengths (Fig. 2b) and energy splitting (Fig. 2c) to the theoretical predictions, we obtain – cm and meV. This substantial electron doping is comparable to that achieved with other doping substrates[26, 27], while the trion binding energy is close to the – separation of the hBN-encapsulated reference, meV, as expected from the similar dielectric screening of the two environments. A loss of electrons from the monolayer with increasing temperature, e.g., by thermal emission into traps of the nitride, would transfer oscillator strength from the AP to the RP and reduce the splitting. Neither occurs: a single, temperature-independent density accounts for both observables (splitting and relative area) up to 90 K.
Figure 3. Temperature dependence of the PL for SiN-encapsulated monolayer and comparison with Fermi-polaron theory. Measured PL spectra (light colors) for different temperatures, from 8.7 to 80 K. Solid dark lines: PL calculated with the Fermi-polaron theory using the parameters obtained from reflectance, and meV. Curves are normalized to their maxima and vertically offset for clarity. The weak emission corresponding to the RP at eV is magnified (for both theory and measured curves) as indicated, except for K.
The doping has a natural origin in sub-stoichiometric SiN that is grown by plasma-enhanced chemical vapor deposition. In this case, it was shown that positively charged dangling-bond centers can transfer an electron density of up to to an adjacent WSe monolayer[24], which is consistent with the extracted densities here. Dielectric-induced doping has also been reported for MoS[25], and charge trapping by the nitride was identified as the main source of inhomogeneous broadening in SiN-supported WS[23].
Photoluminescence and the polaron-to-trion crossover
Turning to PL, we see in Fig. 3 that the emission at low temperatures is dominated by the AP, while the RP progressively gains emission intensity as the temperature increases. This agrees with the expectations based on theory, since the emission spectrum is related to the exciton spectral function through the so-called detailed-balance relation [35] (see Method section). In other words, the relative intensities of the two branches are determined by their oscillator strengths, weighted by the corresponding thermal occupation factors. We find that the magnification required to resolve the RP emission decreases from at 8.7 K to at 60 K, while at 80 K it is directly resolved. Over the same range, the AP line broadens from 10 to 22 meV FWHM, mostly above 50 K, and its integrated intensity falls by a factor of (Supplementary Fig. 4). The low-temperature linewidth is the same over the whole scanned monolayers and independent of the explored excitation powers (Supplementary Figs. 2 and 3).
With the parameters extracted from reflectance, the theory reproduces the PL spectra in the entire range from 8.7 to 80 K (Fig. 3): namely, the relative energy between AP and RP peaks, the width of the AP line, its thermal broadening, and the emergence of the weak RP emission from 40 K upwards. Within the explored range of doping density and trion binding energy , the calculated spectra differ mainly in the weak RP branch, which is best seen on a logarithmic scale (Supplementary Note 7): a larger shifts the RP further from the AP and lowers its intensity. The PL model therefore supports the parameter set obtained from reflectance. At 8.7 and 30 K, small discrepancies between theory and experiment arise in the low-energy tail of the AP emission, which is broader in the experimental spectra. This broadening can be attributed to a low-energy disorder tail induced by the SiN encapsulation, which is not accounted for in the model. Furthermore, the discrepancy between theory and experiment in the RP emission intensity at low temperatures may arise from a slow thermalization rate between different polarons such that the populations have not fully equilibrated before the exciton decays.
Extracting both the doping density and the trion binding energy allows us to compare the relevant energy scales of the problem, namely the Fermi energy , and , and thus to identify the physical regime probed by our experiments. Finite-temperature Fermi-polaron theory predicts that, for , the nature of the AP is controlled by the ratio [28]. In the quantum-degenerate regime, , the AP is a well-defined quasiparticle. When , however, the trion–hole continuum subsumes the attractive branch such that the AP loses its quasiparticle character. This temperature-driven polaron-to-trion crossover implies a qualitative change in the shape of the AP PL emission profile, even though the AP–RP splitting and oscillator strengths remain relatively constant[28]. In our SiN-encapsulated monolayer, taking , the extracted density corresponds to – meV, and the crossover condition is reached at – K. Therefore, the attractive branch evolves from a polaron quasiparticle towards the trion-continuum dominated regime across the temperature series of Figs. 2 and 3. Indeed, both the thermal broadening of the PL line reproduced in Fig. 3 and the observed onset of the AP broadening above 50 K (Supplementary Fig. 4) are direct signatures of the polaron-to-trion crossover.
Concluding remarks
To conclude, we have found that encapsulating monolayer MoSe in amorphous SiN drives it into a doped Fermi-polaron regime. A Fermi-polaron theory with a temperature-independent electron density, – cm, accounts for the distribution of oscillator strength, the AP–RP splitting, and the PL spectra up to 80 K, demonstrating that the charge transferred from the nitride persists over this range. Importantly, doping from charged defects of the amorphous nitride is expected to depend on the stoichiometry and the deposition conditions, which must therefore be controlled when integrating TMD monolayers on silicon-nitride photonic circuits. Combined with SiN resonators and metasurfaces that already reach exciton-photon strong coupling[11], the Fermi sea provided by the nitride offers a gate-free route to Fermi polaron-polaritons on a platform compatible with wafer-scale integrated photonics.
Methods
Sample fabrication
Both heterostructures were assembled by deterministic transfer techniques. The SiN membranes are commercial, 50-nm-thick, ultra-low-stress amorphous silicon nitride windows (PELCO, Ted Pella), with a specified surface roughness of 0.65 nm[36]. They were released from their Si frames with the nail-polish technique [37] and placed on the target substrate. The hBN crystals were grown by the high-pressure, high-temperature method [38], and the MoSe crystals were purchased from HQ Graphene. MoSe and hBN flakes were mechanically exfoliated with adhesive tape onto PDMS stamps and dry-transferred at room temperature [39]. Both samples were then annealed in vacuum at 200 °C for 2 h. Throughout this work, SiN denotes this amorphous, chemical-vapor-deposited nitride, from which our membranes are made, as opposed to crystalline SiN. We note that, unlike atomically flat hBN, the membranes leave roughness as a possible source of inhomogeneous broadening [23]. The supplier describes the films as non-stoichiometric but considers their deposition method and composition proprietary, and has no data on their refractive index or fixed-charge density; we treat these properties as unknown (Supplementary Note 5).
