Metasurface lasers programmed by optical pump patterns

TL;DR

This work demonstrates a reconfigurable plasmonic metasurface laser platform in which spatially structured optical pumping defines gain boundaries to program lasing geometry in hexagonal nanoparticle lattices. It reveals robust $K$-point lasing with spontaneous symmetry breaking between $K$ and $K'$ modes that is insensitive to pump geometry, alongside on-demand control of $M$-point lasing via asymmetric pumping; coupled gain regions further show phase and amplitude synchronization through Bloch-mode mediated coupling. A density-matrix–based stochastic model reproduces the observed real-space and Fourier-space patterns and the SSB statistics, validating the mechanism and enabling exploration of non-Hermitian and topological photonics in programmable lattices. The results point to applications in vortex-beam shaping, on-chip optical logic, neuromorphic computing, and true random-number generation, highlighting the practical impact of spatially programmable metasurface lasers.

Abstract

Metasurface lasers offer unprecedented control over light emission, yet their spatial and modal characteristics are typically fixed post-fabrication. Here, we introduce a reconfigurable plasmonic metasurface laser platform in which the lasing area geometry, and thus the emission properties, are dynamically programmed via spatially structured optical pumping. Using hexagonal arrays of silver nanoparticles embedded in dye-doped waveguides, we demonstrate lasing at high-symmetry points of the Brillouin zone, including the K and M points. K-point lasing exhibits spontaneous symmetry breaking (SSB) in relative intensity between degenerate K and K' modes, with no bias induced by pump geometry, even for geometries that explicitly break symmetry. In contrast, M-point lasing allows deterministic control over emission channels via asymmetric pumping. We further show that spatially separated K-point lasers synchronize in phase and amplitude, undergoing SSB in lockstep. A theoretical density matrix approach cast into stochastic differential equations reproduces the observed real- and Fourier-space intensity distributions and SSB behavior. Our findings establish spatially programmable metasurface lasers as a versatile platform for exploring dynamic phenomena in photonic lattices, with potential applications in vortex beam shaping, optical logic, and true random number generation.

Figures (8)

