Collimated unidirectional laser beams from notched elliptical resonators.

T he whispering gallery is located in the dome of St Paul’s Cathedral in London. If two people stand on opposite sides of the gallery and one of them whispers into the wall, the other can hear his or her words. The reason for this strange effect is that sound bounces along the wall as a surface wave with low loss, and so can be heard far away. This phenomenon, known as whispering gallery mode (WGM), was discovered by Lord Rayleigh in the 18th century. It has found new notoriety in the past few decades in the optics domain: ultra-high Q-factor, i.e., ultralow loss, optical resonators, based on WGMs have been demonstrated, opening up exciting new research frontiers in low-power nonlinear optics, optomechanics, optical processing, and sensing (1, 2). In PNAS, Wang et al. manage to demonstrate what eluded researchers for many years: a semiconductor microlaser emitting a unidirectional low-divergence beam based on optical WGMs (3). To understand the significance of this achievement, consider that in the optical version of WGMs, light bounces around the edge of a glass sphere or a transparent disk if the angle of incidence at the glass–air interface exceeds the so-called critical angle (Fig. 1A). When this condition is satisfied, almost 100% of the light is reflected (100% for planar interface), whence the name total internal reflection comes—a very useful effect to reduce losses. Thus, when light is traveling around the edge it will be “totally” reflected at each bounce and propagate with little loss. Because the light will make a large number of roundtrips (typically well in an excess of 1 million) before being absorbed, it will undergo interference with itself. Therefore, only an integer number of wavelengths of light can “fit” around the edge. These WGMs are of the lowest loss ever measured in the optical domain (1, 2). What attracted early researchers to opticalWGMswas the possibility ofmaking microdisk diode lasers with ultra-low threshold current by virtue of their small size and very low optical losses of WGMs (4). A serious drawback, however, is that in circularly symmetric cavities WGMs can be only weakly and isotropically coupled out through evanescent wave scattering by surface roughness and diffraction. Therefore, it quickly became the “holy grail” of microlaser research to combine the highly directional emission of most lasers with the high Q-factor and the resulting low lasing threshold of WGMs. In the PNAS paper by Wang et al. (3), the authors brilliantly solve this problem of seemingly conflicting requirements by designing a new microlaser resonator (Fig. 1B) that utilizes the advantages of WGM lasers while achieving unidirectional laser action with very low divergence and a concomitant substantial increase in optical power output. Previous deformation strategies to increase performance of WGM lasers had resulted in the emission of multiple optical beams or lower directionality (5–8). The authors measured a record low divergence of approximately 5° in the plane of the resonator in midinfrared (10-μm wavelength) quantum cascade lasers, a major improvement compared with the state of the art. The relative insensitivity of the device divergence to different notch sizes and drive currents demonstrates the robustness of this type of resonator. The aspect ratio ε of the resonator was used as a design parameter to minimize the beam divergence. A peak optical power of 5 mW was measured. Substantial room for optimization exists: for example, to increase the output power of the device, one can choose to moderately increase ε or change the notch size to scatter more power out of the cavity. The Q-factor of quantum cascade lasers, however, is much less than the simulated ones when optical losses caused by absorption by free electrons are included, which reduces the output power and prevents a major laser threshold reduction. Simulations by the authors, however, show that their resonator design should be much more effective at shorter wavelengths such as the ones used in telecommunications near a λ of 1 μm. The high directionality was shown to be preserved for the transverse electric modes of these shorter wavelength resonators, with the advantage that now the contribution of free carrier absorption to optical losses is negligible and the latter are expected to be very small (∼0.5 cm), limited by sidewall roughness and residual absorption. The other exciting achievement of the paper by Wang et al. (3) is that it introduces an important resonator for the controlled study of chaotic ray dynamics (5–7). By breaking circular symmetry, it introduces some amount of chaos in the corresponding ray dynamics. Indeed, many works in the past two decades have explored this regimen known as “wave chaos” or “quantum chaos” in optical resonators of increasing complexity, leading to the discovery of a rich phenomenology (5–8). The elliptical notched resonator design will allow the study of diffractive effects on wave chaos in a controlled way by varying the notch size and thus the size of the obstacle off which light is bounced. Fig. 1. (A) Schematic of a whispering gallerymode circulating around the border of a circular optical resonator, trapped by total internal reflection. (B) The new resonator design exploits the well-known property of an elliptical disc that a beam of incident light rays parallel to the long axis is refracted into one of its foci. The authors used a construction based on an auxiliary ellipse (black dashed curve) to design an elliptical resonator with a wavelengthsize notch (O) at the focus of the auxiliary ellipse and with eccentricity optimized to best match the boundary of the latter, within the largest possible range of angles. In this way, a large fraction of the light of circulating whispering gallery modes scattered by the notch is collected and collimated into a very low divergence beam, thus overcoming the problem of lack of directionality and negligible output power of microlasers using rotationally symmetric resonator designs.