Optical spectroscopy
Optical measurements were performed in an open-loop helium cryostat (CryoVac) with the sample held at temperatures between 8 and 90 K. Spectra were recorded with a SpectraPro 2500i spectrometer (Acton Research) coupled to a PIXIS 1024 × 1024 CCD. PL was collected through a 20×, NA=0.45 objective (Olympus LCPlan N IR) following off-resonant excitation with a green, continuous-wave laser (532 nm); residual laser light was rejected with a 550 nm long-pass filter. For differential reflectivity, the sample was illuminated with a white lamp and was obtained from spectra taken on () and off () the monolayer area. Spatially resolved maps were acquired by scanning the excitation spot across a fixed collection path using a relay and a steering mirror, which images the mirror onto the back focal plane of the objective [40], with the PL signal collected by a 20×, NA=0.40 objective (Mitutoyo M Plan Apo NIR). For momentum-resolved polarization tomography, a flip-mounted lens images the back focal plane of the objective onto the spectrometer slit, and the emission is analysed by motorized half- and quarter-wave plates followed by a fixed linear polarizer; the normalized Stokes parameters are reconstructed pixel by pixel from six polarization projections of the spectrally integrated AP line. The polarization properties of the AP emission are further discussed in Supplementary Note 6. Full schematics of both optical configurations are given in Supplementary Fig. 1.
Spectral analysis
The oscillator strengths were obtained as the integrated areas of the two resonances above a local baseline, normalized to unit sum, which removes the common optical prefactor of the stack, and as the separation of the two reflectivity minima. The uncertainties of and reflect the choice of baseline and of the integration limits of the two resonances. They were estimated by a Monte Carlo procedure in which these choices are varied randomly within reasonable bounds and the analysis is repeated; each error bar is the standard deviation of the resulting distribution. The same procedure was applied to the calculated reflectance spectra. Details of the analysis of the PL maps and of the azimuthal analysis of the Stokes parameters are given in Supplementary Notes 2 and 6.
Fermi-polaron theory
We model the optical response of an electron-doped MoSe monolayer using the finite-temperature Fermi-polaron framework developed in Ref. 28, extending earlier zero-temperature theories of exciton polarons [30, 31]. In this framework, the optically excited excitons are dressed by particle-hole excitations of the thermal Fermi sea of excess electrons, leading to AP and RP resonances whose energies and optical spectral weights are controlled by the electron density , the temperature, and the trion binding energy. The optical response is determined by the exciton Green’s function, , where the energy is measured relative to the bare exciton resonance. The self energy encodes the many-body dressing of the exciton by the surrounding electron gas, and is given by in terms of the exciton-electron matrix,
Here, are the electron and exciton dispersions, and are the corresponding effective masses, and is the electron Fermi-Dirac distribution. We use to calculate the spectral function , while the PL spectrum follows from the detailed-balance relation[28, 35] . To account for residual broadening not captured by the polaron model, is then convolved with a Lorentzian with full width at half maximum of 10 meV, fixed by the low-temperature experimental linewidth (Supplementary Note 7).
Data availability
The data that support the findings of this study are openly available in e-cienciaDatos at https://doi.org/XX.XXXXX/YYYYYY.
Acknowledgements
C.P., L.V., M.D.M., F.M.M., and C.A.-S. acknowledge the support from the projects from the Spanish Ministry of Science, Innovation and Universities through the “Maria de Maeztu” Programme for Units of Excellence in R&D (CEX2023-001316-M), projects PID2023-148061NB-I00, PCI2024-153425, and MCIN/AEI/10.13039/501100011033, FEDER UE, project No. PID2023-150420NB-C31 (Q). K.W. and T.T. acknowledge support from the CREST (JPMJCR24A5), JST and World Premier International Research Center Initiative (WPI), MEXT, Japan. C.A.-S. acknowledges the support from the Comunidad de Madrid fund “Atracción de Talento, Mod. 1”, Ref. 2020-T1/IND-19785 and the project from the Ministerio de Ciencia e Innovación CNS2025-165107. M.M.P. is supported through Australian Research Council Future Fellowship FT200100619, and J.L. through Australian Research Council Discovery Project DP240100569. C.P. and C.A.-S. acknowledge the technical support of Fabiola Mogollón and Antonio Flores and the General Support Services for Experimental Research of the UAM (SEGAINVEX).
Author contributions
A.C.-G. proposed the study of SiN capping and supervised the sample assembly. R.L.M. assembled all the samples. K.W. and T.T. provided the hBN crystals. C.P. performed the experiments, analysed the data, and prepared the figures. A.C.L. assisted with the experiments, analysed the data, and prepared the figures. S.P.V. assisted with the experiments and the data analysis. B.C.M. developed the theory, performed the simulations, and helped prepare the figures. J.L., M.M.P., and F.M.M. supervised the theory development, and F.M.M. provided theoretical feedback during the experiments. M.D.M. co-supervised C.P. C.A.-S., F.M.M., M.M.P. and J.L. co-wrote the manuscript, with contributions from all authors. C.A.-S. supervised the whole project. All authors discussed the results.
Competing interests
The authors declare no competing interests.
Experimental setups
Supplementary Figure 1 shows the two optical configurations used in this work, described in the Methods of the main text. Both share the same open-loop helium cryostat and detection system. Configuration a is used for the PL and reflectance spectra and, with the lens inserted, for the momentum-resolved polarization tomography of Supplementary Note 6. Since the open-loop cryostat leaves no space for cryogenic nanopositioners, the sample-scan configuration b is used for the spatially resolved maps of Supplementary Note 2: a relay images the steering mirror onto the back focal plane of the objective, so that the excitation spot is translated across the sample while the collection path remains fixed[40].
Spatially resolved PL of the SiN-encapsulated sample
To establish whether the SiN-induced doping inferred in the main text is spatially uniform, we mapped the low-temperature PL over the region enclosed by the green dashed box in Supplementary Fig. 2a, which contains several disconnected 1L MoSe islands within the same SiN/MoSe/SiN stack. At each point of the map, a linear spectral background was subtracted using the average intensities on the low- and high-wavelength sides of the measured spectral window. The dominant emission energy was obtained from the intensity-weighted average of the spectral points around the emission maximum. The FWHM was extracted from the two half-maximum crossings using linear interpolation between two adjacent spectral points. Lastly, the integrated PL intensity was obtained by numerical integration of the background-subtracted spectrum. The resulting integrated intensity, peak energy, and FWHM are shown in Supplementary Fig. 2b, c, and d, respectively.
The islands are readily identified in the intensity map (Supplementary Fig. 2b). The peak energy (Supplementary Fig. 2c), lies within 1.625–1.632 eV everywhere, i.e. the emission is the attractive-polaron line AP identified in Figure 1d of the main text over the entire mapped area; the neutral exciton is nowhere the dominant feature. Although a small spatial variation of the emission energy is observed across and between the individual MoSe islands, no region exhibits a shift towards the neutral-exciton energy. This demonstrates that the predominance of the AP emission is not restricted to a particular position on the sample, but persists throughout the spatially separated monolayer regions within the same SiN/MoSe/SiN stack.
