Search

Recent developments in computational inverse design offer the promise of significantly reducing the footprint and allowing complex optical functionality in silicon photonic components as compared to existing conventional building blocks. However, reliable fabrication of such components is one of the major bottlenecks in its widespread adoption. A common characteristic of such designs is the presence of small features that have meaningful impact on the optical performance. Current approaches to tackle this problem consider designing for robustness, such as by co-optimizing for over- and under-etched geometries at the design stage. This is often followed with a design-for-manufacturing optimization step to meet specifications of a foundry such as minimum feature size and curvature radii. Those approaches often incur additional significant computational costs as well as a reduction in peak optical performance. In this work, we highlight our recent progress to bridge the gap between inverse design methods and their ability to deliver reliable and manufacturable designs. We observe that the so-called parameterized shape optimization methods are more likely to produce robust designs for certain classes of components, as showcased in integrated mode converter designs. For components that benefit from topological inverse design such as wavelength demultiplexers, we propose a new optimization penalty that naturally leads the optimizer towards more robust designs. In a new research direction, we also consider improving fabrication reliability by the development and use of data-driven predictive models for fabrication. Leveraging deep learning tools, we present prediction and correction models that improve fabrication outcomes for a variety of components made at an e-beam prototyping foundry., SPIE OPTO 2023 - Integrated Optics: Devices, Materials, and Technologies XXVII, January 28- February 3, 2023, San Francisco, California, United States, Series: Proceedings of SPIE

Computational inverse design techniques have shown potential to become reliable means for designing compact nanophotonic devices without compromising the performance. Much effort has been made to reduce the computation cost involved in the optimization process and obtain final designs that are robust to fabrication imperfections. In this work, we experimentally demonstrate TE0-TE1 and TE1-TE3 mode converters (MCs) on the silicon-on-insulator platform designed using the computationally efficient shape optimization method. These MCs have mode conversion efficiencies above 95%, and the insertion loss ranges from 0.3 dB to 1 dB over a wavelength span of 80 nm ranging from 1.5 µm to 1.58 µm. Maximum modal crosstalk found experimentally in the C-band is -19 dB. The conversion efficiency drops at most by 2.2% at 1.55 µm for 10 nm over/under etch, implying good robustness to dimensional variations. We present the mode conversion mechanism of these MCs by studying the simulated electromagnetic field patterns and validate with supportive data. We also demonstrate their performance in the time domain with a 28 Gbps OOK and a 20 GBaud PAM-4 payload transmissions, which supports their utility for high throughput data communications. The open eye diagrams exhibit Q-factors of 8 dB.

The performance of integrated silicon photonic devices is sensitive to small structural variations that arise from imperfections in the nanofabrication process. This sensitivity is exacerbated for next-generation devices that require fine feature sizes to push the limits of performance. In this work, we present a deep convolutional neural network model to predict fabrication variations in planar silicon photonic devices and verify their manufacturing feasibility prior to prototyping. Our model is trained on a modest set of scanning electron microscope images of structures that experience dimensional inaccuracies stemming from combined contributions from proximity effects in lithography and loading effects in dry etching. Our model quickly and accurately predicts over/under-etching, corner rounding, filling of narrow channels and holes, and washing away of small features in a photonic device. With this, the expected performance of a device can be predicted through an extra simulation and any necessary design corrections can be made prior to fabrication.

