Localized emission in MoSe$_2$ monolayers on GaN nanopillars

TL;DR

This study probes whether strain alone or dielectric context governs localized emitters in $MoSe_2$ on GaN nanopillars. By combining hyperspectral micro-PL with co-registered AFM-derived bending-strain maps and AFM phase contrast, the authors map emitter positions relative to nanoscale curvature and dielectric interfaces. They find that most localized states cluster near suspended–supported interfaces around pillar apices and that LS populate a broad strain range with no clear threshold, implying a cooperative strain–dielectric mechanism. The work establishes a framework for structure–property mapping in 2D quantum materials and advocates co-design strategies—via pillar geometry, apex roughness, and spacer layers—for deterministic sub‑λ emitter arrays in quantum photonics.

Abstract

Solid-state quantum emitters (QEs) in two-dimensional semiconductors offer compact, chip-compatible sources for quantum photonics. In transition-metal dichalcogenides (TMDs), nanopillars are widely used to induce localized emission, yet the underlying confinement mechanism and the relative roles of strain versus dielectric environment remain unclear. The general problem addressed here is whether strain alone explains quantum emitter formation and placement in MoSe$_2$, or whether dielectric contrast at suspended-supported interfaces is also required. Here, we combine hyperspectral superlocalization of photoluminescence with co-registered AFM topography and phase to map the positions of localized states (LS) in MoSe$_2$ suspended on GaN pillars and correlate them with bending strain and the local dielectric context. Contrary to the common assumption of purely strain-driven activation, LS frequently occur at suspended--supported interfaces around the pillar apex and span a broad strain range without a clear threshold, while being scarce along high-strain ripples. Our data indicate that deterministic emitter positioning in Mo-based TMDs benefits from co-engineering both strain gradients and nanoscale dielectric heterogeneity, rather than strain alone. More broadly, this combined optical-mechanical characterization approach provides a general framework for mapping structure-property relationships in 2D quantum materials at the single-emitter level.

Figures (8)

• Figure 1: Localized quantum emitters in monolayer MoSe$_2$ on AlN/GaN nanopillars.a: Schematic of the sample and $\mu$-PL setup. The AlN/GaN substrate is patterned to form nanopillars, onto which monolayer MoSe$_2$ is exfoliated. Micro-photoluminescence is performed at 2.8 K. b–d: Optical micrograph (b) and atomic force micrographs (c, d) of the fabricated device. e: Photoluminescence spectra recorded with the laser focused off (green) and on (orange) a nanopillar, with positions indicated in c. The spectra show neutral (X$_0$) and charged (X$^-$) excitons, along with a dense set of sub-meV localized states (LS) observed on the pillar. f: Hyperspectral PL maps showing the spatial distribution of LS (top) and the X$_0$ intensity (bottom). LS appear exclusively at the pillar positions, highlighted by dashed circles. g: Time-resolved PL spectrum of a single LS at 1566 meV, exhibiting spectral jumps of $\sim$500 $\mu$eV. h: Power dependence of the PL intensity of a single LS, showing saturation behavior. i: polarization-resolved PL revealing the linear polarization of LS emission (red), in contrast to the X$^-$ state (grey).
• Figure 2: Bending-induced strain in monolayer MoSe$_2$ on a patterned GaN surface.a: Schematic of a monolayer MoSe$_2$ flake (thickness $t$) transferred onto a GaN nanopillar. Depending on the balance between local adhesion energy and bending energy, MoSe$_2$ can conformally adhere to the substrate. Around the nanopillar, the monolayer becomes partially suspended, forming a tent-like shape. The local radius of curvature is denoted $R(x,y)$. b: Linecut of the MoSe$_2$ topography across a nanopillar (magenta), with position indicated in c. For comparison, a reference profile measured on a bare GaN nanopillar (grey, uncovered region) is overlaid without vertical shift. c: Topography micrograph of the sample. The logarithmic scale enhances the visibility of transfer-induced features such as ripples, bubbles, and suspended regions around the pillar at the center. d: (Top) Extracted bending-induced strain $\epsilon$ across a ripple (see inset). (Bottom) Corresponding topography profile. The strain exhibits local minima (marked with *) corresponding to inflection points where the curvature changes sign—from convex (bending upward) to concave (bending downward). e: 2D map of the bending-induced strain in MoSe$_2$. The grey region corresponds to an area without MoSe$_2$, where strain extraction is not applicable.
• Figure 3: Super-localization of emitters in MoSe$_2$.a: Hyperspectral photoluminescence (PL) map integrated over the full spectral range; one spectrum is acquired at each scan position. b: Atomic force microscopy (AFM) phase micrograph of the same region. c: PL excitation map with spectral filtering of the emission according to the color code in (d); spot centroids are obtained from 2D spatial Gaussian fits. d: Representative PL spectrum near a nanopillar position showing multiple localized states. e: Super-localization map from 100 repeated scans with fitted emission centroids. Projected 1D histograms (top/right) show Gaussian-fitted spreads of $\sigma_x \approx 51~\mathrm{nm}$ and $\sigma_y \approx 86~\mathrm{nm}$ (laser spot: 600 nm). f: Histogram of the frame-to-frame displacement $\Delta r_n = \lVert \mathbf{r}_{n+1}-\mathbf{r}_n\rVert$; the distribution peaks near $17~\mathrm{nm}$, consistent with the single-step increment of the translation stage.
• Figure 4: Impact of strain and dielectric environment on the positions of localized emitters in monolayer MoSe$_2$.a: Bending-induced strain map overlaid with the extracted positions of localized states (LS). Dot color encodes emission energy (yellow–red: 1540–1610 meV). b: Histogram of the bending-induced strain. Colored markers indicate the strain at LS centroids; arrows show the local maximum strain within a 50 nm neighborhood of each centroid ($\epsilon_{\mathrm{max}}^{(50)}$), highlighting that emitters may experience higher strain than at the centroid itself. The distribution of $\epsilon_{\mathrm{max}}^{(50)}$ (grey) shows LS occurring across the full strain range. c: AFM phase ($\theta$) image of the same region with LS positions overlaid. Blue regions correspond to suspended MoSe$_2$ (tent-like geometry from pillar top to substrate), whereas warmer tones indicate supported areas; the dielectric environment varies sharply across this interface. d: Histogram of AFM phase $\theta$. The vertical dashed line marks the suspended–supported boundary. Colored points denote the phase at each LS centroid (same color code as in a); horizontal bars show the phase range within a 50 nm neighborhood. Most bars cross the boundary, indicating that LS frequently lie near suspended–supported interfaces, consistent with dielectric-contrast–assisted localization.
• Figure S1: Temperature-dependent photoluminescence spectra of MoSe$_2$ on top of a pillar. (a) PL spectra showing narrow emission lines red-shifted from the free-exciton energy, associated with localized states. The spectra are shown for temperatures ranging from 2.74 K to 21 K. Above 20 K, no localized state emission is observed. At each temperature, emission maps were acquired to reposition the laser spot at the same location and compensate for thermal drift of the sample holder (although drift is minimal in the 0-30 K range for this specific setup).