The linewidth map (Supplementary Fig. 2d), likewise shows that the AP emission remains with a similar FWHM throughout the mapped monolayer regions. Together, the peak-energy and linewidth maps demonstrate that the spectral characteristics assigned to the SiN-induced electron doping in the main text are reproduced across multiple spatially disconnected MoSe islands, rather than arising from a single locally doped region of the sample.
Excitation-power dependence
Supplementary Fig. 3 shows the dependence of the SiN-encapsulated sample emission on excitation power at K, over the range 0.015–0.265 mW. The spot size presents a diameter of around 2 µm. Two points relevant to the main text follow from it:
First, the inset of Supplementary Fig. 3a resolves the weak RP emission near 1.655 eV that is at the limit of detection in Figure 1d of the main text. Its integrated intensity is smaller than that of AP by roughly two orders of magnitude at every power investigated, and the ratio does not change appreciably over the range studied. The inversion of spectral weight is therefore an intrinsic property of the doped monolayer and not an artifact of the excitation conditions.
Second, the emission linewidth of AP (Supplementary Fig. 3b), is constant within ±0.2 meV over an eighteen-fold change in excitation power, while the integrated intensity (Supplementary Fig. 3c), grows monotonically. The measurements reported in the main text are thus taken in a regime where neither excitation-induced heating nor photo-induced screening of the resident carriers modifies the line shape, so that the temperature-dependent linewidths of Supplementary Fig. 4f can be attributed to the intrinsic temperature dependence.
To quantify the excitation-power dependence, the integrated intensity of the AP emission was evaluated over the energy interval 1.590–1.644 eV. The upper integration bound was chosen to isolate the AP branch and exclude the weak higher-energy RP feature. A constant baseline defined by the minimum measured intensity of each spectrum was used for the integration. No additional background subtraction was applied, since the remaining background could not be reliably determined without substantially affecting the very weak higher-energy emission. The resulting integrated intensity was fitted with a power law yielding , consistent with a linear increase of the AP emission with excitation power throughout the investigated range. The uncertainties associated with both the extracted FWHM and the integrated PL intensity are smaller than the corresponding data markers and are therefore omitted from Supplementary Fig. 3b,c for clarity. Together, the approximately linear power dependence and nearly power-independent linewidth show no evidence of pronounced saturation or excitation-induced modification of the AP emission over the range of powers investigated.

Figure 2 (PDF p. 16). a Optical micrograph of the SiN_x/1L MoSe_2/SiN_x stack; the green rectangle marks the region mapped in b–d. The white circle corresponds to the monolayer region used for the PL and reflectance experiments. Scale bar, 40 µm. Inset: schematic cross section of the heterostructure. b Spectrally integrated PL intensity, revealing the spatially separated 1L MoSe_2 islands within the mapped region. The displayed intensity range was chosen to provide optimal contrast across the different PL-emitting regions; consequently, a small number of high-intensity pixels exceed the upper limit of the colour scale. c Spatial variation of the peak energy of the dominant emission line, shown as in meV, where eV is the minimum energy of the range of interest. The emission remains within the energy range associated with the attractive-polaron line AP across the different MoSe_2 islands. d Full width at half maximum (FWHM) of the dominant emission line, showing the spatial variation of the AP linewidth across the individual MoSe_2 islands. The PL maps were acquired using a Olympus objective with a spatial step size of 1 µm and an estimated excitation power of 0.140 mW at the sample plane. Grey pixels in b–d denote points where the PL intensity was below the threshold used for reliable extraction of all the spectral parameters. Isolated pixels observed away from the PL-emitting MoSe_2 islands, consistently appearing in b–d, are attributed to cosmic-ray events detected by the CCD and do not represent sample emission. Scale bars, 20 µm.
Figure 3 (PDF p. 17). a PL spectra for excitation powers between 0.015 and 0.265 mW. The spectra are dominated by the attractive-polaron line AP. Inset: magnified view ( ordinate scale) of the high-energy region containing the weak repulsive-polaron emission RP. b Full width at half maximum of the AP line versus excitation power. c Integrated PL intensity of the AP line versus excitation power, integrated over the energy range 1.590–1.644 eV to exclude the higher-energy RP feature. The solid line shows a fit to a power law, yielding . Error bars in b and c are smaller than the corresponding data markers and are therefore omitted.
Temperature dependence of the SiN-encapsulated sample
Supplementary Fig. 4 shows the spectral parameters extracted from the reflectivity and PL spectra of Figs. 2a and 3 of the main text, which are not repeated here.
Energies are those of the dip minima in reflectivity (Supplementary Fig. 4a) and of the peak maxima in PL (Supplementary Fig. 4d). The FWHM (Supplementary Fig. 4c,f), is the full width at half maximum above the local baseline. The reflectivity areas (Supplementary Fig. 4b), are normalized to sum to unity, which removes the common optical prefactor of the stack, and the PL intensity (Supplementary Fig. 4e), is the area of the AP line over a window excluding the RP. Two limitations apply. The dips are slightly asymmetric because of a dispersive admixture from the stack, which gives the absolute energies a systematic uncertainty of a few meV that should largely cancel in the temperature dependence and in the AP–RP splitting. The widths of the weak AP dip and of the 80 K PL line, which overlaps with a low-energy tail and the RP emission, are poorly constrained, as the error bars show. The PL intensity is comparable between temperatures only if the collection efficiency stays constant. The error bars of the AP–RP splitting and of the normalized areas (Fig. 2b,c of the main text), reflect the uncertainty associated with the choice of baseline and of the integration limits of the AP and RP resonances; they were estimated by the Monte Carlo procedure described in the Methods of the main text.
In reflectivity both resonances redshift in parallel (Supplementary Fig. 4a), from 1.660 to 1.650 eV (RP) and from 1.629 to 1.618 eV (AP) between 8 and 90 K, accelerating above K as expected for a semiconductor band gap[34], so that the splitting stays at –32 meV (Fig. 2c of the main text). The RP keeps –90% of the weight with no discernible trend, Supplementary Fig. 4b; a thermal release of carriers into the nitride would instead transfer weight from the AP to the RP, so the persistent AP weight is consistent with a constant density. The RP FWHM is –18 meV without a clear trend, whereas the AP FWHM (–27 meV, errors up to meV) is too uncertain to show one (Supplementary Fig. 4c).
In PL the AP and RP lines redshift[34] (Supplementary Fig. 4d), from 1.623 to 1.617 eV and from 1.657 to 1.648 eV between 8.7 and 80 K, and lie and meV below the reflectivity minima at low temperature. The AP intensity (Supplementary Fig. 4e), falls by a factor of , slowly below K and steeply above, with an apparent activation energy of meV between 50 and 80 K that we do not assign to a specific channel. The RP emission grows relative to the AP, as the magnification needed to display it in Fig. 3 of the main text falls from at 8.7 K to at 60 K. The AP FWHM (Supplementary Fig. 4f), stays at 10–11 meV up to 50 K and then grows to 14 and 22 meV at 60 and 80 K, the latter less certain. This onset coincides with the temperature, –60 K, at which the crossover condition is met for the extracted density (–3.6 meV), and is consistent with the crossover to the trion-dominated regime discussed in the main text. Altogether, the AP persists in absorption up to 90 K with an almost constant weight and splitting from the RP, while the emission is quenched and broadened thermally.