In this paper, we present the experimental results for integrated photonic devices optimized with an energy-constrained inverse design method. When this constraint is applied, optimizations are directed to solutions that contain the optical field inside the waveguide core medium, leading to more robust designs with relatively larger minimum feature size. We optimize three components: a mode converter (MC), a 1310 nm/1550 nm wavelength duplexer, and a three-channel C-band wavelength demultiplexer for coarse wavelength division multiplexing (CWDM) application with 50 nm channel spacing. The energy constraint leads to nearly binarized structures without applying independent binarization stage. It also reduces the appearance of small features. In the MC, well-binarized design, improved insertion loss, and cross talk are obtained as a result. Furthermore, the proposed constraint improves the robustness to fabrication imperfections as shown in the duplexer design. With energy constraint optimization, the corresponding spectrum shifts for the duplexer under ± 10 nm dimensional variations are reduced from 105 nm to 55 nm and from 72 nm to 60 nm for the 1310 nm and 1550 nm channel, respectively. In the CWDM demultiplexer, robustness toward ± 10 nm fabrication error is improved by a factor of 2. The introduction of the energy constraint into topological optimization demonstrates computational gain with better-performing designs.

A novel and energy efficient mode insensitive switch building block is proposed and experimentally demonstrated on a silicon-on-insulator platform. Based on a Mach-Zehnder interferometer, the switch uses a relatively compact mode insensitive phase shifter which includes a mode exchanger. The novel structure realizes the exact same phase shift for all modes by exchanging the modes midway within the phase shifter. The design approach leads to reduced power consumption otherwise not possible. Switching the first two quasi transverse electric (TE) modes simultaneously consumes 25.6 mW of power, an approximately 30% reduction from previous reported demonstrations. The measured insertion loss is 3.1 dB on average with a worst-case crosstalk of −14.9 dB over a 40 nm optical bandwidth from 1530 nm to 1570 nm. The design methodology enables scalability up to four optical modes.

Silicon photonics spectrometers have great potential for applications in medicine and hazard detection. However, silicon spectrometers are very sensitive to fabrication imperfections and environmental conditions. Here, we study the use of deep-learning algorithms to improve tolerance of Fourier-transform spectrometers against fabrication imperfections and temperature variations., 2021 IEEE 17th International Conference on Group IV Photonics (GFP), December 7-10, 2021, Malaga, Spain

Design of modern integrated nanophotonic components requires increasingly sophisticated optical simulation and optimization tools. Modeling and computational challenges arise with the increase in the number of design parameters and the introduction of multiple and often competing performance criteria. In such high dimensional design parameter spaces, it becomes difficult to navigate, explore, and visualize the best candidate designs that satisfy all the requirements. We present our recently developed approach that leverages dimensionality reduction - an area of machine learning – to identify and efficiently investigate only the most relevant portion of the design space. Once this reduced space is found, mapping and optimization can often be achieved several orders of magnitude faster than in the original design space. We showcase our approach on several design scenarios focusing on components such as optical grating couplers and power splitters. We employ principal component analysis for linear dimensionality reduction, achieving impressive performance despite its simplicity. We also demonstrate the use of a non-linear technique, i.e. neural-network based autoencoders, which can improve the effectiveness of dimensionality reduction even further. All components have nontrivial regions of interest in their design space that are identified and explored through the evaluation of various performance metrics. Visualizations of these regions offer a global picture of device behavior. Different component geometries can then be chosen depending on specific performance requirements or fabrication constraints. The proposed framework can be easily integrated into various design toolkits., Nanophotonics and Micro/Nano Optics VII, Oct. 10-20, 2021, Nantong, China, Series: Proceedings of SPIE

We present a novel, to the best of our knowledge, remote gas detection and identification technique based on correlation spectroscopy with a piezoelectric tunable fiber-optic Fabry–Perot filter. We show that the spectral correlation amplitude between the filter transmission window and gas absorption features is related to the gas absorption optical depth, and that different gases can be distinguished from one another using their correlation signal phase. Using a previously captured telluric-corrected high-resolution near-infrared spectrum of Venus, we show that the radial velocity of Venus can be extracted from the phase of higher order harmonic lock-in signals. This correlation spectroscopy technique has applications in the detection and radial velocity determination of weak spectral features in astronomy and remote sensing. We experimentally demonstrate a remote C O 2 detection system using a lock-in amplifier, fiber-optic Fabry–Perot filter, and single channel avalanche photodiode.

