We demonstrate a novel mesh of Mach-Zehnder interferometers (MZIs) for programmable optical processors. We thoroughly analyze the benefits and drawbacks of previously known meshes and compare our newly proposed mesh with these prior architectures, highlighting its unique features and advantages. The proposed mesh, referred to as Bokun mesh, is an architecture that merges the attributes of the prior topologies Diamond and Clements. Similar to Diamond, Bokun provides diagonal paths passing through every individual MZI enabling direct phase monitoring. However, unlike Diamond and similar to Clements, Bokun maintains a minimum optical depth leading to better scalability. Providing the monitoring option, Bokun’s programming is faster improving the total energy efficiency of the processor. The performance of Bokun mesh enabled by an optimal optical depth is also more resilient to the loss and fabrication imperfections compared to architectures with longer depth such as Reck and Diamond. Employing an efficient programming scheme, the proposed architecture improves energy efficiency by 83% maintaining the same computation accuracy for weight matrix changes at 2 kHz.

A mode insensitive switch is proposed and experimentally demonstrated on a silicon-on-insulator platform using a balanced Mach–Zehnder interferometer structure with a mode insensitive phase shifter for on-chip mode division multiplexing interconnects. Switching the first three quasi-transverse electric (TE) modes, consuming less than 40 mW power is demonstrated. The whole system exhibits approximately − 2 , − 3.7 , and − 5.2 d B insertion loss for the TE0, TE1, and TE2 modes at 1550 nm, respectively. The corresponding crosstalk is less than − 8.6 ( − 9 ) , − 8 ( − 10.3 ) , and − 10 d B ( − 10.3 d B ) within the wavelength range of 40 nm (1535–1575 nm) for the cross (bar) states, respectively. The extinction ratios (ERs) for the cross (bar) states are 20.1 (19.5), 22.8 (33.7), and 15.4 dB (18.1 dB) for the TE0, TE1, and TE2 modes at 1550 nm, respectively. The payload transmission is also conducted using non-return-to-zero pseudorandom binary sequence (PRBS)-31 data signals at 10 Gb/s for single-mode transmission and simultaneous three-mode transmissions. For all the scenarios, open eyes are observed.

Mode Division Multiplexing (MDM) is a technique used over the past decade in Silicon Photonics (SiPh) to incorporate more data into communication links by employing higher-order transverse electric or transverse magnetic modes. MDM was primarily used in optical communication; however, in recent years, there have been several applications of MDM in optical computing, including both classical and quantum computing. Although MDM has shown great promise for increasing the throughput of optical communication and the accuracy and fidelity of optical computation, there are a few challenges towards expanding its applications. One major challenge is the lack of process design kits (PDKs) and building block libraries compatible with standard SiPh foundries. Here, we present a comprehensive library of MDM components developed using classical and inverse design, compatible with standard 220 nm SiPh foundries. The library includes thermo-optic phase shifters, mode multiplexers and demultiplexers, mode converters, mode exchangers, and multi-mode interference couplers. We also discuss our recent achievements in MDM for datacom, classical and quantum optical computing, including a mode-selective switch for mode-reconfigurable optical add-drop multiplexer (ROADM), multimode multiply-accumulate operation, and multimode photonic quantum processors.

The design, fabrication, and characterization of an 8 × 8 lossless optical switch, based on semiconductor optical amplifier (SOA) gates, is reported. It comprises three stages of 2 × 2 switches into an 8 × 8 Banyan switch, for a total of 48 SOAs. Three SOAs on each optical path provide gain to compensate for on-chip and fiber coupling loss, thereby making the optical switch lossless. All 64 optical paths demonstrate error-free 10 Gbps NRZ PRBS-31 transmission with at least 30 dB signal-to-noise ratio and less than 0.9 dB power penalty.

Reverse gate-current of AlGaN/GaN heterostructure field-effect transistor is studied over a wide range of lattice-temperatures from 150 to 490 K. For gate-source voltages approaching zero, volt signatures of gate-to-2-D electron gas leakage through the sidewalls of the mesa are observed. For exploring this leakage path, a set of devices built on a number of alternative isolation features of different geometries, and with different number of gate-covered sidewalls, are investigated. Among these devices, which were realized on an identical layer structure (produced following the same fabrication technology with identical gate-length and -width), as the number of isolation-feature sidewalls overlapping with the gate metal increases, an increase in the gate-current is observed. The identified sidewall leakage is not only consequential in devices built on isolation-feature geometries presenting more than two sidewalls, but it can also compete with mechanisms such as Poole-Frenkel (PF), Fowler-Nordheim, and trap-assisted tunneling (TAT), which are traditionally considered in devices built on cubic mesas. In this paper, the relevance of these other transport mechanisms is also re-evaluated. It is observed that at temperatures below 320 K, among all of the explored devices, sidewall leakage becomes more dominant than the PF and trap-assisted tunneling (TAT) processes.

We propose and experimentally demonstrate a novel design approach for scalable 3-port and 4-port mode insensitive multimode switching matrices for the first two quasi-transverse electric (TE) modes. The mode insensitive phase shifter ensures less power consumption for simultaneous multimode signal transmission in a mode division multiplexing (MDM) network. At 1550 nm, the 3-port switch exhibits approximately 2.6 dB and 3.3 dB insertion loss for the longest path with a crosstalk of at most 10 dB and 8 dB over a bandwidth of 40 nm (1530 nm to 1570 nm) for ${\mathrm {TE}}_{\mathrm {0}}$ and ${\mathrm {TE}}_{\mathrm {1}}$ modes, respectively. The insertion loss measured for the 4-port switch is approximately 2.7 dB and 3.6 dB at 1550 nm with a corresponding crosstalk less than 8 dB for the two TE modes, respectively. The payload transmission is also performed using both 10 Gb/s non-return-to-zero (NRZ) and 14.0625 Gbaud 4-level pulse-amplitude modulation (PAM4) pseudorandom binary sequence (PRBS)-31 data signal to analyze the performance of the switches. Clear open eyes are observed for both single-mode and simultaneous two-mode transmissions. The low insertion loss, low intermodal crosstalk over a large optical bandwidth, and the clear open eye validate the scalability of the proposed switches for larger port count and higher-order modes.