Dielectric environment and range of trion binding energies
The trion binding energy enters the Fermi-polaron model as an input parameter, and no calculation of it is available for MoSe encapsulated in amorphous silicon nitride. This section summarizes the arguments that restrict the range explored in the main text, –28 meV.
Each membrane is 50 5 nm thick [36], about 45 times the r.m.s. radius of the MoSe exciton, nm [41], and well above the nm range in which the screening of the nitride is reduced near its surfaces [42]. Each membrane therefore screens the monolayer as a bulk dielectric, characterized by its optical dielectric constant , with the refractive index below the gap. The supplier considers the deposition method and stoichiometry of the films proprietary and has no data on their refractive index; must therefore be bracketed with literature values, which depend on stoichiometry and deposition method. At the exciton wavelength, nm, stoichiometric CVD and LPCVD give –[43, 44], i.e. –, and a 100-nm PECVD film gives [45], whereas Si-rich, low-stress LPCVD nitride, the type commonly used for membranes, reaches in 100-nm films[45]. The supplier describes its ultra-low-stress films, which include our 50-nm membranes, as non-stoichiometric and closer to SiN than to [36], so we expect – for our nitride, similar to the value of hBN, –[41].
Replacing hBN with should therefore change only slightly. For comparison, the trion binding energy of changes only from meV on [32] to meV in our hBN reference (Figure 1d of the main text), although the screening changes much more between these two environments. Since the – separation measured in PL includes the doping effects of residual electrons, it is an upper bound for the trion binding energy in hBN[33]. The range – meV, centered close to this value, therefore covers the expected binding energy in the nitride. The value extracted from the AP–RP splitting, meV, lies in the middle of this range.
Momentum-resolved polarization tomography
Encapsulation is not polarization-neutral. In angle-resolved measurements on monolayers, hBN encapsulation was found to imprint a pronounced linear polarization on the exciton emission at large emission angles, reaching a degree of linear polarization of , an effect traced by finite-difference time-domain modelling to the uniaxial optical anisotropy of the hBN layers themselves; in a suspended monolayer, where that extrinsic contribution is absent, the residual momentum-dependent polarization is several times smaller and appears only at cryogenic temperatures, where it is attributed to the intrinsic -dependent electron–hole exchange coupling that splits the exciton dispersion into TE and TM branches46. Amorphous has no crystalline optic axis, so the hBN-like extrinsic contribution should be absent here; conversely, residual stress in the membranes could in principle imprint a preferred in-plane axis. Both possibilities are tested by the same measurement.
We performed a full Stokes tomography of the AP emission in the Fourier plane, using the configuration of Supplementary Figure 1 with the lens inserted. The full AP line is spectrally integrated at each point, so the analysis is sensitive to the polarization texture in momentum space but not to any underlying energy splitting of the emitting branches. The normalized Stokes parameters , and , reconstructed pixel by pixel from six polarization projections, are shown as functions of in-plane momentum in Supplementary Figure 5–Supplementary Figure 5, and the degree of polarization DOP in Supplementary Figure 5.
Supplementary Figure 1. Optical setups. a Polarization-tomography configuration. (1) 532 nm continuous-wave excitation laser; (2) beam-expanding telescope; (3) CCD used for white-light imaging of the sample through a flip-mounted pellicle beam splitter fed by an 880 nm light-emitting diode; (4) , objective focusing onto the 1L MoSe heterostructure held in the cryostat; (5) polarization-analysis stage comprising motorized half-wave (HWP) and quarter-wave (QWP) plates followed by a fixed linear polarizer, with the flip-mounted lens and the relay – selecting between real-space and Fourier-space imaging; (6) 550 nm long-pass filter and spectrometer. F.P. and I.P. denote Fourier and image planes, respectively. b Sample-scan configuration. A relay images the steering mirror (7) onto the back focal plane of the objective, so that the excitation spot is translated across the sample while the collection spot on the spectrometer slit stays fixed.
Supplementary Figure 5. Momentum-resolved Stokes tomography of the SiN-encapsulated sample. a–c Normalized Stokes parameters , and of the AP emission as functions of the in-plane momentum . The parameter showed no significant polarization features, indicating the absence of a pronounced circular-polarization component. d Corresponding degree of polarization, DOP . The ring-shaped feature observed near in the maps a–d is attributed to an experimental artifact, possibly arising from a dust particle or an etalon fringe in the detection path. e Azimuthal dependence of (red circles) and of the DOP (green diamonds), obtained by binning the maps in a and d as a function of the in-plane angle . The red dash-dotted line is a fit to with a -periodic modulation; the green dashed line marks the angle-averaged degree of polarization, 0.064. The Stokes analysis was performed at 10 K with an excitation power of 0.215 mW at the sample plane. The -space scale was determined considering PL emission centered at 759 nm and an objective with a numerical aperture of NA . The corresponding maximum in-plane momentum was estimated to be .
All three Stokes parameters remain below 0.15 in magnitude everywhere within the collection numerical aperture, and the circular component (Supplementary Figure 5), is featureless at the noise level, consistent with the non-resonant, above-gap excitation used here, which does not preserve valley polarization. Azimuthally binning the maps (Supplementary Figure 5), reveals a weak but systematic modulation of with a period of in the in-plane angle , of amplitude , superposed on a degree of polarization that is independent of and averages 0.066.
A fixed in-plane axis, such as the one that would be imprinted by uniaxial residual stress in the membranes, produces a Stokes vector that is constant in the laboratory frame and therefore a -independent . The -periodic observed here, accompanied by a -independent DOP, is instead the signature of a polarization axis that rotates with the azimuthal angle, i.e. of an imbalance between the - and -polarized (equivalently TM- and TE-like) components of the far field, which is precisely the decomposition used in Ref.46. The residual anisotropy is thus of the same nature as that reported for suspended monolayers, and not a uniaxial strain axis.
The amplitude found here, , is roughly three times smaller than the reported for hBN-encapsulated at the edge of a comparable collection cone, and is instead of the order of the suspended case 46. This is what the absence of a crystalline optic axis in amorphous SiN would predict: the nitride does not add the extrinsic, birefringence-driven polarization anisotropy that hBN does. In this regard, SiN encapsulation is the less perturbative of the two environments, which is a favourable property for the polarization- and valley-resolved experiments that an integrated platform is ultimately meant to support.