We demonstrate inverse-designed mode converters with conversion efficiencies higher than 98.3% and insertion losses below 0.07 dB at the wavelength of 1.55μm. Despite their compact footprints, simulation results show tolerance to fabrication errors up to 10 nm under/over etch., 2021 IEEE Photonics Conference (IPC), October 18-21, 2021, Vancouver, BC, Canada

We show how existing iterative methods can be used to efficiently and accurately calculate Bloch periodic solutions of Maxwell’s equations in arbitrary geometries. This is carried out in the complex-wavevector domain using a commercial frequency-domain finite-element solver that is available to the general user. The method is capable of dealing with leaky Bloch mode solutions, and is extremely efficient even for 3D geometries with non-trivial material distributions. We perform independent finite-difference time-domain simulations of Maxwell’s equations to confirm our results. This comparison demonstrates that the iterative mode finder is more accurate, since it provides the true solutions in the complex-wavevector domain and removes the need for additional signal processing and fitting. Due to its efficiency, generality and reliability, this technique is well suited for complex and novel design tasks in integrated photonics, and also for a wider range of photonics problems.

The absorption spectra of distant targets can be used for remote gas detection. However, the detection of very faint absorption features is difficult since noise can dominate the molecular signal in the acquired spectrum. Rather than using typical dispersive spectroscopy, high sensitivity remote gas sensing is possible by correlation spectroscopy using a spectral matching filter. In this work, we use a thermo-optically tunable nanophotonic silicon waveguide ring resonator as an integrated spectral filter to obtain a correlation signal proportional to the absorption spectrum of a specific gas. We demonstrate the ability to resolve very small absorption depth changes using integrated photonics at the level required for the characterization of exoplanet atmospheres during transits of their host stars., 2021 Photonics North (PN), May 31 - June 2, 2021, Toronto, ON, Canada

We present fully apodized and perfectly vertical surface grating couplers in 300 nm silicon waveguides. We achieve ultrahigh coupling efficiency of -0.35 dB at 1550 nm for an optical fiber with 10.4 µm mode-fiber-diameter. To this end, we followed a two stage process in which, we first optimized a pool of periodic grating couplers using machine learning techniques, and then apodized them over 100 parameters using the gradient based adjoint- method. Using a simple fabrication tolerance analysis, we also show that segment variations mostly causes a wavelength shift for the maximum coupling efficiency of the apodized grating couplers, similar to those typically observed for periodic grating couplers., Integrated Optics: Devices, Materials, and Technologies XXV, March 6-12, 2021, Online Only, United States, Series: Proceedings of SPIE

Modern design of photonic devices is quickly and steadily departing from classical geometries to focus more and more on non-trivial structures and metamaterials. These devices are governed by a multitude of parameters and the optimal design requires to simultaneously consider different figure of merits. In this invited talk we will present our recent work on the application of machine learning tools to the multi-objective optimization of multi-parameter photonic devices. In particular, we will demonstrate the potentiality of dimensionality reduction for the analysis of the complex design space of subwavelength metamaterials devices.

We present a compact silicon-based surface grating antenna design with a high diffraction efficiency of 89% (-0.5 dB) and directionality of 0.94. The antenna is designed with subwavelength-based L-shaped radiating elements in a 300-nm silicon core, maintaining high efficiency with a compact footprint of 7.6 µm × 4.5 µm. The reflectivity remains below -10 dB over the S, C and L optical communication bands. A broad 1-dB bandwidth of 230 nm in diffraction efficiency is achieved with a central wavelength of 1550 nm.

