In recent years silicon photonics has become a considerable mainstream technology, especially in telecommunications fields to overcome the limitations imposed by copper-based technology. Nanoscale photonic technologies have attracted a lot of attention to co-develop photonic and electronic devices on silicon (Si) to provide a highly integrated electronic–photonic platform. Silicon-on-insulator (SOI) technology that relies heavily on the contrasted indices of Si and SiO2, enables the design and integration of these photonic devices in submicronic scales, similar to the devices produced by a standard CMOS fabrication platform in the electronics industry. One of the key challenges with these submicronic waveguide devices is to enable efficient coupling with fibre, which is mainly due to the mode-field differences between fibre and the waveguide, and their relative misalignments. To overcome this challenge, various techniques including prism, butt and grating coupling have been proposed. Among them, although butt coupling is an elegant solution for low loss and wideband operation, it often requires post-processing for accurate polishing and dicing to taper the waveguide edges. Therefore, it is not suitable for wafer-scale testing. Grating couplers, which mostly perform out of the plane coupling between a fibre and a waveguide, are also an attractive solution as light can be coupled in and out everywhere on the chip, opening the way for wafer-scale testing. However, despite such advantages, grating couplers often exhibit low coupling efficiency (CE) due to downward radiation of light that propagates towards substrate through buried oxide (BOX) which comprises 35%-45% of total incident light. Grating couplers are also very sensitive to the wavelength of the light as different wavelengths exhibit specific diffraction properties at the grating, which cause a narrow coupling bandwidth. In this thesis we have studied various techniques to improve the coupling efficiency and coupling bandwidth of the grating couplers. We have used the finite difference time domain (FDTD) and Eigenmode Expansion (EME) methods to study the interaction of light with grating. The directionality of the coupler which determines the coupling efficiency has been improved by means of silicon mirrors in the BOX layer that essentially redirect the light propagates toward substrate. For improvement of directionality, an ultra-subwavelength grating coupler has also been developed with an engineered grating structure which exhibits high coupling efficiency and bandwidth without the need for bottom mirrors. The grating coupler only converts vertical dimension into nano scale, leaving the lateral width in micrometre range typically >15 μm. In order to connect the grating coupler with a nanophotonic waveguide, the grating structure needs to be matched in dimensions both vertically and laterally. Conventionally, to meet the requirement the width of grating structure is gradually tapered to nano scale. The coupling efficiency relies highly on the taper length, which is typically hundreds of micrometres. Such a long taper waveguide causes an unnecessarily large footprint of the photonic integrated circuits. In order to minimise the length of the taper while retaining high coupling efficiency, we have designed two different types of tapered waveguides. One of them is a partially overlaid tapered waveguide and the other is a hollow tapered waveguide.

Light coupling in dimension mismatch waveguides for silicon photonic integrated circuits

A directional-coupler-based 1 
 $$\times$$
  1 silicon-on-insulator photonic ON–OFF switch with a complementary metal–oxide–semiconductor (CMOS)-compatible driving voltage is proposed in this paper. The directional coupling of the switch is accomplished by a carrier injection method with the help of a P-i-N diode phase shifter. The advantage of using the directional coupler as a switch is its smaller layout requirement, which makes it more suitable for use in integrated photonic applications. The ON–OFF switch has potential applications in programmable photonic switch fabrics, where the OFF-state behavior is used to prevent the input from being propagated to the output. The proposed switch offers a wide range of cross power-coupling coefficient $$(\kappa ^2)$$
 values, ranging from 0 to 0.3 for the ON-state and from 0.79 to 1 for the OFF-state. The finite-difference beam propagation and two-dimensional (2D) finite-difference time-domain methods are used for photonic simulations. The results are compared analytically using the coupled mode theory with the help of the MATLAB software package.

The design and analysis of a CMOS-compatible silicon photonic ON–OFF switch based on a mode-coupling mechanism

Diffusion doped p-i-n/p-n diodes in SOI substrate is proposed for the fabrication of active silicon photonics devices with scalable waveguide cross-sections. The p-type and n-type diffusion doping parameters are optimized for the fabrication of tunable single-mode waveguide phase-shifters with microns to submicron cross-sectional dimensions. The simulations results show that the shape of depletion layer can be effectively engineered by suitably positioning the rib waveguide with respect to the gap between doping windows. We could thus introduce an additional control parameter to optimize over-all figure of merits of the phase-shifter for various applications. For an optimized set of diffusion parameters, the VπLπ of single-mode waveguides designed with 1μm, 0.5μm, and 0.25μm device layers (under reverse bias operating in TE-polarization at λ ~ 1550 nm) are found as 2.7 V-cm, 2.1 V-cm, and 1.6 V-cm, respectively. The typical p-n junction capacitance of an optimized 0.25μm single-mode waveguide is estimated to be < 0.5 fF/μm, which is comparable to that of ion-implanted p-n waveguide junctions.