Finally, the same Fourier-plane data bound the orientation of the emitting dipole. An out-of-plane dipole component radiates no power along the surface normal and contributes purely radially polarized light that grows steeply towards the edge of the collection cone, so that the in- and out-of-plane dipole weights can be extracted from the momentum dependence of the polarized far field 47,48. We observe no such steep radial component within our numerical aperture, consistent with the in-plane dipole orientation established for monolayer MoSe 47 and for MoSe–WSe interlayer excitons, which are in plane 48. The attractive polaron therefore radiates as an in-plane dipole, like the neutral exciton, and contact with the nitride does not introduce out-of-plane dipole character.
Azimuthal analysis of the Stokes parameters and
To investigate the momentum dependence of the polarization texture, the and maps were analyzed along annuli of fixed . Each annulus corresponds to a constant emission angle , with its radius in the Fourier-plane image determined according to , where is the radius corresponding to the numerical aperture of the objective (). Eight equally spaced emission angles were considered, ranging from to the largest angle for which the complete annulus remained within the objective NA. An annular half-width of two pixels was used for each emission angle, i.e. pixels around the central radius. Within each annulus, the Stokes parameter was evaluated as a function of the in-plane azimuthal angle in momentum space.
The data were grouped into azimuthal bins, and the mean Stokes parameter within each bin was calculated as , where is the Stokes parameter of the th pixel within the corresponding azimuthal bin and is the number of contributing pixels. Only bins containing at least three valid pixels were retained. We take the standard deviation of the Stokes parameter to calculate the uncertainty of the mean Stokes parameter of each bin as the standard error of the mean (SEM) .
For each emission angle, the azimuthal dependence was first fitted directly to the binned mean Stokes parameters using the twofold angular modulation , , where is the modulation amplitude, is the phase of the angular modulation, and is a constant offset. In this initial fit, , , and were treated as free parameters. When valid SEM values were available for at least four azimuthal bins, the fit was weighted by the corresponding SEM values; otherwise, an unweighted fit was performed. This initial fitting step therefore determined the best-fit parameters , , and directly from the azimuthally-binned mean Stokes parameters and their associated SEM values. The modulation amplitude was defined to be positive, with the corresponding phase adjusted accordingly.
The asymmetric uncertainties in were determined in a subsequent step using the SEM of the azimuthally binned data. First, the binned Stokes values were separated into positive and negative lobes according to whether they lay above or below the best-fit offset . An outer SEM envelope was then constructed by adding the SEM to the positive-lobe values and subtracting the SEM from the negative-lobe values, thereby increasing their excursions from . Conversely, an inner SEM envelope was constructed by subtracting the SEM from the positive-lobe values and adding the SEM to the negative-lobe values, thereby reducing their excursions from .
The outer and inner SEM envelopes were subsequently fitted independently while fixing and , such that only the modulation amplitude was allowed to vary. The fit to the outer envelope yielded , whereas the fit to the inner envelope yielded .
The asymmetric uncertainties of the best-fit modulation amplitude were then defined as and . The resulting best-fit modulation amplitudes and their asymmetric uncertainties were finally plotted as a function of the emission angle to examine the evolution of the polarization modulation with increasing emission angle and, equivalently, increasing in-plane momentum.
Supplementary Figure 6. Momentum-resolved azimuthal analysis of the Stokes parameters and . a shows the map together with the fixed- annuli used for the analysis, spanning increasing emission angles from the inner (dashed line) to the outer region (solid line) of the accessible -space. b shows the corresponding modulation amplitude (dark circles) extracted from the azimuthal fits as a function of emission angle , with asymmetric error bars determined from the SEM-envelope analysis. c and d show the equivalent analysis for , where the modulation amplitude is illustrated in dark squares.
The amplitude of the azimuthal modulation increases with emission angle and, consequently, with the in-plane momentum . In particular, the fitted modulation amplitude increases from 1.65% at to 4.95% at , corresponding to an approximately threefold increase over the investigated momentum range. At the largest emission angle, the modulation reaches a comparable value of 5.06%, independently confirming the maximum linear-polarization modulation obtained from . Furthermore, the fitted phases of and differ by approximately on average, close to the phase difference expected between the two Stokes components of a rotating linear-polarization texture.
The amplitude of the modulation can be directly compared with the -resolved degree of linear polarization (DOLP) reported by Han et al. ⁴⁶, as both provide a measure of the az imuthmally varying linear-polarization contrast in momentum space. Han et al. observed a similar momentum dependence in monolayer MoSe, with the amplitude of the DOLP increasing monotonically towards larger 46. In particular, a DOLP modulation amplitude of approximately 6.7% was reported at for suspended monolayer MoSe at 10 K, in close agreement with the modulation of 4.95% measured here at the comparable emission angle of . For suspended MoSe, this momentum-dependent polarization response was associated with the TE–TM exciton splitting arising from long-range exchange interactions. The increase of the modulation with emission angle therefore supports a momentum-dependent origin of the observed linear-polarization texture, with both its magnitude and momentum dependence being comparable to those reported for suspended MoSe, while remaining substantially smaller than the enhanced response observed for the hBN-encapsulated structure.
Impact of the parameters of the Fermi-polaron model on the PL simulations
In the main text, we extracted the doping density and the trion binding energy by comparing the temperature dependence of the RP–AP energy splitting and the oscillator strengths extracted from reflectance data with the predictions of the Fermi polaron model [28]. This analysis allowed us to extract the parameter ranges – and –. Here, we show in Fig. 7 that the expected PL spectra, calculated for these ranges of doping density and trion binding energy, exhibit very similar features and are in good agreement with the experimental data.
Figure S7. Normalized PL spectra on a logarithmic scale from 8.7 to 80 K for electron densities (a) and (b) . Colored lines: experimental spectra after subtraction of a linear background and normalization to the maximum of the AP peak. Black lines: calculated spectra for trion binding energies (dashed), 26 (solid) and 28 meV (dotted), with a broadening of meV. The calculated spectra are shifted in energy at each temperature so that their AP peak coincides with the experimental one. The logarithmic scale reveals the weak RP branch at higher energy, whose position and intensity depend on ; for K the calculated RP intensity is below and therefore not visible.