• Figure 1: Single-shot microscopy with structured pump illumination. (a) Illustration of the plasmon lattice laser, which is a hexagonal lattice of silver nanoparticles embedded in a dye-doped slab waveguide. (b) Fluorescence microscope setup used to study plasmon lattice lasers. We place a digital micromirror device (DMD) in the pump-illumination track in the focus of the epi-lens, which enables us to project any pattern in the real-space plane of the sample. The emission signal is sent to the imaging setup where we record real- and Fourier-space images simultaneously. (c) 54 $\upmu$m sized triangle projected by the DMD revealed by fluorescence image recorded in real space. (d, e) Fourier images of fluorescence from the hexagonal plasmon lattices with pitches a = 375 nm and 500 nm, with bandpass filter centered at 580 nm wavelength, at which the repeated waveguide mode crossings occur at the $\mathit{M}$- and $\mathit{K}$-points, respectively. The first Brillouin zone is shown in black in panel (c), with high-symmetry points $\mathit{M}$, $\Gamma$, and $\mathit{K}$ indicated.
• Figure 2: $\mathit{K}$-point lasing with boundary conditions in gain, averaged over many lasing shots. (a-c) Fourier- and real-space images of $\mathit{K}$-point lasing ($a=500$ nm) from DMD-projected hexagons of short diagonal $L=\{46,34.5,23\}$$\upmu$m (radius $\{26.6,19.9,13.3\}$$\upmu$m). (d-f) Lasing from triangles with side length $L=\{57.5,46,34.5\}$$\upmu$m. (g-i) Lasing from rectangles of shape $\{(28.9\times50), (23.1\times40),(17.3\times30)\}$$\upmu$m and (j-l) the same rectangles rotated by 90$\degree$. In real space, lasing is characterized by finely spaced speckles, overlayed with intensity envelopes that each have a minimum at the center. In Fourier space, six $\mathit{K}$-point lasing spots are donut-shaped (see insets; the extent of each inset is $\Delta k/k_0=0.06$) but relate to the real-space laser shapes and sizes by Fourier transform, increasing in their Fourier-space extent as the real-space area is reduced. (m) Fourier-space lasing spot size versus the inverse of real-space length $L$. This data is obtained by measuring the Fourier-space extent of the spots, averaging over six spots and taking standard deviation for error estimation. In blue, we display the results for radial and azimuthal crosscuts through donut-spots for the hexagon pump (panels a-c), whereas red points show the results for horizontal and vertical crosscuts through donut-spots for the rectangle pump (panels g-i). Both datasets are plotted against 1/$L$ with $L$ being either the hexagon's short diagonal or the relevant rectangle's side length. The linear relation $\Delta k \propto 1/L$ clearly holds. Caption continues on the next page.
• Figure 2: Continuation of caption from the previous page. (n,o) To quantify the propagation length of laser light beyond the pumped area, we take horizontal and linear crosscuts (at $y_c,x_c=(35,38.7)$$\upmu$m and summing 20 pixels wide) through the real-space hexagon shape in panel (c), for pump fluences $\{1.5,2.1,2.7 \}$ mJ/cm$^2$. The lowest fluence (dark purple curve) belongs to fluorescence, the higher two (green/red curves) to lasing. We fit the line $y = a\exp{b(x-x_1)}$ (cyan curves) to the region beyond the pumped area and obtain $b = -1.2$$\upmu$m$^{-1}$ for fluorescence and $b = -0.4$$\upmu$m$^{-1}$ for lasing. The laser emission thus propagates about 2 to 3 $\upmu$m beyond the gain boundary. (p,q) Analogous crosscut results for the rectangular pump (panel (i)), with fluences $\{1.8,2.3,2.8 \}$ mJ/cm$^2$ -- fitting leads to similar $b$-values as for the hexagons.
• Figure 3: Single-shot $\mathit{K}$-point lasing and SSB. Panels (a,c,e,g,i,k): Simultaneously recorded Fourier-space (top) real-space (bottom) single shots, in which SSB leads to dominant $\mathit{K}$-mode lasing, for a variety of lasing area shapes. For each shape, in addition to displaying the intensity dip in the center, the real-space envelope forms three high-intensity lobes at the edges of the lasing area, forming an equilateral triangle. The orientation of that triangular envelope pattern is mirrored with respect to the $\mathit{K}$ direction of the lasing mode in Fourier space. At the same time, the real-space envelope is directly imprinted on the donut-shaped emission spots. Panels (b,d,f,h,j,l): Single shots for the same pump patterns but with SSB giving rise to dominant $\mathit{K'}$-mode lasing. Pump patterns: (a-f) Triangle, side length 53 $\upmu$m. (g,h) Hexagon, short diagonal 34.5 $\upmu$m. (i-l) Rectangle, shape $(17.3\times30)$$\upmu$m. (m) Single-shot statistics of $\mathit{K}/\mathit{K'}$ SSB parameter $\theta$ that traces relative amplitude, as a function of pump fluence for the upward triangle (panels a,b), and (n) for the downward triangle (panels c,d). Note the absence of bias towards $\mathit{K}/\mathit{K'}$ lasing, despite breaking their symmetry by the gain distribution.
• Figure 4: DMD-controlled $\mathit{M}$-point lasing. (a) 20-shot average Fourier- and real-space images of lasing at the $\mathit{M}$ points under symmetric pumping (LCP-polarized pump forming hexagonal shape in real space with short diagonal 69 $\upmu$m). SSB between the three frequency-degenerate $\mathit{M}$ modes leads to intensity fluctuations of $15\%$. (b-g) Active switching between $\mathit{M}$ modes by rotating a rectangular pump shape $(50\times28.9)$$\upmu$m. The bias is achieved by overlapping the gain pattern with the spatially distinct mode profiles belonging to one (c,e,g) or two (b,d,f) of the three $\mathit{M}$ modes. The insets magnifying the lasing spots in the Fourier images have extents $k_{||}/k_0=0.1$.