We demonstrate a high-performance, thermally-robust, C-band GeSi Franz-Keldysh electro-absorption modulator (FK-EAM) featuring a compact 0.5 x 50 μm2 active region on a 300-mm silicon photonics platform. The static electrical and optical performance is experimentally evaluated over wide operating ranges, with wafer-scale results yielding an extinction ratio (ER) of 6.44 ± 0.38 dB and a transmitter penalty (TP) of 9.25 ± 0.15 dB, while the best dies achieve an insertion loss (IL) of ∼ 5 dB, figure-of-merit (FOM = ER/IL) of 1.23, and TP of 8.8 dB at λ = 1550 nm under a 2 V peak-to-peak drive, representing the best reported EAM performance. The temperature-dependent measurements from 25-85 ∘C show a monotonic red-shift rate of 0.8 nm/∘C, while maintaining stable performance with ER > 6 dB, FOM > 1, and TP ∼ 9-10 dB across the wafer, demonstrating temperature-robust modulation performance comparable to state-of-the-art room-temperature EAMs. Furthermore, robust EAM operation beyond the nominal design window (∼1590 nm) is sustained even at 85 ∘C. These results establish GeSi EAMs as a key enabling technology for high-speed, low-power, dense silicon-photonic transceivers, operating reliably in thermally exposed dynamic environments, paving the way for next-generation optical I/O applications.

Silicon-based wavelength division multiplexing (WDM) filters are essential for scaling optical communication capacity in data centers and telecommunications networks. However, extending silicon WDM systems beyond 32 channels with 100 GHz spacing poses significant challenges due to limitations in conventional filter architectures. Here we present the first silicon 64 & times; 100 GHz WDM filter by introducing a novel ring-Mach-Zehnder interferometer (MZI) cascade architecture. Our design utilizes third-order polynomial interconnected circular (TOPIC) bends to construct low-loss half-ring waveguides, facilitating an MZI configuration where the arm length difference is determined entirely by half of the ring structure. This approach ensures precise alignment between the MZI interference peaks and the ring resonator wavelengths, with the MZI FSR being exactly double that of the ring, eliminating the need for dynamic tuning between the MZI and the ring. We demonstrate the concept through a 16 & times; 400 GHz WDM filter with insertion loss of 1.3 +/- 0.6 dB and channel isolation >= 14.3 dB. The 64 & times; 100 GHz implementation, realized using a 4-channel interleaver followed by four 16 & times; 400 GHz WDM filter, achieves insertion loss of 3.2 +/- 1.1 dB and channel isolation >= 10.7 dB. This work opens new possibilities for high-density silicon photonic WDM systems, addressing the growing bandwidth demands of artificial intelligence and machine learning applications.

Reliability-improved GaAs nano-ridge lasers are fabricated on 300 mm silicon. A novel contact-FIN-based design, including an anisotropic InGaP passivation, ensures carrier injection while pulling down the optical modes. Single-facet power exceeding 10 mW is achieved at 25 °C.

We present a novel compact asymmetric bent directional coupler polarization beam splitter (PBS) fabricated on a silicon-on-insulator (SOI) platform using third-order polynomial interconnected circular (TOPIC) bends. The TOPIC bend design provides continuous curvature and curvature derivatives throughout the structure, which minimizes mode transition losses at connection interfaces. This approach allows for the implementation of tight bend radii to enhance extinction ratio performance without compromising insertion loss, thereby resolving the conventional trade-off limitation in bent-coupler PBS designs. The device performance was characterized through cascaded measurements involving up to 40 PBS units, with single-device loss parameters extracted via linear regression analysis. Comprehensive wafer-level testing across 61 dies confirmed excellent design reproducibility and manufacturing tolerance. The optimized PBS achieves insertion losses as low as 0.017 +/- 0.007 dB for TE and 0.019 +/- 0.018 dB for TM polarizations, accompanied by extinction ratios of 25.5 +/- 0.5 dB and 29.6 +/- 6.1 dB, respectively, at 1295 nm wavelength. These results establish new performance benchmarks for silicon photonic PBS devices and demonstrate significant potential for advanced polarization-handling applications in integrated photonic systems.

We demonstrate a silicon ring-based 8-channel WDM filter that achieves a record-low thermal tuning power of 3.80 mW/π (= 420 GHz/mW) per channel and an ultra-compact footprint of 10×160 μm2.