References
- [1]Daniel J. Blumenthal, Rene Heideman, Douwe Gezebroek, Arne Leinse, and Chris Roelofsz en. Silicon Nitride in Silicon Photonics. Proceedings of the IEEE, 106(12):2209–2231, December 2018. ISSN 0018-9219, 1558-2256. doi: 10.1109/JPROC.2018.2861576. URL https://ieeexplore.ieee.org/document/8472140/.DOI
- [2]Toby Bi, Shuangyou Zhang, Egemen Bostan, Danxian Liu, Aditya Paul, Olga Ohletz, Irina Harder, Yaojing Zhang, Alekhya Ghosh, Abdullah Alabbadi, Masoud Kheyri, Tianyi Zeng, Jesse Lu, Kiyoul Yang, and Pascal Del’Haye. Inverse-designed silicon nitride nanophotonics. Nature Communications, 17(1):6943, 2026. ISSN 2041-1723. doi: 10.1038/s41467-026-73390-9. URL https://www.nature.com/articles/s41467-026-73390-9.DOI
- [3]Yue-De Yang, Yu Li, Yong-Zhen Huang, and Andrew W. Poon. Silicon nitride three-mode division multiplexing and wavelength-division multiplexing using asymmetrical directional couplers and microring resonators. Optics Express, 22(18):22172–22183, September 2014. ISSN 1094-4087. doi: 10.1364/OE.22.022172. URL https://opg.optica.org/oe/abstract.cfm?uri=oe-22-18-22172.DOI
- [4]Shuqi Xiao, Junlei Qi, Yi Wang, Gaolei Hu, Yue Qin, Zhenzhou Cheng, Zunyue Zhang, Qiyuan He, and Hon Ki Tsang. A high-speed infrared tellurium photodetector on a silicon nitride platform. Nanoscale, 16(47):22423–22430, 2024. ISSN 2040-3364, 2040-3372. doi: 10.1039/D4NR03900A. URL https://pubs.rsc.org/en/content/articlelanding/2024/nr/d4nr03900a.DOI
- [5]Ipshita Datta, Sang Hoon Chae, Gaurang R. Bhatt, Mohammad Amin Taday on, Baichang Li, Yiling Yu, Chibeom Park, Jiwoong Park, Linyou Cao, D. N. Basov, James Hone, and Michal Lipson. Low-loss composite photonic platform based on 2D semiconductor monolayers. Nature Photonics, 14(4):256–262, April 2020. ISSN 1749-4885, 1749-4893. doi: 10.1038/s41566-020-0590-4. URL https://www.nature.com/articles/s41566-020-0590-4.DOI
- [6]Swati Joshi and Brajesh Kumar Kaushik. Transition metal dichalcogenides integrated waveguide modulator and attenuator in silicon nitride platform. Nanotechnology, 31(43):435202, October 2020. ISSN 0957-4484, 1361-6528. doi: 10.1088/1361-6528/aba579. URL https://iopscience.iop.org/article/10.1088/1361-6528/aba579.DOI
- [7]Kamyar Parto, Shaimaa I. Azzam, Nicholas Lewis, Sri Divya Patel, Seiya Umezawa, Kenji Watanabe, Takashi Taniguchi, and Galan Moody. Cavity-Enhanced 2D Material Quantum Emitters Deterministically Integrated with Silicon Nitride Microresonators. Nano Letters, 22(24):9478–9756, December 2022. ISSN 1530-6984, 1530-6992. doi: 10.1021/acs.nanolett.2c03151. URL https://pubs.acs.org/doi/10.1021/acs.nanolett.2c03151.DOI
- [8]Hannah Rarick, Abhinav Kala, Sinabu Pumulo, Arnab Manna, David Sharp, Christopher Munley, Xiaodong Xu, and Arka Majumdar. Enhanced Second-Harmonic Generation in a Monolayer Tungsten Diselenide Integrated Silicon Nitride Nanocavity. ACS Photonics, 11(11):4635–4641, November 2024. ISSN 2330-4022, 2330-4022. doi: 10.1021/acsphotonics.4c01029. URL https://pubs.acs.org/doi/10.1021/acsphotonics.4c01029.DOI
- [9]Kishor Kumar Mandal, Yashika Gupta, Brijesh Kumar, Mandar Sohoni, Achanta Venu Gopal, and Anshuman Kumar. A photonic integrated chip platform for interlayer exciton valley routing. Journal of Applied Physics, 133(12):123104, March 2023. ISSN 0021-8979, 1089-7550. doi: 10.1063/5.0139880. URL https://pubs.aip.org/jap/article/133/12/123104/2881181/A-photonic-integrated-chip-platform-for-interlayer.DOI
- [10]Xingyu Huang, Hanlin Fang, Shima Kadkhodazadeh, Monia Runge Nielsen, Qiaoling Lin, Zhipei Sun, Martijn Wubs, and Sanshui Xiao. On-chip integrated light sources with MoS₂/WSe₂ moiré superlattices at telecom wavelengths, 2025. URL https://arxiv.org/abs/2504.03267.
- [11]Shun Cao, Hongguang Dong, Jinlong He, Erik Forsberg, Yi Jin, and Sailing He. Normal-Incidence-Excited Strong Coupling between Excitons and Symmetry-Protected Quasi-Bound States in the Continuum in Silicon Nitride–WS₂ Heterostructures at Room Temperature. The Journal of Physical Chemistry Letters, 11(11):4631–4638, June 2020. ISSN 1948-7185. doi: 10.1021/acs.jpclett.0c01080. URL https://pubs.acs.org/doi/10.1021/acs.jpclett.0c01080.DOI
- [12]Chih-Zong Deng, Sunhao Shi, Chun-Hao Chiang, Mu-Hsin Chen, Shuaicheng Liu, Haruyuki Sakurai, Jui-Han Fu, Kuniaki Konishi, Masanobu Iwanaga, Vincent Tung, and Ya-Lun Ho. Freestanding Polymer Metasurface Supporting Higher-Order Optical Resonances for Strong Field Enhancement in TMD Monolayers. Small, 22(32):e13320, 2026. ISSN 1613-6810, 1613-6829. doi: 10.1002/smll.202513320. URL https://onlinelibrary.wiley.com/doi/10.1002/smll.202513320.DOI
- [13]Henna Farheen, Yuheng Chen, Peigang Chen, Pranshu Maan, Samuel Peana, Alexander Senichev, Vladimir M. Shalaev, Alexandra Boltasseva, Jens Forstner, and Alexander V. Kildishev. Efficient Silicon Nitride Quantum Interconnect for Intrinsic Silicon Nitride Single-Photon Emitters. IEEE Journal of Selected Topics in Quantum Electronics, pages 1–12, 2026. ISSN 1077-260X, 1558-4542. doi: 10.1109/JSTQE.2026.3722063. URL https://ieeexplore.ieee.org/document/11651461/.DOI
- [14]Amy S. McKeown-Green, Helen J. Zeng, Ashley P. Saunders, Jiayi Li, Jiaojian Shi, Yuejun Shen, Feng Pan, Jenny Hu, Jennifer A. Dionne, Tony F. Heinz, Stephen M. Wu, Fan Zheng, and Fang Liu. Millimeter-Scale Exfoliation of hBN with Tunable Flake Thickness for Scalable Encapsulation. ACS Applied Nano Materials, 7(6):6574–6582, March 2024. doi: 10.1021/acsanm.4c00412. URL https://pubs.acs.org/doi/10.1021/acsanm.4c00412.