Diffusion doped p-i-n/p-n diodes for scalable silicon photonics devices

This study focuses on the Optomechanical bichromatic wavelength switching system as an indirect two-color up-conversion process that relies on optical force and nanorod scattering effects. This system is used to control light coupling between four parallel optical waveguides made of silicon nitride (Si3N4) which form two identical parts. The parallel waveguides with 0.5 µm × 0.5 µm cross-section and 220 µm lengths are suspended on a silica (SiO2) substrate embedded with the array of square silicon (Si) nanorods. By mid-IR plane wave illumination, as control light, with different intensities and different wavelengths on nanorods, scattering would increase and result in an improvement in attractive gradient optical force exerted on waveguides. Via bending waveguides toward each other, caused by optical gradient force, two different visible lights, as probe signals, propagating in the first waveguide of each section would couple to the adjacent waveguide. Simulation results reveal that when the distance between the parallel waveguides in the equilibrium position is 100 nm and the intensity of mid-IR light is 1.28 mW/µm2 total coupling would occur in two situations: 1- when the control light is 4.5 µm, the probe light with 713 nm wavelength is transmitted to the output, 2- when the control light is 3 µm, the probe light with 609 nm wavelength is transmitted to the output. In the first case 1.92 pN/µm optical force is needed to bend each waveguide by 9 nm and in the second one, 1.28 pN/µm optical force is needed to bend each waveguide by 6 nm for total coupling. The efficiency of the coupled waveguides system is %88.6 for 609 nm probe light injection and %96.5 for 713 nm probe light injection. 

/pdf/a-proposal-for-optomechanical-bichromatic-wavelength-2vvqnsxx.pdf

A proposal for optomechanical bichromatic wavelength switching for two-color up-conversion application

Polarization dependent plasma dispersion effect is investigated for single-mode SOI waveguides (λ ∼ 1550 nm) with laterally diffused p-n junction. The lateral p-n junction of desired length is formed initially in 1.5 μm thick SOI layer (BOX-2 μm) by diffusion doping and then the single-mode waveguides with appropriate design parameters (width W, height H, and slab height h) is defined along the junction for simulations studies to estimate V π and L π . It is observed that the device becomes more and more polarization sensitive as the device layer thickness (H) decreases, which is justified appropriately by calculating polarization dependent confinement factors as well as free carrier overlaps with the corresponding guided modes.

Polarization dependent electro-optic effect in SOI waveguides with laterally diffused P-N junction

Theoretical and experimental studies reveal that a predefined single-mode rib waveguide fabricated in silicon-on-insulator (SOI) substrate with a device layer thickness of 2 μm can be adiabatically trimmed down to submicron waveguide dimensions (<1  μm), resulting in regional modification of waveguide properties. The fabrication process involves physical trimming/removal of a waveguide surface by plasma etchants that is spatially filtered by a shadow mask with a rectangular aperture inside a reactive ion etching system. The exact position of a shadow mask above a sample surface has been optimized (∼500  μm) to obtain the desired adiabatic spot-size converters of length up to 1 mm at both ends of the trimmed waveguides. For experimental demonstration, three different sets of 15-mm-long single-mode waveguides fabricated in 2-μm SOI were adiabatically trimmed in the middle for three different lengths of 3, 5, and 7 mm, respectively. Excess propagation loss and group index of a trimmed submicron waveguide section were extracted by analyzing the wavelength-dependent Fabry–Perot transmission characteristics of the device with polished input/output end facets. The insertion loss of a typical spot-size converter designed for the guidance of TE-like polarization has been recorded to be ∼0.25  dB for a wide range of wavelengths (1500  nm≤λ≤1600  nm). As predicted by numerical simulation, no polarization rotation has been observed in all the trimmed submicron waveguides. The proposed surface trimming technique can be potentially used to tune the waveguide cross-section/geometry for phase error correction and/or to avail stronger light-matter interactions at a desired location of an integrated optical circuit.

Design and fabrication of surface trimmed silicon-on-insulator waveguide with adiabatic spot-size converters

Single-mode silicon p-i-n waveguides with varying cross-sections have been studied experimentally for on-chip photodetection at an operating wavelength λ ~ 1550 nm. It has been shown that the quantum efficiency increases with decreasing waveguide cross-section. The performance of such a photodetector can be modelled in terms of density of surface states, bulk two photon absorption co-efficient, and waveguide loss parameters.

Sreevatsa Kurudi

Papers

Design and fabrication of surface trimmed silicon-on-insulator waveguide with adiabatic spot-size converters

Diffusion doped p-i-n/p-n diodes for scalable silicon photonics devices

Scaling of Silicon PIN Waveguide Photodetector at 1550 nm Wavelength