The 2 μm wavelength band has become a promising candidate to be the next communication window. We demonstrate high-speed modulators based on a 220 nm silicon-on-insulator platform working at a wavelength of 1950 nm, using the free carrier plasma dispersion effect in silicon. A Mach–Zehnder interferometer modulator and a microring modulator have been characterized. At 1950 nm, the carrier-depletion modulator operates at a data rate of 20 Gbit/s with an extinction ratio of 5.8 dB and insertion loss of 13 dB. The modulation efficiency (${\mathrm{V}}_{\pi }·{\mathrm{L}}_{\pi }$) is 2.68 V·cm at 4 V reverse bias. The device operation is broadband, and we also characterize its performance at 1550 nm. At 1550 nm, an open eye is obtained at 30 Gbit/s. The difference in bandwidth is caused by the bandwidth limit of the 2 μm measurement setup. We also show a ring modulator paired with a low power integrated driver working in hybrid carrier depletion and injection mode at a data rate of 3 Gbit/s with power consumption of 2.38 pJ/bit in the 2 μm wavelength range. This work is a proof of principle demonstration and paves a route toward a full silicon-based transceiver in the 2 μm window.

The explosive growth of deep learning applications has triggered a new era in computing hardware, targeting the efficient deployment of multiply-and-accumulate operations. In this realm, integrated photonics have come to the foreground as a promising energy efficient deep learning technology platform for enabling ultra-high compute rates. However, despite integrated photonic neural network layouts have already penetrated successfully the deep learning era, their compute rate and noise-related characteristics are still far beyond their promise for high-speed photonic engines. Herein, we demonstrate experimentally a noise-resilient deep learning coherent photonic neural network layout that operates at 10GMAC/sec/axon compute rates and follows a noise-resilient training model. The coherent photonic neural network has been fabricated as a silicon photonic chip and its MNIST classification performance was experimentally evaluated to support accuracy values of >99% and >98% at 5 and 10GMAC/sec/axon, respectively, offering 6× higher on-chip compute rates and >7% accuracy improvement over state-of-the-art coherent implementations.

In recent years, silicon nitride has drawn attention for the realisation of integrated photonic devices due to its fabrication flexibility and advantageous intrinsic properties that can be tailored to fulfill the requirements of different linear and non-linear photonic applications. This paper focuses on our progress in the demonstration of enhanced functionalities in the near infrared wavelength regime with our low temperature (<350°) SiN platform. It discusses (de)multiplexing devices, nonlinear all optical conversion, photonic crystal structures, the integration with novel phase change materials, and introduces applications in the 2μm wavelength range.

In this review we present some of the recent advances in the field of silicon nitride photonic integrated circuits. The review focuses on the material deposition techniques currently available, illustrating the capabilities of each technique. The review then expands on the functionalisation of the platform to achieve nonlinear processing, optical modulation, nonvolatile optical memories and integration with III-V materials to obtain lasing or gain capabilities.

We propose a reconfigurable and non-volatile Bragg grating in the telecommunication C-band based on the combination of novel low-loss phase-change materials (specifically Ge₂Sb₂Se₄Te₁ and Sb₂S₃) with a silicon nitride platform. The Bragg grating is formed by arrayed cells of phase-change material, whose crystallisation fraction modifies the Bragg wavelength and extinction ratio. These devices could be used in integrated photonic circuits for optical communications applications in smart filters and Bragg mirrors and could also find use in tuneable ring resonators, Mach–Zehnder interferometers or frequency selectors for future laser on chip applications. In the case of Ge₂Sb₂Se₄Te₁, crystallisation produces a Bragg resonance shift up to ∼ 15 nm, accompanied with a large amplitude modulation (insertion loss of 22 dB). Using Sb₂S₃, low losses are presented in both states of the phase change material, obtaining a ∼ 7 nm red-shift in the Bragg wavelength. The gratings are evaluated for two period numbers, 100 and 200 periods. The number of periods determines the bandwidth and extinction ratio of the filters. Increasing the number of periods increases the extinction ratio and reflected power, also narrowing the bandwidth. This results in a trade-off between device size and performance. Finally, we combine both phase-change materials in a single Bragg grating to provide both frequency and amplitude modulation. A defect is introduced in the Sb₂S₃ Bragg grating, producing a high quality factor resonance (Q ∼ 10⁴) which can be shifted by 7 nm via crystallisation. A GSST cell is then placed in the defect which can modulate the transmission amplitude from low loss to below -16 dB.