- [15]Jaewon Wang and Soon-Yong Kwon. Wafer-Scale Single-Crystal Boron Nitride: Synthesis and Integration in 2D Electronics. Advanced Materials Interfaces, 13(9):e01079, May 2026. ISSN 2196-7350, 2196-7350. doi: 10.1002/admi.202501079. URL https://advanced.onlinelibrary.wiley.com/doi/10.1002/admi.202501079.DOI
- [16]Kin Fai Mak, Changgu Lee, James Hone, Jie Shan, and Tony F. Heinz. Atomically Thin MoS₂: A New Direct-Gap Semiconductor. Physical Review Letters, 105(13):136805, September 2010. ISSN 0031-9007, 1079-7114. doi: 10.1103/PhysRevLett.105.136805. URL https://link.aps.org/doi/10.1103/PhysRevLett.105.136805.DOI
- [17]Gang Wang, Alexey Chernikov, Mikhail M. Glazov, Tony F. Heinz, Xavier Marie, Thierry Amand, and Bernhard Urbaszek. Colloquium: Excitons in atomically thin transition metal dichalcogenides. Reviews of Modern Physics, 90(2):021001, April 2018. ISSN 0034-6861, 1539-0756. doi: 10.1103/RevModPhys.90.021001. URL https://link.aps.org/doi/10.1103/RevModPhys.90.021001.DOI
- [18]Alexey Chernikov, Timothy C. Berkelbach, Heather M. Hill, Albert Rigosi, Yilei Li, Ozgur Burak Aslan, David R. Reichman, Mark S. Hybertsen, and Tony F. Heinz. Exciton Binding Energy and Nonhydrogenic Rydberg Series in Monolayer WS₂. Physical Review Letters, 113(7):076802, August 2014. ISSN 0031-9007, 1079-7114. doi: 10.1103/PhysRevLett.113.076802. URL https://link.aps.org/doi/10.1103/PhysRevLett.113.076802.DOI
- [19]Daniel Rhodes, Sang Hoon Chae, Rebeca Ribeiro-Palau, and James Hone. Disorder in van der Waals heterostructures of 2D materials. Nature Materials, 18(6):541–549, June 2019. ISSN 1476-1122, 1476-4660. doi: 10.1038/s41563-019-0366-8. URL https://www.nature.com/articles/s41563-019-0366-8.DOI
- [20]C. R. Dean, A. F. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K. Watanabe, T. Taniguchi, P. Kim, K. L. Shepard, and J. Hone. Boron nitride substrates for high-quality graphene electronics. Nature Nanotechnology, 5(10):722–726, October 2010. ISSN 1748-3387, 1748-3395. doi: 10.1038/nnano.2010.172. URL https://www.nature.com/articles/nnano.2010.172.DOI
- [21]F. Cadiz, E. Courtade, C. Robert, G. Wang, Y. Shen, H. Cai, T. Taniguchi, K. Watanabe, H. Carrere, D. Lagarde, M. Manca, T. Amand, P. Renucci, S. Tongay, X. Marie, and B. Urbaszek. Excitonic Linewidth Approaching the Homogeneous Limit in MoS₂-Based van der Waals Heterostructures. Physical Review X, 7(2):021026, May 2017. ISSN 2160-3308. doi: 10.1103/PhysRevX.7.021026. URL http://link.aps.org/doi/10.1103/PhysRevX.7.021026.DOI
- [22]Obafunso A Ajayi, Jenny V Ardelean, Gabriella D Shepard, Jue Wang, Abhinandan Antony, Takeshi Taniguchi, Kenji Watanabe, Tony F Heinz, Stefan Strauf, X-Y Zhu, and James C Hone. Approaching the intrinsic photoluminescence linewidth in transition metal dichalcogenide monolayers. 2D Materials, 4(3):031011, September 2017. ISSN 2053-1583. doi: 10.1088/2053-1583/aa6aa1. URL https://iopscience.iop.org/article/10.1088/2053-1583/aa6aa1.DOI
- [23]Fuhui Shao, Steffi Y. Woo, Nianjheng Wu, Robert Schneider, Andrew J. Mayne, Steffen Michaelis De Vasconcellos, Ashish Arora, Benjamin J. Carey, Johann A. Preuß, Noémie Bonnet, Mauro Och, Cecilia Mattevi, Kenji Watanabe, Takashi Taniguchi, Zhichuan Niu, Rudolf Bratschitsch, and Luiz H. G. Tizei. Substrate influence on transition metal dichalcogenide monolayer exciton absorption linewidth broadening. Physical Review Materials, 6(7):074005, July 2022. ISSN 2475-9953. doi: 10.1103/PhysRevMaterials.6.074005. URL https://link.aps.org/doi/10.1103/PhysRevMaterials.6.074005.DOI
- [24]Kevin Chen, Daisuke Kiriya, Mark Hettick, Mahmut Tosun, Tae-Jun Ha, Surabhi Rao Madhvapathy, Sujay Desai, Angada Sachid, and Ali Javey. Air stable n-doping of WSe₂ by silicon nitride thin films with tunable fixed charge density. APL Materials, 2(9):092504, September 2014. ISSN 2166-532X. doi: 10.1063/1.4891824. URL https://pubs.aip.org/apm/article/2/9/092504/150620/Air-stable-n-doping-of-WSe2-by-silicon-nitride.DOI
- [25]Leyi Loh, Han Yan, Lixin Liu, Yuan Chen, Esther Lilian Gray, Kamal Kumar Paul, Deepnarayan Biswas, Tien-Lin Lee, Takashi Taniguchi, Kenji Watanabe, Mauro Brotons-Gisbert, Yan Wang, and Manish Chhowalla. Dielectric-induced charge doping in monolayer mos₂. Nano Letters, 26(34):11583–11590, August 2026. ISSN 1530-6992. doi: 10.1021/acs.nanolett.6c03443. URL http://dx.doi.org/10.1021/acs.nanolett.6c03443.
