ABSTRACT
Traditional end-fire coupling involves aligning a single-mode fiber to the butt end of a waveguide at the edge of a photonic chip. A taper on the waveguide allows for better optical mode matching between the external fiber and the on-chip waveguide. This reduces losses and back reflections upon coupling.32,33 Although simple to design and implement, a disadvantage of this method
is that alignment can only take place at the edge of the chip. It therefore limits device locations and densities on the sample.Coupling light onto a chip can also be accomplished using optical gratings.34-39 Gratings allow light to be coupled perpendicularly from a fiber onto a chip, and this removes the edge alignment requirement as in end-fire coupling. This leads to the importance of grating couplers in large-scale device integration. The density of addressable coupling locations is no longer dependent on the perimeter length of the chip. The grating consists of alternating regions with high and low indices of refraction. These can be fabricated either by etching the waveguide34-36 or by depositing a periodic array of metal material.37 Light is channeled into the waveguide at nearly a 90° angle due to diffraction, and it can do this with both high efficiency and large bandwidth. Normally, the grating requires an adiabatically tapered waveguide to connect the 10 µm-wide grating to the 500 nm-wide waveguide. The grating must be large due to the size mismatch between the optical modes of a single-mode fiber and an integrated waveguide. Focused gratings can also be designed to remove the need for this taper and decrease the device footprint.38 This permits the dense packing of nanophotonic devices across an entire wafer, which is highly desirable for integrated photonic circuits. Multiaxis fiber positioners are often used to align the two components, as coupling efficiency is dependent on both the position and input angle incident on the grating.36 Grating couplers offer an effective way to couple light into high-density structures, and, consequently, it is one of the most commonly used methods.Looking next at a technique to address single devices, dimpled fiber taper probe coupling40,41 is a simple non-destructive method to locally probe individual devices on a larger optical chip. This method has been adapted from straight42,43 and curved44,45 fiber taper probes to address the issues of parasitic loss and mechanical instability experienced by these methods, respectively. The dimpled fiber is created by simultaneously heating and pulling a telecommunications fiber to first create a straight single-mode tapered fiber. Pulling the fiber adiabatically reduces the fiber diameter until it is small enough to become a single-mode fiber with evanescent tails extending significantly into the area surrounding the waveguide. The dimpled fiber is then created from this straight
fiber by placing it into a mould with the desired fiber radius. The fiber retains the shape of the mold as it is subsequently heated and cooled. In this way, the fiber can be mounted with high tension to achieve good mechanical stability, while still having a smaller probing area to easily probe individual, high-density devices with lower parasitic losses. The dimpled fiber tapered probe can be mounted on a three-axis stage to position it in relation to the desired optical device, or the device itself can be aligned to the tapered fiber.46For the highest sensitivities, nanomechanical resonators must operate in vacuum to eliminate air damping. Therefore, fiber positioning/alignment must occur in vacuum. This complicates the experimental setup which then requires a larger vacuum chamber with vacuum safe positioning systems. The alternative is to suffer a loss of positioning flexibility if alignment is completed prior to chamber pump down.47 To circumvent these issues, a free-space confocal laser scanner system can be used to align both the position and angle of incident light on a photonic chip.48 This system is not much more complicated than a free-space interferometric setup24,49,50 since the photonic chip sample is the only component required to be placed under vacuum.The confocal lens system allows for the adjustment of the position and input angle (k-vector) of a laser onto the entrance pupil of a microscope objective. Since a set of Fourier transform planes is formed by this entrance pupil and the microscope objective’s focus plane, a change in the position on the entrance pupil defines the k-vector of the output beam and hence the angle of incident light onto the chip. Conversely, the k-vector of the input beam defines the location of the beam at the output focus plane. As such, the confocal system allows the position and input angle on the photonic chip to be controlled independently. A fiber collimator is used to input the laser light into the free-space confocal lens system. Light from the output grating coupler is captured by this microscope objective to couple back into the system. A diagram of this flexible free-space confocal lens setup is shown in Fig. 10.1.48 This type of system gives the user both the efficiency of input grating couplers with the flexibility of a free-space optical design for probing nanophotonic devices in vacuum.