A novel silicon photonic CWDM filter design is proposed and experimentally demonstrated. The design has achieved flat-top transmission across all dies on a wafer, with a device footprint of 48*25 μm2, an insertion loss of 0.24 ± 0.18 dB, and a channel central wavelength standard deviation of 0.77 nm.

We experimentally demonstrate a silicon photonic S-bent directional coupler achieving a coupling variation of only 0.065 over an 80 nm wavelength range, showing superior broadband performance with a fabrication-friendly minimum feature size of 200 nm.

We present the extraction of recombination coefficients for electrically injected monolithic nano-ridge laser diodes by first determining the effective carrier capture time from the small signal modulation response. The effect of the nano-ridge box size on the recombination coefficients is investigated.

A lifetime model is presented to study the diffusion-driven gradual degradation and the recombination-enhanced rapid failure in monolithic InGaAs/GaAs-on-Si nano-ridge lasers induced by high current density at p-contacts. Design guidelines are provided for improving reliability.

The static and dynamic characteristics of electrically injected monolithic nano-ridge lasers emitting around the wavelength of 1030 nm are comprehensively investigated, providing critical insights into their performance and identifying pathways for future improvement. Key laser parameters such as the D-factor, the K-factor, the differential gain and the gain compression factor are extracted. Recombination coefficients and carrier escape times are determined by taking the effective carrier capture times derived from the small-signal modulation response. Additionally, the impact of the nano-ridge box size on the recombination coefficients is evaluated, highlighting the role of structural design in optimizing device performance and reliability.

This study investigates the thermal crosstalk in an InP reflective semiconductor optical amplifier (RSOA) array, flip-chip bonded to a Si photonics wafer. Previous studies [1], [2] have focused on single-channel operation, and in the present study we experimentally characterise all-channel operation and support our findings with thermal simulation results.

We present broadband silicon-nitride to polymer waveguide adiabatic coupling interfaces with sub-1 dB loss around 1310 nm, validated using two integration methods: lithography and flip-chip bonding, achieving sub-2 dB chip-to-chip and chip-to-fiber coupling loss.

As future computing systems increasingly demand higher bandwidth and energy efficiency, the development of faster and more efficient modulators is becoming a critical component of optical transceivers. In this paper, we present O-band silicon ring modulators (RMs) which are highly optimized in terms of electro- and thermo-optic modulations for an energy efficient optical I/O link. A highly optimized vertical p-n junction design is used to enhance the electro-optic efficiency, and fabricated RMs have 2 dB better transmitter penalty and bandwidth trade-off than previously reported lateral p-n junction based RMs. To improve the thermo-optic efficiency, Si substrate undercut (UCUT) process module is introduced. The efficiency increases more than 2.7 times with UCUT as 106-GHz/mW (0.608 nm/mW) with 5 mu m radius RM, and no performance degradation is observed with 50-Gb/s NRZ operation.

We propose a measurement system that enables the rapid measurement of insertion loss and polarization-dependent loss using a parallel test setup with a fiber array, and the calibration procedure to be used within this system. By applying rough scan methods, we have developed a calibration algorithm that efficiently finds the accurately optimized state of polarization in minimal time. Through conducting on-wafer spectral optical power measurements, we compared conventional applications and our proposed algorithms. The results demonstrate that our method enables the almost simultaneous measurement of the spectral responses of multiple optical components. Moreover, the method enables to measure these responses with a well-defined input state of polarization (SOP) applied to each path individually. This novel approach holds promise for enhancing accuracy and cost-effectiveness in insertion loss and polarization-dependent loss measurements.

Co-Packaged Optics applications require scalable and high-yield optical interfacing solutions to silicon photonic chiplets, offering low-loss, broadband, and polarization-independent optical coupling while maintaining compatibility with widely used approaches for electrical redistribution. We present two heterogeneous integration techniques that enable high-density electrical and optical I/O connections, utilizing adiabatic coupling between on-chip silicon nitride (SiN) waveguides and package-level polymer optical waveguides. In the first approach, polymer waveguides are patterned using standard lithography directly on the surface of the photonic chip, ensuring compatibility with chip embedding as commonly employed in chip-first fanout wafer-level packaging. In the second approach, photonic chips are flip-chip bonded to the package substrate. Both techniques have been experimentally validated, achieving a coupling efficiency near 1 dB between SiN and polymer waveguides in O-band, for both TE and TM polarizations. SiN tapers were designed using the "Mono" method to optimize phase-matching conditions between the two waveguides, a critical requirement for integrating diverse optical components. These results demonstrate the potential of polymer waveguides in Co-Packaged Optics applications, achieving sub-2 dB chip-to-chip and chip-to-fiber coupling losses.