- [26]Hangyong Shan, Max Waldherr, Diksha, Ghada Missaoui, Seyma Esra Atalay, Martin Zinner, Ana M. Valencia, Kenji Watanabe, Takashi Taniguchi, Seth Ariel Tongay, Caterina Cocchi, Brendan C. M<ulkerin, Jesper Levinsen, Francesca Maria Marchetti, Meera M. Parish, Niklas Nilius, Christian Schneider, and Sven Höfling. Optical properties of Fermi polarons in a GaInP/MoSe₂ monolayer heterostructure. 2D Materials, 13(2):025016, March 2026. ISSN 2053-1583. doi: 10.1088/2053-1583/ae4761. URL https://iopscience.iop.org/article/10.1088/2053-1583/ae4761.DOI
- [27]Tingting Wu, Francesca Maria Marchetti, Antonio Tiene, Antonio I. Fernández-Domínguez, Miao Qi, Zhe Wang, Lin Liu, Lei Wei, Francisco J. Garcia Vidal, Mengxiao Chen, Qi Jie Wang, and Yu Luo. Room-temperature tuning and probing of fermi polarons in atomically thin semiconductors on a plasmonic metasurface. ACS Nano, 20:18395–18405, 2026. doi: 10.1021/acsnano.6c04930.DOI
- [28]Antonio Tiene, Brendan C. M<ulkerin, Jesper Levinsen, Meera M. Parish, and Francesca Maria Marchetti. Crossover from exciton polarons to trions in doped two-dimensional semiconductors at finite temperature. Physical Review B, 108:125406, 2023. doi: 10.1103/PhysRevB.108.125406. URL https://link.aps.org/doi/10.1103/PhysRevB.108.125406.DOI
- [29]Brendan C. M<ulkerin, Antonio Tiene, Francesca Maria Marchetti, Meera M. Parish, and Jesper Levinsen. Exact quantum virial expansion for the optical response of doped two-dimensional semiconductors. Physical Review Letters, 131:106901, 2023. doi: 10.1103/PhysRevLett.131.106901. URL https://link.aps.org/doi/10.1103/PhysRevLett.131.106901.DOI
- [30]Meinrad Sidler, Patrick Back, Ovidiu Cotlet, Ajit Srivastava, Thomas Fink, Martin Kroner, Eugene Demler, and Atac Imamoglu. Fermi polaron-polaritons in charge-tunable atomically thin semiconductors. Nature Physics, 13(3):255–261, March 2017. ISSN 1745-2473, 1745-2481. doi: 10.1038/nphys3949. URL https://www.nature.com/articles/nphys3949.DOI
- [31]Dmitry K. Efimkin and Allan H. MacDonald. Many-body theory of trion absorption features in two-dimensional semiconductors. Physical Review B, 95(3):035417, January 2017. ISSN 2469-9950, 2469-9969. doi: 10.1103/PhysRevB.95.035417. URL https://link.aps.org/doi/10.1103/PhysRevB.95.035417.DOI
- [32]Jason S. Ross, Sanfeng Wu, Hongyi Yu, Nirmal J. Ghimire, Aaron M. Jones, Grant Aivazian, Jiaqiang Yan, David G. Mandrus, Di Xiao, Wang Yao, and Xiaodong Xu. Electrical control of neutral and charged excitons in a monolayer semiconductor. Nature Communications, 4(1):1474, February 2013. doi: 10.1038/ncomms2498. URL https://www.nature.com/articles/ncomms2498. Number: 1.DOI
- [33]Kin Fai Mak, Keliang He, Changgu Lee, Gwan Hyoung Lee, James Hone, Tony F. Heinz, and Jie Shan. Tightly bound trions in monolayer MoS₂. Nature Materials, 12(3):207–211, March 2013. ISSN 1476-1122, 1476-4660. doi: 10.1038/nmat3505. URL https://www.nature.com/articles/nmat3505.DOI
- [34]L. Viña, S. Logothetidis, and M. Cardona. Temperature dependence of the dielectric function of germanium. Physical Review B, 30(4):1979–1991, 1984. doi: 10.1103/PhysRevB.30.1979.DOI
- [35]E. H. Kennard. On the thermodynamics of fluorescence. Physical Review, 11(1):29–38, 1918. doi: 10.1103/PhysRev.11.29.DOI
- [36]Ted Pella, Inc. PELCO silicon nitride support films for TEM: technical specifications. https://www.tedpella.com/grids_html/silicon-nitride.aspx, 2026. Accessed October 2026.
- [37]Kristine L. Haley, Joshua A. Cloninger, Kaitlyn Cerminara, Randy M. Sterbentz, Takashi Taniguchi, Kenji Watanabe, and Joshua O. Island. Heated assembly and transfer of Van der Waals heterostructures with common nail polish. Nanomanufacturing, 1(1):49–56, 2021. doi: 10.3390/nanomanufacturing1010005.DOI
- [38]T. Taniguchi and K. Watanabe. Synthesis of high-purity boron nitride single crystals under high pressure by using Ba–BN solvent. Journal of Crystal Growth, 303(2):525–529, 2007. doi: 10.1016/j.jcrysgro.2006.12.061.DOI
- [39]Andres Castellanos-Gomez, Michele Busema, Rianda Molenaar, Vibhor Singh, Laurens Janssen, Herre S J van der Zant, and Gary A Steele. Deterministic transfer of two-dimensional materials by all-dry viscoelastic stamping. 2D Materials, 1(1):011002, April 2014. ISSN 2053-1583. doi: 10.1088/2053-1583/1/1/011002. URL http://dx.doi.org/10.1088/2053-1583/1/1/011002.
- [40]Woei Ming Lee, Peter J Reece, Robert F Marchington, Nikolaus K Metzger, and Kishan Dholakia. Construction and calibration of an optical trap on a fluorescence optical microscope. Nature Protocols, 2(12):3226–3238, December 2007. ISSN 1750-2799. doi: 10.1038/nprot.2007.446. URL http://dx.doi.org/10.1038/nprot.2007.446.
- [41]M. Goryca, J. Li, A. V. Stier, T. Taniguchi, K. Watanabe, E. Courtade, S. Shree, C. Robert, B. Urbaszek, X. Marie, and S. A. Crooker. Revealing exciton masses and dielectric properties of monolayer semiconductors with high magnetic fields. Nature Communications, 10:4172, 2019. doi: 10.1038/s41467-019-12180-y.DOI
- [42]T. Anh Pham, Tianshu Li, Sadasivan Shankar, Francois Gygi, and Giulia Galli. Microscopic modeling of the dielectric properties of silicon nitride. Physical Review B, 84(4):045308, July 2011. ISSN 1098-0121, 1550-235X. doi: 10.1103/PhysRevB.84.045308. URL https://link.aps.org/doi/10.1103/PhysRevB.84.045308.DOI
- [43]H. R. Philipp. Optical properties of silicon nitride. Journal of the Electrochemical Society, 120(2):295–300, 1973. doi: 10.1149/1.2403440.DOI
- [44]Kevin Luke, Yoshitomo Okawachi, Michael R. E. Lamont, Alexander L. Gaeta, and Michal Lipson. Broadband mid-infrared frequency comb generation in a Si₃N₄ microresonator. Optics Letters, 40(21):4823–4826, 2015. doi: 10.1364/OL.40.004823.DOI
- [45]Leonid Yu. Beliaev, Evgeniy Shkondin, Andrei V. Lavrinenko, and Osamu Takayama. Optical, structural and composition properties of silicon nitride films deposited by reactive radio-frequency sputtering, low pressure and plasma-enhanced chemical vapor deposition. Thin Solid Films, 763:139568, 2022. doi: 10.1016/j.tsf.2022.139568.DOI