Design of modern photonic devices requires to handle a large number of parameters and figures of merit. By scaling down the complexity of the problem, machine learning dimensionality reduction enables the discovery of better performing devices, higher integration scale, and efficient evaluation of fabrication tolerances., 2021 European Conference on Optical Communication (ECOC), Sept 13-16, 2021, Bordeaux, France

Exoplanetary biosignatures, molecular compounds which indicate a likelihood of extraterrestrial life, can be detected by highly sensitive spectroscopy of starlight which passes through the atmospheres of exoplanets towards the Earth. Such sensitive measurements can only be accomplished with the next generation of telescopes, leading to a corresponding increase in cost and complexity spectrometers. Integrated astrophotonic instruments are well-suited to address these challenges through their low-cost fabrication and compact geometries. We propose and characterize an integrated photonic gas sensor which detects the correlation between the near-infrared quasi-periodic vibronic absorption line spectrum of a gas and a silicon waveguide ring resonator transmittance comb. This technique enables lock-in amplification detection for real-time detection of faint biosignatures for reduced observation timescales and rapid exoplanetary atmosphere surveys using highly compact instrumentation.

We present a comparative experimental study of three silicon photonic echelle grating demultiplexers that are integrated with a Mach-Zehnder interferometer (MZI) launch structure. By appropriate choice of the MZI configuration, the temperature induced shift of the demultiplexer channel wavelengths can be suppressed (athermal) or enhanced (super-thermal) or be controlled by an on-chip micro-heater. The latter two configurations allow the channel wavelengths to be actively tuned using lower power than possible by temperature tuning a conventional echelle demultiplexer. In the athermal configuration, the measured channel spectral shift is reduced to less than 10 pm/°C, compared to the 83 pm/°C shift for an unmodified echelle device. In super-thermal operation an enhanced channel temperature tuning rate of 170 pm/°C is achieved. Finally, by modulating the MZI phase with an on-chip heater, the demultiplexer channels can be actively tuned to correct for ambient temperature fluctuations up to 20 °C, using a drive current of less than 20 mA.

Absorption spectroscopy is widely used in sensing and astronomy to understand molecular compositions on microscopic to cosmological scales. However, typical dispersive spectroscopic techniques require multichannel detection, fundamentally limiting the ability to detect extremely weak signals when compared to direct photometric methods. We report the realization of direct spectral molecular detection using a silicon nanophotonic waveguide resonator, obviating dispersive spectral acquisition. We use a thermally tunable silicon ring resonator with a transmission spectrum matched and cross-correlated to the quasi-periodic vibronic absorption lines of hydrogen cyanide. We show that the correlation peak amplitude is proportional to the number of overlapping ring resonances and gas lines, and that molecular specificity is obtained from the phase of the correlation signal in a single detection channel. Our results demonstrate on-chip correlation spectroscopy that is less restricted by the signal-to-noise penalty of other spectroscopic approaches, enabling the detection of faint spectral signatures.

We present an integrated photonic correlation filter for the detection of remote HCN gas. We demonstrate sensing of absorption due to HCN using thermo-optic tuning of the drop port transmission spectrum in a silicon waveguide ring resonator., 2020 IEEE Photonics Conference (IPC), September 28 - October 1, 2020, Vancouver, BC, Canada

We study the fabrication tolerance of two types of grating antennas featuring U-shape and L-shape structures. The L-shape design is found to be more fabrication tolerant, showing performance degradation of less than 2% assuming common fabrication imperfections, making it promising for optical phased array applications., 2020 IEEE Photonics Conference (IPC), September 28 - October 1, 2020, Vancouver, BC, Canada

The optimization of complex high-dimensional photonic structures is often limited by computational resources. Current techniques based on global optimization algorithms or shape/topology inverse design treat design variables as entirely independent. However, there is often correlation between the input variables and patterns in the design outcomes. We review our strategy of using machine learning pattern recognition for building the performance map of a high-dimensional design space, thereby quickly guiding the search to a small region of interest and significantly improving the computational efficiency. This strategy is found beneficial in both forward and inverse design process flow., 2020 Photonics North (PN), May 26-28, 2020, Niagara Falls, Ontario, Series: Photonics North