We present a C-band GeSi EAM fabricated on a 300mm silicon photonics platform, exhibiting a bandwidth beyond 110 GHz. Transmission of 200, 212.5 and 224 GBaud PAM-4 is demonstrated, meeting the 25% OH SD-FEC threshold, highlighting the device's potential for next-generation optical scale-up fabrics. (c) 2025 The Author(s)

This article presents an electrothermal model for a InGaAs/GaAs quantum well (QW) nanoridge (NR) laser diode monolithically integrated in a Si photonics wafer. Heat conduction inside the laser is modeled using both the finite element (FE) method and the Boltzmann transport equation (BTE) for phonon transport. The BTE model captures nanoscale thermal effects and is used to calibrate the FE model, which captures large-scale effects, such as heat spreading in the Si substrate. Two methods are used to obtain the power distribution inside the laser: first, electrical TCAD simulation is carried out for the p-i-n structure and second, an equivalent electrical circuit is extracted from measurement data. The simulation results of the different thermal and electrical models are compared and finally experimentally validated using spectral measurements. We conclude that the simulated laser thermal resistance R-th,R-sim=74 +/- 8.9 mm-K/W is in good agreement with the experiments R-th,R-exp=78.2 +/- 15.2 mm-K/W; furthermore, the model predicts a local hot spot at the p-contact, which is a bottleneck for laser reliability. This insight will drive future design iterations.

Optical transceivers for data center applications require multi-wavelength light sources, which can either be integrated or external from the transceiver die. Scaling up the number of communication channels implies the need for large laser arrays. Since the energy efficiency of semiconductor lasers is very sensitive to temperature, it is imperative to employ a thermal-aware design and minimize self-heating and thermal crosstalk. In this paper, a thermal scaling analysis is performed on hybrid, flip-chip integrated InP-on-Si lasers. A finite element thermal model of a single gain section laser is validated with experimental measurement of the laser thermal resistance and extrapolated to accomodate multi-section operation. The impact of adding a top-side heat sink as well as increasing laser length and width are investigated. The detailed 3D simulation results are used to build a compact, coupled thermo-optic model of a large array of multiple lasers, considering thermal crosstalk. Finally, this model is applied to a test case with 8 WDM channels and 8 ports. Depending on the configuration (integrated vs. external) and ambient temperature, different optimal designs arise based on both energy efficiency and module footprint. The presented modelling framework is generic; it can be applied to different types of lasers and systems.

We present a semi-analytical model that can accurately explain the working principle behind the recently reported electrically injected In0.2Ga0.8As/GaAs monolithic nano-ridge lasers and more importantly show how the model can be used to study the effect of device parameters on the spectral behavior, the slope efficiency and the threshold gain. We show that mode beating between the fundamental mode and a higher order mode is fundamental in the operation of these lasers. Analytical expressions for codirectional mode coupling are used in developing the round-trip laser model. Results from analytical expressions are verified by comparisons with simulations and the model is supported later by measurement results.

We present assembly results for a variety of single channel and array-based InP lasers built on the Sivers Photonics InP100 platform, combined with imec's 200 mm SiPho platform, and leveraging the sub-micron placement capabilities of the ASMPT AMICRA NANO. This focuses on the scaling from single devices ($0.3 \times 0.35 \text{mm}$) to $4 \times(0.8 \times 0.5 \text{mm})$ and $8 \times(1.6 \times 0.5 \text{mm})$ array devices, considering the interaction between the bond tool, device, landing site and processing conditions. Furthermore, we discuss how alignment strategies and the choice of alignment markers and bonding parameters can be fine-tuned to minimize misalignment in the XY plane and in the rotation of the bonded devices. In addition, we utilise white light interferometry to determine any warp or bow present in the device and how that may affect the individual channels in a multi-channel device. Finally, we use LIV curves as our primary determinant of how successful the bonding is, with both inter- and intra-device comparisons. We report post-bond results with $<500 \text{nm}$ misalignment in the XY plane, $<0.065^{\circ}$ rotational misalignment and $<500 \text{nm}$ tilt across a variety of device shapes and sizes and best device outputs of 25 mW and 2.5 dB coupling loss.