Single-photon emitters in 2D materials

Pioneering work during the last decade has propelled color centers in diamond, silicon carbide, and other wide-bandgap semiconductors as reference platforms in the search for optimal chip-integrated solid-state quantum processors; it has also exposed, however, some serious difficulties stemming from limitations of the point defects themselves or the associated bulk crystal serving as the host. Circumventing the limitations of traditional hosts, we are exploring novel semiconductor platforms in the form of thin (30-50 nm), transfer-compatible membranes. Unlike single atomic layers (where emitters are subject to environmental fluctuations difficult to control), these ‘quasi-2D’ hosts provide bulk-like crystalline environments, while simultaneously allowing the experimenter to gain a more direct control of the qubit properties, for example, through the use of metamaterial structures, local strain, or electric fields. The versatility inherent to transfer-ready semiconducting epilayers gives us the opportunity to exploit the most evolved heterostructure geometries to better extract and guide light from the rare-earth emitters, and/or tune their emission properties via the use of electrical potentials or local strain. The latter, in turn, opens intriguing theoretical and experimental possibilities to investigate defect response in regimes difficult or impossible to reach with bulk crystals.

As an introductory example, the figure below shows confocal imaging and optical spectroscopy from single color centers in a thin (~20 nm) flake of CVD-grown hexagonal boron nitride (hBN), a ~6 eV van der Waals semiconductor isomorphic to graphene. Optical spectroscopy reveals narrow zero-phonon lines (ZPL) and comparatively small phonon sidebands, an advantage over point defects such as the NV center in diamond whose fluorescence is largely phonon shifted.

Fig. 1: (a) In a wide-bandgap semiconductor, the relevant energy levels of a point defect serving as a qubit lie within the material bandgap. The defect, therefore, can be looked at as a ‘virtual atom’. (b,c) Example color center in hBN, a ~6eV bandgap van der Waals material. We address single color centers in hBN using confocal microscopy (bright spots in the image, lower right). (d) Optical spectroscopy of individual, strain-shifted color centers in hBN. The upper inset is a typical auto-correlation trace, revealing single photon emitters.

Related work is showcased in the figure below, where we examine the response of emitters in 20-nm-thick films of hBN overlaid on a pillar array. Van der Waals forces reshape the hBN membrane to conform to the SiO2 substrate and, in so doing, introduce intense local strain, whose effect on the emitters in the film is dramatic: In particular, confocal imaging shows fluorescence selectively originates from the pillar sites (or, more precisely, from the strained regions around the pillars). Numerical modeling supports the notion of defect activation via charge trapping in deformation potential wells at these highly strained locations, i.e., emitters become bright because they change their charge state due to strain-induced local doping.

Fig. 2: (a) AFM image of a 20-nm-thick hBN on a Si pillar structure; the film is ripped off in some sections (0L) and folded on itself in others (2L). (b) Confocal image of the structure in (a). (c,d) AFM and confocal images of a 2-µm-wide pillar. (e) AFM cross section of the pillar in (c), with and without the hBN film (black and red traces, respectively). (f) Deformation potential for holes and electrons (blue and orange traces); also included for reference are the deformation and fluorescence cross sections (red and green traces, respectively).

But perhaps a better illustration of the possibilities afforded by the use of thin semiconductor membranes is that summarized in Fig. 3: Here, we combine an Si3N4 microdisk optical resonator and a ~20-nm-thick, color-center-hosting layer of hBN overlaid on the resonator surface. The hBN film folds rip-free around the microdisk and undercut so as to maximize the contact with the substrate material (Fig. 3a). Consistent with the results of Fig. 2, the local strain that develops at the cavity edges deterministically activates a low density of photon emitters within the whispering gallery mode volume of the microdisk. These conditions allow us to demonstrate non-local, cavity-mediated excitation of hBN emitters and incipient coupling of individual emitters and microdisk cavity modes (Figs. 3b and 3c).

Fig. 3: (a) (Top) Lateral view of a microdisk resonator showing ripped hBN wrapped around its contour. (Bottom) Top view of similar disk resonators; the wrapped hBN film takes different geometries (white contours on the lower, left corner). (b) (Left) Confocal image of an array of micro-disk resonators with overlaid hBN flakes. (Right) Zoomed confocal image of the resonator in the lower right corner; emitters activate selectively at the edge of the resonator (white circle). The orange square indicates the location of a scatterer for cavity mode out-coupling. (c) The blue trace is the optical spectrum of the emission collected at the scatterer upon exciting the emitters at the circled spot in the right image in (b). The dashed red trace represents the hBN emission spectrum in the absence of a cavity. We observe emission only at the resonator modes, demonstrating incipient coupling of the emitters to the cavity.