Article Highlight | 21-Jul-2025

Thermal scaling analysis of large hybrid laser arrays for co-packaged optics, published in ieee journal of selected topics in quantum electronics

This study investigates how scaling the size of laser arrays has strong implications for self-heating and optical performance

Institute of Electrical and Electronics Engineers

The trend of ever-increasing data center traffic has pushed for more efficient I/O links. Silicon photonics is a promising route to enable communication in the optical domain using fiber links. Si photonics has the main advantage that it leverages decades of CMOS processing developments to fabricate photonic integrated circuits. One important drawback however is the lack of on-chip light sources. Silicon’s indirect band gap makes the material unfit for creating a laser gain medium. The work-around in industry and academia is the integration of III-V materials on Si to create a laser. Lattice mismatch between these materials makes it challenging to create high-quality crystal structures. At IMEC, one of the laser integration approaches we are developing is flip-chip bonding of InP laser dies on our Si photonic wafers. This is referred to as ‘hybrid integration’.  The laser dies are picked up and carefully placed with sub-300 nm precision on the Si photonic wafer [1]. The light is butt-coupled to a SiN waveguide. The laser dies are supplied by Sivers Photonics and the bonding tools by ASMPT.

Optical transceivers employ light at multiple wavelengths to increase the data that can be sent per fiber. To enable this, multi-wavelength light sources are required. Unfortunately, scaling the laser array size increases thermal crosstalk. As both laser efficiency and reliability are very sensitive to temperature [2], it is important to consider thermal performance during the design stage. In this paper, we analyze thermal scaling trends in arrays of hybrid InP-Si DFB lasers.

In a new study published in IEEE Journal of Selected Topics in Quantum Electronics, Dr. David Coenen and his team adopted a systematic approach to develop an experimentally validated thermo-optic laser model. The idea is to use this model to explore the laser array design space and document the impact of design choices on laser self-heating and overall energy efficiency. The model is demonstrated for a case study where a transceiver with 8 physical ports (fibers) and 8 wavelengths per fiber are required. In total, 64 laser output channels are required. There exist many possible configurations which meet these specifications, but which one is both energy efficient, reliable and occupies a small area?

To answer this question, the following input parameters are studied: how many lasers can fit in one die, laser die size, output power per laser gain, ambient temperature, thermal management strategy and finally integrated vs. external laser. We found several interesting conclusions: there exists a clear trade-off between laser array area and overall thermal resistance. A smaller array area will drastically increase the thermal crosstalk and temperature. Furthermore, increasing the laser length allows the generation of more light per gain section and decreases laser thermal resistance. This must, however, be balanced against the additional optical losses induced by the long gain section. Increasing laser width, and putting more lasers in one die, drastically increases thermal crosstalk.

Finally, external lasers, which need to overcome fiber coupling losses, suffer at high ambient temperatures and have more difficulty reaching the required output power. However, an advantage of an external laser is that it can be thermally decoupled from any high-power electronic chips, e.g. a network switch with co-packaged optics. The results from this publication will help optical system designers to understand the trade-offs in laser array design and provide them with the necessary tools to evaluate the impact of design choices and key performance metrics. More experimental results, which confirm these models, will be published later this year at the CLEO conference [3].

[1] A. Marinins et al., ‘Wafer-scale hybrid integration of InP DFB lasers on Si photonics by flip-chip bonding with sub-300 nm alignment precision’, IEEE JSTQE, Vol. 29, No. 3, 8200311, Nov. 2022.

[2] D. Coenen et al., ‘Thermal characterisation of hybrid, flip-chip InP-Si DFB lasers’, Micromachines, Vol. 14, No. 2, pp. 381, Feb. 2023.

[3] D. Coenen et al., ‘Experimental and theoretical investigation of thermal crosstalk in RSOA array’, CLEO, accepted, 2025.

***

Reference

Authors: David Coenen et al.                 

Title of original paper: Thermal Scaling Analysis of Large Hybrid Laser Arrays for Co-Packaged Optics

Journal: IEEE Journal of Selected Topics in Quantum Electronics

DOI: https://doi.org/10.1109/JSTQE.2024.3444923

Affiliations: IMEC, Leuven, Belgium

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.